1) On headline results, the key narrative this quarter is the U.S.–Iran conflict driving higher oil and broader goods prices, lifting both revenue and costs.

Revenue rose 11.6% YoY, accelerating vs. 9.2% in Q2 and marking the first double‑digit growth since 2022, beating expectations. Operating profit was $2.82bn, up 11.3% YoY; amid inflation, OPM dipped just 1bps YoY, with no sign of revenue growing without profit.

2) Comp growth re‑accelerated, still price‑led: On core metrics, total comps rose 9.8% YoY, a clear step‑up from 7.4% in Q2. Structurally, traffic rose only 2.4% YoY, decelerating from 3.1%. Average ticket jumped 7.3% vs. 4.2% last quarter.

It is clear that higher comps were driven by surging fuel and broadly higher goods prices, while consumption willingness/capacity among Western middle‑income consumers weakened (as seen in traffic).

3) U.S. most resilient: By region, ex‑FX and fuel, U.S. comps rose 6.8% YoY, the only market to accelerate this quarter. The lift was still price‑led, as U.S. traffic growth slowed notably from 2.4% to 1.8%.

By contrast, Canada and Other Intl comps both slowed QoQ, underscoring the stronger U.S. resilience to macro shocks.

4) E‑com still leads: Unlike slowing store traffic, Costco e‑com sales rose ~22% YoY, roughly stable vs. Q2. Meanwhile, online traffic accelerated to 37% this quarter.

With inflation showing signs of re‑acceleration, U.S. consumption is again shifting online, a trend widely seen across grocery channels.

5) Member growth off the bottom; price‑hike tailwinds fade: Membership fee revenue was ~$1.37bn, +10.7% YoY, with growth decelerating QoQ. That said, paid members rose by 0.8mn QoQ vs. +0.7mn in Q2, a modest improvement.

Also, North America renewal rate ticked up 10bps after several quarters of declines, suggesting better momentum. The slower growth in fee revenue mainly reflects the membership price hike that began in Q3 last year, with tailwinds fading on a higher base; Avg. fee per member YoY growth narrowed to 6.3% from 8%+. Further narrowing is likely.

6) Cost discipline offsets GP pressure; profit held up: Inflation lifted tickets and top‑line growth, but the impact on gross profit and costs was larger. Retail GPM was 11.04%, down 21bps YoY. Core grocery GPM fell by a larger 46bps, partially offset by ancillary businesses and a positive LIFO impact.

The company has said that under inflation, it first absorbs pricing pressure rather than fully passing it through, to defend price leadership. This quarter bears that out.

At the same time, tight opex control offset GP pressure, with opex ratio down 20bps YoY, mainly from in‑store efficiency gains.

With GP and opex ratio narrowing by similar magnitudes, OPM dipped just 1bps YoY, keeping profit growth broadly aligned with revenue — a solid outcome.

Longbridge Dolphin Research view:

Costco remains highly resilient. Amid geopolitical volatility and a fresh inflation uptick, it benefited from inflation on sales while holding margins via cost control, effectively enjoying the upside without meaningful downside — testament to best‑in‑class execution.

This print should be a positive for the stock. That said, as we have noted, Costco’s revenue and profit growth typically fluctuate within low‑to‑mid single digits, so single‑quarter results rarely move the stock much absent a structural shift. Valuation swings and market style tend to be bigger drivers.

On valuation, the Street expects FY27 profit of roughly $10bn, implying ~44x on the current mkt cap, around the 80th percentile vs. historical peaks, which is not obviously compelling.

Hence, we find it more useful to read U.S. consumer signals through Costco’s print.

Both company data and U.S. Census statistics show that post U.S.–Iran conflict, higher inflation did not dent retail sales; growth has instead accelerated.

Beyond the nominal lift from early‑stage inflation, equities’ wealth effect and Mar–Apr tax refunds also supported spending.

More importantly, U.S. in‑store traffic growth at Costco continues to slow to below 2%. While some of this reflects natural channel shift to online, persistently high prices could bite more after a lag, posing a risk to consumption.

Details below:

I. U.S. keeps weakening; comps lack true improvement

1) Big step‑up in comps — the inflation effect

Core metric: total comps rose 9.8% YoY vs. 7.4% in Q2, above expectations and optically strong. Yet traffic was up only 2.4% YoY, slower than 3.1% in Q2, while avg. ticket surged 7.3% vs. 4.2%.

Together, higher comps were driven mainly by the fuel spike and broad price inflation; under price pressure, Western middle‑income consumers’ willingness to spend has softened (as seen in traffic).

Ex‑fuel and FX, comp growth was 6.6%, slightly below 6.7% in Q2, reinforcing the point.

2) In the conflict backdrop, U.S. demand proves more resilient

By region and on a like‑for‑like basis excluding FX and fuel, the largest market — U.S. comps rose 6.8% YoY, the only true acceleration (vs. 6.4% in Q2), implying broader price inflation even ex‑fuel.

U.S. traffic growth fell to 1.8% from 2.4% last quarter, marking five straight quarters of deceleration and highlighting pressure on North American retail (at least the mid‑income cohort).

In Canada and Other Intl, ex‑fuel/FX comp and traffic growth both slowed QoQ, showing U.S. outperformance in inflation resilience.

3) Online continues to outperform

E‑com sales rose 21.5% YoY, roughly steady QoQ, and ~21% ex‑FX, remaining a clear growth engine.

Per management, online traffic accelerated to 37% YoY, and as seen at Walmart, e‑com’s role should keep rising.

4) With higher avg. ticket on inflation, merchandise sales rose 11.6% YoY, up from 9.8% in Q2, the first return to double‑digit growth since 2022.

II. Membership fee growth slowed, but renewals and adds improved off the bottom

On subscriptions, membership fee revenue was ~$1.37bn, +10.7% YoY, with a notable QoQ slowdown. On quantity, paid members rose by 0.8mn QoQ, still modest but improving.

Meanwhile, North America renewal rate ticked up 10bps after several quarters of declines. Overall renewal held at 89.7%.

Thus, the slowdown in fee revenue mainly reflects last year’s Q3 fee hike, so price tailwinds fade on a higher base; avg. fee per member YoY growth narrowed to 6.3% vs. 8.5% in Q2.

That does not signal weaker operations, but fee growth will likely converge toward member growth (Approx. 4%–5%).

Combining retail and membership, total revenue was ~$69.2bn, +11.6% YoY and slightly ahead of estimates.

III. GP narrowed under inflation, but cost control kept margins steady

While price inflation accelerated nominal revenue, the impact on GP and costs was larger. Retail GPM was 11.04%, down 21bps YoY. Core grocery GP fell 46bps YoY, partially offset by ancillary businesses and a positive LIFO contribution.

Management has reiterated that under inflation, it absorbs cost pressure first and delays pass‑through to protect price leadership. This quarter followed that playbook.

As a result, retail GP was ~$7.64bn, +9.6% YoY, lagging top‑line growth.

Fortunately, opex ratio fell 20bps YoY on cost discipline, with 12bps from store ops, 3bps from HQ, and 5bps from normalized vacation accruals.

Thus, most of the GPM compression was offset by opex, and OPM fell just 1bps YoY, a minimal impact. Operating profit was $2.82bn, +11.3% YoY, with no sign of negative operating leverage.

Net income rose 13.8% to $2.19bn, helped by higher interest income ($160mn vs. $90mn last year) and a lower tax rate (25.2% vs. 26.2%).

<End of Text>

Longbridge Dolphin Research prior work on Costco:

Earnings reviews:

Dec 12, 2025 review: U.S. Gov. shutdown ‘victim’? Is Costco the last leg down

Sep 26, 2025 review: Middle class under strain — headwinds for Costco?

Sep 26, 2025 transcript: Costco (Trans): 35 new clubs in FY26; minimize price hikes

May 30, 2025 review: Trump tariff blitz; Costco remains rock solid

May 30, 2025 transcript: Costco (Trans): Tariff impact to last all year, but manageable

Mar 7, 2025 review: Volatile U.S. market? Costco stays steadfast

Mar 7, 2025 transcript: Costco (Trans): No clear inflation yet; watch tariffs

1. Core results: $Dell Tech.US revenue was $43.8bn (+87% YoY), well above consensus ($39.0bn). QoQ revenue rose by $10.5bn, driven mainly by AI server shipments.

Gross margin was 17.8%, down 240bps QoQ but ahead of consensus (16%). The margin contraction reflected higher storage costs and mix shift toward lower-margin hardware.

2. ISG (Infrastructure Solutions): Q1 revenue was $29.0bn, up $9.4bn QoQ and far above consensus ($24.9bn). The QoQ step-up came from AI server growth and a rebound in traditional servers.

① AI servers: AI server revenue was about $16.1bn, topping raised buy-side expectations ($15.6bn).

New AI orders reached $24.4bn in the quarter, taking AI backlog to $51.3bn by quarter-end. Dolphin Research estimates next-quarter AI server revenue at $15.5bn+ amid clear supply constraints.

② Other: Beyond AI servers, legacy areas also rebounded. Traditional server-related revenue was roughly $8.5bn (+92% YoY), well above consensus ($5.0bn) and the key source of the beat. Storage contributed about $4.3bn (+8% YoY) .

3. CSG (Client Solutions): Q1 revenue was $14.6bn (+17% YoY), above consensus ($13.9bn). By customer type: commercial revenue was $13.0bn (+18% YoY); consumer was $1.6bn (+9% YoY).

4. Next-quarter guidance: DELL guides FY2027 Q2 revenue of $44–45bn, above consensus ($37.6bn). GAAP EPS is guided at $4.5, beating consensus ($3).

Dolphin view: Broad beat as AI and traditional servers both take off

DELL delivered an all-round beat, with revenue up $10.5bn QoQ and growth accelerating. The quarter was propelled by ISG (servers) shipments.

AI revenue reached $16.1bn, up $7.2bn QoQ and above raised expectations ($15.6bn). New AI orders were $24.4bn (vs. $10–15bn expected), pushing AI backlog to $51.3bn by quarter-end and underpinning high growth ahead.

Heading into earnings, the Street had already raised AI server expectations. The biggest surprise came from traditional servers: Q1 traditional server revenue hit $8.5bn (+92% YoY), smashing consensus ($5.0bn) and driving the beat. Consensus had assumed only a steady uptick, but demand is clearly inflecting.

Recent share gains have been fueled by rising AI server expectations. DELL is transitioning from GB200 to GB300, which should lift ASPs. It benefits from both volume and price tailwinds, plus a legacy server recovery; AI momentum, in particular, helped the multiple break above the traditional range ceiling.

Management raised full-year guidance: FY revenue to $165–169bn, up by $17bn (prior $138–142bn). Q1 itself was about $10bn above the prior quarterly run-rate, and Dolphin Research sees the full-year guide as conservative with room for further raises.

Outside the print, investors focused on the following:

a) Ongoing tightness in storage: The biggest downstream impact is on PCs, where the cost-to-price-to-demand transmission cooled end-market growth to 4% in Q1.

Meanwhile, rising storage prices pressured hardware margins across PCs and servers. GPM fell to 17.8%, reflecting cost inflation, yet still beat the 16% Street view, underscoring solid cost control.

b) AI servers: the key swing factor as legacy faces pressure.

Hyperscalers have lifted capex outlooks, and Dolphin Research estimates the four core cloud players (Alphabet, Meta, Microsoft, and Amazon) could spend $700bn+ in 2026, up ~80% YoY.

With capex accelerating, DELL had struggled to break $10bn in quarterly AI servers, capping AI re-rating potential. This quarter’s $16.1bn, up $7.2bn QoQ, showcased tangible high growth, reflecting robust AI server demand and higher ASPs from the GB200→GB300 transition.

DELL has deep ties with NVIDIA, offering Blackwell and Vera Rubin options and collaborating on the next-gen Feynman platform. Especially in Tier-2 CSPs (Neocloud) and public sector/enterprise, these customers typically avoid direct ODM sourcing given integration needs; they prefer OEMs with full solutions.

DELL provides full-stack, end-to-end services, spanning desktop AI workstations (GB10/GB300 DGX), rack servers, storage, networking, software orchestration (OpenManage), and services (installation, tuning, 24/7 support). This end-to-end model positions DELL to win more in Tier-2 CSP and public sector markets.

DELL’s market cap is $204.7bn, implying ~16x FY2027 post-tax core OP on assumptions of 60% revenue growth, 17% GPM, and a 17% tax rate. Historically, the multiple has ranged 8–20x PE, and after the sharp estimate revisions, today’s PE sits just above the mid-range.

After the recent rally, the market has priced in faster AI growth. Leading sell-side shops now model $65bn of AI revenue (+163% YoY), implying investors expect a guide-up post-print.

Management raised full-year AI guidance to $60bn (from $50bn), which Dolphin Research still views as conservative: ① 1H AI revenue should reach $31.6bn, and with Rubin upgrades in 2H, revenue should not be lower; ② backlog already exceeds $50bn, and grew even as AI revenue surged, signaling persistent undersupply.

Overall, AI is accelerating, traditional servers are clearly rebounding, and PCs/margins beat expectations. Operations are improving across the board, and there is scope to raise the full-year outlook again. With AI tailwinds and high growth, DELL deserves to trade beyond traditional frameworks, and its multiple could keep breaking past historical ceilings.

Details on DELL (DELL.N) below:

I. DELL overall results

1.1 Revenue

In FY2027 Q1 (26Q1), DELL delivered revenue of $43.8bn (+87% YoY), beating consensus ($39.0bn) . QoQ revenue rose $10.5bn, led by ISG growth; within that, AI revenue rose $7.2bn QoQ and traditional servers rose $2.7bn QoQ.

1.2 Gross profit

Gross profit was $7.8bn (+35.6% YoY) in FY2027 Q1 (26Q1).

GPM was 17.8%, down 240bps QoQ but above consensus (16%) .

The decline reflected: ① higher storage costs; ② mix shift toward lower-margin ISG, which diluted blended GPM.

1.3 Opex

Operating expenses were $4.1bn in FY2027 Q1 (26Q1), up 9% YoY. Scale benefits drove the opex ratio down to 9.4% this quarter.

Breakdown: 1) R&D was $980mn, up 22% YoY, with increased investment momentum. 2) SG&A was $3.1bn, up 6% YoY.

1.4 Net profit

Core operating profit was $3.66bn in FY2027 Q1 (26Q1), up 214% YoY, with a core OPM of 8.3%. Growth on the bottom line was driven by the sharp revenue increase.

II. Core segments: AI accelerates, traditional rebounds

By segment, ISG expanded on AI server strength, reaching 66% of revenue this quarter.

Based on management’s growth framework, FY2026–FY2030 CAGR for ISG (11–14%) will outpace CSG (2–3%), lifting ISG’s mix further.

ISG is the most important part of the company. By sub-segment:

2.1 ISG (Infrastructure Solutions)

ISG revenue was $29.0bn in FY2027 Q1 (26Q1), up 181% YoY, far above consensus ($24.9bn) .

Detail: ① AI servers were about $16.1bn, up $7.2bn QoQ, and the main ISG driver. ② Traditional servers and related were $8.5bn (+92% YoY). ② Storage was $4.3bn (+8% YoY).

ISG growth was powered by both AI and traditional servers. Traditional servers began a sharp rebound, while AI’s high growth now accounts for over half of ISG.

Beyond reported AI revenue, management disclosed leading indicators in new AI orders and backlog. New AI orders were $24.4bn (ahead of the $10–15bn range), and AI backlog reached $51.3bn, setting up sustained FY2027 growth.

On supply, DELL has deep collaboration with NVIDIA, offering Blackwell, Vera Rubin, and working on the next Feynman platform. As GB200→GB300→Rubin upgrades roll through, AI server ASPs should keep rising.

On the customer side, Tier-2 CSPs (Neocloud) and public sector/enterprise typically do not buy directly from ODMs due to integration complexity; they need OEMs that offer full-stack solutions. DELL’s end-to-end capabilities make it a preferred choice and should drive order wins in these markets.

Although management raised full-year AI revenue guidance to $60bn (from $50bn), Dolphin Research still sees it as conservative. 1H can deliver $31.6bn, and with Rubin upgrades, 2H should not be lower. With $50bn+ in AI backlog, further guide-ups are likely.

2.2 CSG (Client Solutions)

CSG revenue was $14.6bn in FY2027 Q1 (26Q1), up 17% YoY, beating consensus ($13.9bn) .

By mix: commercial remained the core at $13.0bn (+18% YoY), while consumer revenue was $1.6bn (+9% YoY).

Global PC shipments were 65.6mn units in Q1 2026 (+4% YoY), with growth decelerating. DELL shipped 10.3mn units (+7% YoY), outpacing the market and lifting share to 15.7%.

While PC demand remains soft overall, DELL’s commercial-heavy mix reduces exposure to the consumer cycle. Commercial PCs still grew 18% this quarter.

Amid ongoing storage price increases, DELL raised PC pricing earlier this year, passing through part of the cost. Shipments still grew YoY, indicating strong brand and execution.

Net-net, storage tightness will continue to weigh on PCs, but given mix and execution, CSG should remain on a steady upward track.

<End here>

Related DELL coverage by Dolphin Research:

Earnings calls and reviews

Feb 27, 2026 Trans: Dell (Trans): AI orders exclude Rubin; legacy also stays positive

Feb 27, 2026 First Take: DELL: Can doubling AI offset storage tightness?

Nov 26, 2025 Trans: Dell (Trans): Value lies in L11+ end-to-end solutions

Nov 26, 2025 First Take: DELL: Storage hikes add friction? AI guide supports

Aug 29, 2025 First Take: Dell: Are the ‘crazy’ AI orders just a flash in the pan?

Deep dives:

Jul 11, 2025 Deep Dive: AI double buffs: A new spring for Dell?

Jul 9, 2025 Deep Dive: Dell: Riding the AI wave with a comeback

1) Revenue beat, but ‘Services & others’ fell from a seasonal high: Total revenue came in at RMB 13.0 bn, above the RMB 12.8 bn consensus, mainly on stronger vehicle sales.However, ‘services & others’ was RMB 2.03 bn, down RMB 1.15 bn QoQ from RMB 3.18 bn, as last quarter recognized milestone licensing from the E/E architecture R&D with Volkswagen, now normalizing this quarter, and there was no extra overseas carbon-credit revenue this quarter.

2) Auto revenue beat; ASP rose against the trend on mix: Auto sales were RMB 11.0 bn (-23% YoY; vs. consensus RMB 10.8 bn). ASP reached RMB 175k, up RMB 11k QoQ and above market expectations of RMB 172k, driven by higher mix of the higher-priced X9 and higher-margin exports, offsetting price cuts.

3) Auto GPM a small positive surprise; mix offset cost headwinds: Q1 deliveries were soft (down 33% YoY to 63k units). Higher BOM on X9, lower scale that limited fixed-cost absorption, early ramp for range-extended trims, and rising prices for key inputs (batteries, chips, aluminum) lifted unit cost by ~RMB 11k to RMB 154k.

Despite aggressive ‘spec bump at same price’ pricing and weaker volume, the market expected auto GPM to fall QoQ to 11.3%. Actual auto GPM was 12.1% (down only 90bps QoQ), beating. Unit GP held at ~RMB 21k, roughly flat QoQ.

4) Overall GPM beat, led by auto GPM: Q1 blended GPM was 20.6%, beating the 20.2% consensus and the 20% guide. Unlike past quarters when software/licensing led the beat, this time it was driven by the core auto business. By contrast, ‘services & others’ GPM fell to 66.5%, slightly below the ~67% consensus, as high-margin tech R&D and carbon-credit contributions declined in mix.

5) Heavy AI and new-model spend drove a big R&D beat; S&M/GA trended lower QoQ: To prepare for a 2026 ‘product year’ and AI ecosystem, XPeng front-loaded strategic investment. R&D was RMB 2.91 bn (vs. consensus RMB 2.53 bn), focusing on the dual-energy strategy (BEV + range-extended) product cycle, in-house ‘Turing’ chips and VLA 2.0 model upgrades, and the ‘Iron’ humanoid robot (mass production targeted by end-2026 with >1,000 units/month capacity).

S&M/GA was RMB 1.88 bn (vs. RMB 2.03 bn consensus), down ~RMB 900 mn QoQ on lower franchise commissions due to weaker volume. Channel build-out continues, with a net addition of 12 stores in the quarter to 733, laying groundwork for future scale-up.

6) Loss widened as volume leverage has yet to kick in: Net loss was RMB -1.78 bn, below the market’s RMB -1.3 bn estimate, pressured by elevated R&D, a QoQ decline of RMB 660 mn in one-off ‘other income’ (e.g., subsidies), and larger FX losses.On core OP (GP minus core Opex), Q1 was RMB -1.87 bn (vs. RMB -40 mn in Q4), implying volume failed to stimulate end-market demand as expected and positive operating leverage has not materialized, leaving the core business bleeding near term.

Longbridge Dolphin Research view:

Overall, XPeng delivered a slight beat against low expectations. A richer mix led by X9 and rising export contribution lifted ASP QoQ, offsetting rising unit costs and driving upside in auto revenue and GPM.

But XPeng is still in a ‘invest now for future payoff’ phase. High R&D plus seasonal volume softness widened operating losses.

Q1 auto sales were 133k units, -33% YoY. Despite the combined push of G7, P7+, G6, and G9, and the full rollout of the dual-energy strategy (BEV + range-extended), legacy base models under-delivered, fueling doubts about new-cycle power that are already reflected in the stock’s drawdown.

As the Q1 trough appears largely priced in, the market focuses on Q2 guidance, especially after the dual-energy strategy faced interim pressure in Q1. Above-consensus Q2 volume and revenue guidance validates the potential of the 2026 MONA M03 refresh and the all-new GX to pull XPeng out of its rut. Details:

a) Volume guide beat, driven by new models

Q2 delivery guidance is 100k–106k units, above the 97k consensus and roughly flat YoY. With 31k units delivered in Apr, implied May/Jun average needs to be 34.5k–37.5k units. The lift should come mainly from two models — the 2026 MONA M03 refresh (launched Apr 2) and the all-new full-size flagship SUV GX (RMB 289.8k–359.8k; launched May 20) — confirming they are helping XPeng overcome early hurdles in the dual-energy rollout.

b) Revenue guide implies flat QoQ ASP; GX offsets higher MONA M03 mix

Q2 revenue guidance of RMB 19.6–20.8 bn also tops the RMB 19.3 bn consensus, largely volume-driven. It implies Q2 ASP around RMB 175k, roughly flat vs. Q1, as the higher-priced GX helps offset the rising mix of the lower-priced MONA M03 (RMB 119.8k–151.8k).

② For full-year 2026:

Despite a bumpy Q1 start for dual-energy, XPeng still has four new SUVs queued up, sustaining a strong cycle, with exports becoming a core growth engine.

Exports as the core growth driver: The company plans to double overseas sales in 2026 to 90k units, prioritizing Israel, Germany, Norway, Thailand, and France, and expand the network to 680 stores across 60+ countries/regions. Exports reached nearly 17.5k units in Jan–Apr. Longbridge Dolphin Research sees execution risk on the aggressive target and, under a neutral case, estimates 80k exports (+78% YoY).

Domestic base supported by a ‘big new-model year’: With two volume MONA SUVs launching through Q2/H2, Longbridge Dolphin Research expects 2026 domestic sales to rise ~9% YoY to 420k units, broadly in line with industry growth.

On AI: XPeng is accelerating the shift from R&D to commercialization:

a. ADAS/AD base upgrade: Hardware: in-house ‘Turing’ chip (750 TOPS per die) is in mass production across the lineup. Algorithms: VLA 2.0 foundation model launched in Mar with a ‘vision-to-action’ architecture, to serve autos, Robotaxi and robotics on a unified stack. A new ‘General Intelligence Center’ was created to speed integration.

b. Robotaxi rollout: Plans to launch three Robotaxi-focused models in 2026, equipped with four Turing chips (3,000 TOPS), start pilot operations, and license SDKs to third parties.

c. Humanoid robot sprint: Next-gen ‘Iron’ is in a critical R&D phase, featuring solid-state batteries and three Turing chips, integrating multimodal foundation models, aimed at commercial, industrial, and home use cases. Mass production is targeted by end-2026 with monthly capacity >1,000 units.

Based on full-year 2026 sales of 500k units (+16% YoY) and revenue of ~RMB 94.0 bn, Longbridge Dolphin Research assigns a conservative 1.0–1.2x P/S, implying a market cap of RMB 94.0–106.6 bn as a safer entry to capture ‘auto + AI’ optionality, with a stronger margin of safety (this mainly values the auto business, excluding AI options in Robotaxi and robotics).

If AI (Robotaxi pilots, humanoid mass production, etc.) executes and monetizes, the market may re-rate XPeng. Assuming 1.5x P/S in 2026 (AI premium), implied market cap would be ~RMB 141.0–160.0 bn, or ~33–50% upside vs. the current U.S. market cap of ~RMB 106.5 bn.

Therefore, with supportive Q2 guidance, two volume MONA models in H2, faster Robotaxi rollout, and year-end humanoid mass production as catalysts, XPeng shares still have room to run. If AI delivers on schedule, the valuation center (P/S multiple) could move higher.

Full text below:

I. Auto GPM: a modest positive surprise

With Q1 deliveries already disclosed, investors focused on auto revenue and margins. Deliveries fell 33% YoY to 63k units, weakening scale, while raw-material costs rose. The market expected auto GPM to drop from 13% in Q4 by 170bps to 11.3% in Q1.

Actual auto GPM printed at 12.1%, above consensus, mainly because ASP kept rising on improved mix and higher export share. This offset cost pressure more than expected.

Unit economics details:

a) ASP: higher-priced trims and exports lifted pricing

ASP reached RMB 175k, up RMB 11k QoQ from RMB 164k and above the RMB 172k consensus. Two drivers:

① Higher X9 mix: The high-ASP X9 rose 6ppts QoQ to 15% of sales. Meanwhile, the budget MONA M03 fell 5ppts QoQ to 33%, shifting domestic mix toward higher price bands.

② Rising export mix with higher margin: Exports were 11.6k units (+52% YoY), while domestic volume fell 41% YoY to 51k. Export mix rose 5.3ppts QoQ to 18.4%, and higher overseas pricing directly lifted ASP.

③ Bigger discounts, offset by better mix and exports

XPeng increased promotions in Q1: cash discounts of RMB 5k/8k/17k/8k/8k/5k for G6/G7/G9/P7+/P7/MONA M03, and 3-year 0% or 5–7-year low-rate financing on select models. Despite heavier promotions vs. Q4, the negative was offset by mix improvement and higher export contribution.

b) Unit cost: scale headwinds and input inflation pushed costs up

Unit cost rose to ~RMB 154k, up ~RMB 11k QoQ from RMB 143k. Key drivers:

① Higher inherent cost of premium trims: A larger X9 mix lifts ASP but, as a large MPV, its BOM is structurally high, raising average unit cost. ② Scale effect stalled: Q1 deliveries were only 63k (-33% YoY), near the low end of the 61–66k guide, limiting depreciation/amortization absorption. Despite sector headwinds after purchase-tax changes (~5% YoY EV industry decline), XPeng’s drop exceeded the industry as it pushed a full dual-energy lineup across G7/P7+/G6/G9.

New models launched since late Jan with ‘range-extended = BEV price’ and ‘spec bump at same price’ did not trigger the expected demand response amid a weak macro and a policy gap, so scale benefits did not materialize. ③ Early ramp and supply volatility: Early ramp losses (especially first-time range-extended), plus rising prices for batteries, memory chips, and aluminum, further increased COGS per unit.

c) Unit GP: roughly flat QoQ

Unit GP was ~RMB 21k, roughly flat QoQ and above the ~RMB 19k consensus. With unit cost up ~RMB 11k QoQ, ASP also rose ~RMB 11k, keeping unit GP stable. Auto GPM printed at 12.1%, down only 90bps QoQ vs. the 11.3% consensus.

II. Q2 guide beat; new models may reverse early dual-energy softness

After interim pressure in Q1 for the dual-energy push, above-consensus Q2 volume and revenue guidance validate the 2026 MONA M03 refresh and the new GX as catalysts to lead XPeng out of the lull. Specifically:

a) Volume guide beat, driven mainly by new models

Q1 auto sales were 133k units, -33% YoY. Despite the G7/P7+/G6/G9 combo and the full dual-energy rollout, legacy base-model uplift lagged, and doubts on the new-cycle firepower have been reflected in the share price.

Q2 delivery guide is 100k–106k, above the 97k consensus and roughly flat YoY. With 31k in Apr, implied May/Jun average is 34.5k–37.5k, mainly driven by the 2026 MONA M03 (launched Apr 2) and the all-new flagship GX (RMB 289.8k–359.8k; launched May 20), confirming these models are reversing early setbacks in the dual-energy strategy.

b) Revenue guide implies flat ASP QoQ; GX offsets higher MONA M03 mix

Q2 revenue guidance of RMB 19.6–20.8 bn also beats the RMB 19.3 bn consensus, driven by volume. It implies ASP at ~RMB 175k, flat vs. Q1, as the higher-priced GX (RMB 289.8k–359.8k) offsets the rising mix of the lower-priced MONA M03 (RMB 119.8k–151.8k).

III. Overall GPM beat, led by auto

Q1 revenue was RMB 13.0 bn (vs. RMB 12.8 bn consensus). Blended GPM was 20.6%, beating the 20.2% consensus and the 20% guide, driven by autos, while other revenue was broadly in line.

① Revenue beat on better auto:

Auto revenue and GPM beat as ASP gains offset cost inflation. Auto sales were RMB 11.0 bn (-23% YoY; vs. RMB 10.8 bn consensus), supported by a higher X9 mix and strong, higher-ticket exports.

‘Services & others’ was RMB 2.03 bn, down RMB 1.15 bn QoQ from RMB 3.18 bn. Two reasons: a) last quarter recognized sizable E/E architecture licensing revenue from the Volkswagen collaboration, normalizing this quarter; b) extra carbon-credit income tied to exports last quarter, with none this quarter.

② Slight GPM beat, mainly on auto GPM

Auto GPM was 12.1%, down 90bps QoQ but above the 11.3% consensus, as richer mix offset weakened scale and input inflation. ‘Other & services’ GPM was 66.5%, down 430bps QoQ and below the ~67% consensus, on lower shares of high-margin tech R&D and carbon credits.

IV. Opex: building for dual-energy and AI, but weighing on earnings

XPeng treats ‘intelligence’ as its moat, requiring sustained heavy spend in AI driving and new platforms. Front-loaded investment for a 2026 product year pushed R&D well above expectations and capped core OP this quarter.

1) R&D: RMB 2.91 bn, a big beat and up QoQ, fully leaning into new models and physical AI

R&D reached RMB 2.91 bn vs. RMB 2.53 bn consensus, up RMB 40 mn QoQ from RMB 2.87 bn, driven by new-model development and AI spend in three areas: ① Front-loaded spend for a dense product cycle: since Q4 2025, XPeng has rolled out dual-energy across BEVs and plans four new SUVs in 2026.② Smarter driving: in hardware, in-house ‘Turing’ (750 TOPS) is shipping across the lineup; in algorithms, VLA 2.0 launched in Mar 2026 with a ‘vision-to-action’ end-to-end design to cut latency, serving autos, Robotaxi, robotics, and flying vehicles on a unified base, supported by a new ‘General Intelligence Center’ that merges AD and cockpit teams.③ Humanoids: next-gen ‘Iron’ uses industry-first solid-state batteries and three ‘Turing’ chips (total 2,250 TOPS), integrating VLT (reasoning), VLA (action), and VLM (interaction), targeting commercial/industrial/home use, with initial deployments in stores and campuses. Mass production is targeted by end-2026 with monthly capacity >1,000 units.

Guidance: 2026 R&D to rise ~26% YoY to ~RMB 12.0 bn, with ~RMB 7.0 bn for AI and ~RMB 5.0 bn for vehicle programs. This underscores the dual-track focus.

2) S&M/GA: sharply lower QoQ on volume, while channel expansion continues

S&M/GA was RMB 1.88 bn, below the RMB 2.03 bn consensus and down ~RMB 900 mn QoQ, mainly on lower commissions to franchise stores amid weaker sales. To support dual-energy penetration and four new SUVs in 2026, XPeng kept expanding: +12 stores QoQ to 733 and +1 city to 256.

V. Operating loss widened; volume leverage yet to show

Net loss was RMB -1.78 bn vs. RMB -1.3 bn consensus. Beyond heavy R&D, a sharp QoQ drop in ‘other income’ (mainly one-off subsidies, -RMB 660 mn QoQ) and larger FX losses weighed on net.

Core OP (GP minus core Opex) was RMB -1.87 bn in Q1 vs. RMB -40 mn in Q4, deteriorating by ~RMB 1.83 bn, mainly because: a) volume slumped (Q1 133k, -33% YoY), limiting fixed-cost absorption and operating leverage; b) R&D rose QoQ from RMB 2.87 bn to RMB 2.91 bn, further suppressing earnings.

<End here>

Longbridge Dolphin Research’s deep dives and tracking on XPeng include:

Earnings season

Aug 19, 2025 earnings take: XPeng: GPM at a new high — ready to spread its wings?

Aug 20, 2025 call transcript: XPeng (Trans): Aiming for New P7 to rank top-3 in RMB 200k–300k BEVs

May 21, 2025: XPeng: After the grind, ready to soar?

May 21, 2025: (1Q25 Trans): Poised for takeoff?

1) Growth remains unremarkable: Core subscription revenue was approx. $10.6bn, +12% YoY in cc, a modest 100bps acceleration vs. last quarter. However, excluding Informatica consolidation effects, underlying growth was broadly unchanged vs. last quarter.

By line of biz, the company revamped disclosures into two buckets: (1) Agentforce Apps, its more traditional applications, grew 7% in cc, unchanged vs. recent quarters. (2) The more PaaS-leaning Data & Platform segment grew 23% in cc, up ~200bps QoQ, which looks solid.Yet stripping out Informatica, the underlying growth is only around 10%, suggesting a much more muted picture.

Across angles, AI has yet to show a meaningful lift to consolidated revenue growth.

2) AI skews upstream; ARR up steadily: Agentforce & Data AI ARR reached $3.4bn this quarter, up $0.5bn QoQ, a modest step-up. Roughly one-third comes from app-side Agentforce, with two-thirds from data/platform layers.This aligns with the notion that AI benefits upstream/mid-layer more visibly, while the lift to end apps is limited or even negative.

AI-related revenue accounted for 8% of total, vs. 6.8% last quarter, inching higher. However, unless AI translates into faster overall growth, the benefit is limited and looks more like internal cannibalization of non-AI revenue.

3) Leading indicators also middling: cRPO was $33.6bn, +13% YoY in cc, identical to last quarter’s growth. However, new bookings were only about $7.2bn, a cliff drop from $20.5bn last quarter; new bookings grew just 4% YoY, the weakest pace since 1Q25 over the past two years.

This likewise indicates that despite rolling out AI & Agent features, customer demand for Salesforce offerings remains tepid.

4) GPM continues to trend down: With revenue growth still lackluster, GPM continued to decline. Overall GPM was 76.9%, down sequentially for four straight quarters, and down 10bps YoY.Core subscription GPM fell more, at 81.6%, down 110bps YoY. This implies AI margins are lower than legacy products (unsurprising given higher compute costs), which means even if AI scales, profit may not rise proportionally.

5) Cost control preserved profit, but cash flow was soft: Positively, opex discipline was solid at ~$6.2bn (+10.6% YoY), below revenue and GP growth. Notably, S&M grew less than 10%, under the 12% street view.As a result, despite ~10bps YoY GPM compression, a 140bps drop in opex ratio lifted GAAP OPM YoY, driving OP of ~$2.35bn (+~21% YoY), ~9% above estimates. Operating profit was decent.

However, FCF grew only 3% YoY, well below the ~8% expected. The gap vs. OP growth stems from light new bookings, with revenue recognition more from consuming prior deferreds, which does not bring in cash.

6) Massive shareholder returns: As we have noted, with growth stuck, capital returns remain a key lever to sustain investor appeal. The company spent $25bn on shareholder returns this quarter alone, vs. just $14.3bn for all of FY26.About $27.1bn went to buybacks, cutting total share count by ~10% YoY this quarter, and less than $0.4bn went to dividends. While unlikely to be the norm, against a ~$145bn market cap, the quarterly shareholder yield was ~17%.

Longbridge Dolphin Research view:

1) As discussed, Salesforce’s results remain lukewarm with few highlights. Key takeaways:

a) Ex-FX and M&A, underlying revenue growth hovers just above 10%, essentially stagnant. This reflects that AI-driven revenue has not materially accelerated overall growth; the data/platform layer logically benefits more, but the actual lift is limited.

b) Meanwhile, higher AI costs pressure margins; with muted top-line progress and declining GPM, profits come under pressure early. Together this is a double whammy of no growth and weaker profitability.

2) Guidance and outlook:

Near term, management guides next-quarter revenue growth of 10% in cc, 200bps slower than this quarter and below consensus. Of that, >4pts come from consolidation, implying organic growth under 6%. cRPO growth is guided at 13% YoY (cc), unchanged vs. this quarter.

Taken together, AI still does not drive a clear re-acceleration next quarter.

On profit, guided GAAP diluted EPS midpoint is $1.75, implying >10% YoY decline, well below the market’s roughly flat expectation. Again, growth lacks a step-up while profit remains pressured.

Midterm, for FY27, the company nudged up the low end of revenue guidance, but implied growth remains only 10%–11%, suggesting growth will slow in coming quarters. This mainly reflects the absence of consolidation tailwinds by Q4, though management indicates ex-M&A organic growth may pick up in FY27.

The FY27 changes center on profit, with GAAP OPM guided down from 20.9% to 20.6%. FCF growth guidance was cut sharply from 9%–10% to 4%–5%, implying continued margin pressure from AI investments.

3) Valuation and investment view: per the prior long-term guide to FY30, GAAP after-tax operating profit is around $18bn; applying a 15x PE for a mature steady state implies ~$270bn market cap and ~$290/share, making the current sub-$180 price look quite cheap.That said, with AI evolving rapidly, the achievability of the long-term plan is highly uncertain. On a conservative FY27 view, with 10%–11% revenue growth and 20.6% GAAP OPM, GAAP OP just above $9bn implies about an 18x PE on today’s market cap; guided FY27 FCF implies a sub-10x P/FCF.

All in, Salesforce’s advantage remains a reasonable valuation, plus outsized buybacks that buttress the downside. Pure derating room seems limited from here.However, both the latest quarter and channel checks suggest muted customer interest in Salesforce’s AI and other products. Reasons include: overall IT budgets are mostly flat with incremental dollars going to AI, crowding out spend on SaaS like Salesforce; on Agents, some enterprises prefer more flexible tools like Claude Code over third-party customized agents; and pricing for products, especially mid-layer Platform services, is not particularly compelling.

Thus, while the stock looks cheap, the lack of clear upside keeps us on the sidelines.

Core charts and biz overview:

I. Scope tweak, greater focus on the mid-layer

Starting FY27, Salesforce made small changes to segment disclosure. The core subscription revenue breakout shifted from five lines — Sales, Service, Marketing & Commerce, Platform, Analytics — to two buckets: Agentforce Apps and Data 360 & Platform & Others.

Specifically, Agentforce Apps are more SaaS-like, end-user-facing applications including Sales, Service and Slack. Data 360 & Platform skews PaaS, providing platform/data services including Data 360, the new Headless Platform, and acquired businesses such as Informatica and MuleSoft.

Longbridge Dolphin Research believes the change helps separate slower-growing, more traditional SaaS apps from faster-growing data/platform businesses in the AI era, guiding investor focus to the latter.

II. Revenue growth still stagnant; AI uplift not visible

On growth, Salesforce’s core subscription revenue was ~$10.6bn, +12% YoY in cc, a 100bps acceleration vs. last quarter. But management indicated Informatica provided a bigger lift this quarter than last, so ex-consolidation growth is roughly flat.

As noted, under the new split, the app-leaning Agentforce Apps grew 7% in cc, essentially unchanged vs. prior quarters, so AI/Agents are providing only minimal lift to legacy apps.

The more PaaS-oriented Data & Platform segment grew 23% YoY in cc, up from 21% last quarter and continuing to accelerate slightly. However, excluding Informatica, the underlying growth is only about 10%, again showing no clear AI-driven acceleration.

Including ~$0.54bn of professional services, total revenue was about $10.6bn, +12% YoY in cc, up 100bps vs. last quarter. Ex-consolidation growth was ~9% vs. ~8% last quarter.

III. New bookings growth hits a new low

On leading indicators, cRPO was $33.6bn, +13% YoY in cc, identical to last quarter. However, new bookings were only ~$7.2bn, plunging from $20.5bn in the prior quarter; YoY growth was just 4%, the lowest since 1Q25, again pointing to soft demand.

III. No growth pickup, and GPM under pressure

With limited revenue acceleration and likely a higher AI revenue mix, GPM continued to slide. Overall GPM was 76.9%, marking a fourth straight sequential decline and a 10bps YoY dip.

Core subscription GPM fell more (pro services’ gross loss improved), reaching 81.6%, down 110bps YoY. We think AI’s lower margin and rising mix are the drivers, with higher compute costs and pricing that has not fully shifted from seat-based to usage-based, creating a revenue-cost mismatch.

As a result, total GP was ~$8.56bn, +13.2% YoY, slightly below the Street.

IV. Cost control protects profit, but cash flow lags

Encouragingly, total opex was about $6.2bn (+10.6% YoY), below revenue and GP growth. S&M rose nearly 10%, under the 12% market view; G&A grew ~6%, and R&D rose 11.4%.

Aside from R&D, which runs higher to fund new AI features, other opex lines were restrained.

Thus, despite ~10bps YoY GPM pressure, a 140bps drop in opex ratio lifted GAAP OPM by ~130bps YoY. OP was ~$2.35bn, +~21% YoY and ~9% above expectations.

However, FCF grew only 3% YoY, well below the ~8% expected, largely because light new bookings meant limited cash inflows, as in-quarter revenue relied more on consuming deferred revenue.

<End of text>

Past Longbridge Dolphin Research on Salesforce:

Earnings reviews:

Sep 4, 2025 review ‘Salesforce: Is AI a savior or a death knell?’

Sep 4, 2025 transcript ‘Salesforce (Trans): AI is an assistant, not a killer; SMEs are the new growth vector’

Sep 4, 2025 review ‘Salesforce: Is AI a savior or a death knell?’

Sep 4, 2025 transcript ‘Salesforce (Trans): AI is an assistant, not a killer; SMEs are the new growth vector’

May 29, 2025 review ‘Salesforce: Is the AI agent dream still out of reach?’

May 29, 2025 transcript ‘Salesforce: SME demand strong; doubling down on Agentforce’

Feb 27, 2025 review ‘Salesforce: Agents must spend first to earn later?’

Feb 27, 2025 transcript ‘Salesforce (Trans): The ‘holy trinity’ of apps, data and agents’

Deep dives:

Jan 15, 2025 Initiation II ‘Salesforce: Can a legacy SaaS sprout new growth?’

Jan 7, 2025 Initiation I ‘Will ‘AI’ replace ‘human’? How much can Salesforce benefit?’

1. Revenue: $2.42bn this quarter, +9% QoQ, in line with consensus ($2.41bn). The ~$0.2bn sequential uplift was essentially all driven by data center.

2. GPM: GAAP GPM was 52.1%, up 40bps QoQ. Because GAAP margins are affected by acquisition-related amortization, they do not fully reflect underlying operations. Ex-amortization, our reference Adj. GPM was 58.3%, down 20bps QoQ. Lower-margin custom ASIC mixed in, keeping overall GPM on a slight downward trend.

3. Data center: the key focus. Segment revenue was $1.83bn, +11% QoQ, mainly on stronger optical interconnect; data center accounted for 75.8% of total.

Previously, the Street’s AI focus was on custom ASIC, but MRVL’s exposure via Amazon’s Trainium underperformed. The company still supplies Trainium 2.5 to Amazon, while the newly announced Trainium 3 is primarily sourced from Alchip, which weighed on the shares at the time.

As AI moves into inference, raw compute matters less while storage, CPUs and interconnects gain importance. Marvell is relatively ahead in interconnects, and even with softer ASICs near term, robust demand for connectivity should support high growth.

4. Next-quarter guide: revenue $2.7bn, above Street ($2.61bn), with QoQ growth led by interconnect demand ; GAAP GPM 52.1%–53.1%, broadly stable.

5. Full-year outlook: FY2027 revenue raised to $11.5bn (prior $11.0bn), +40% YoY. The lift is mainly from interconnect growth now guided to >70% (prior 50%+), while ASIC growth stays >20%; FY2028 revenue raised to $16.5bn (prior $15.0bn), +43% YoY.

Longbridge Dolphin Research view: well-positioned for the AI super-cycle; interconnect underpins growth

The print broadly matched expectations, with growth driven mostly by data center (interconnect). Ex-amortization, Adj. GPM was 58.3%, down 20bps QoQ.

Last quarter management recut disclosure from five lines to two (data center; communications & others). Data center was the primary growth engine this quarter, up 11% QoQ. ASICs were relatively weak, while growth was led by interconnect demand.

Marvell guides next-quarter revenue to $2.7bn (+11% QoQ), above Street at $2.61bn. With full-year guidance now in place, the next-quarter guide carries less incremental importance.

The recent share rally reflects a shift in the narrative. The market had worried about Trainium share loss, 1.6T DSP ramp timing and Broadcom (AVGO) competition. But in an optical super-cycle, even if custom ASICs lag, interconnect products (e.g., DSPs) still offer high-visibility growth.

In the inference era, the AI stack is better defined as storage + XPU + connectivity. Marvell is strong in optical/electrical interconnects and has augmented AI networking via M&A. Sell-side models were already moving up, and management’s FY2027/FY2028 revenue guides to $11.5bn/$16.5bn largely meet those higher expectations.

Beyond the quarterly numbers, Marvell’s growth vectors include:

1) Rack-scale networking: a recent M&A spree—XConn (PCIe/CXL switch), Celestial AI (photonic interconnect), Polariton (silicon photonics). This builds out three AI networking layers—Scale-Out, Scale-Up and Scale-Across—delivering a full high-speed interconnect portfolio from within-rack to cross-DC.

2) CXL memory expansion/pooling (Structera): not for cold data storage, but for adding a tier in the active memory hierarchy. AI inference needs massive KV cache; HBM is costly and DRAM is constrained, so NAND/storage-class memory is required for expansion.

Marvell’s CXL lineup comprises Structera A, Structera X and Structera S for near-memory acceleration, memory expansion control, and memory pooling & switching, respectively.

3) Full-stack optical interconnect

Connectivity is evolving from copper to pluggable optics (with a relative lead in PAM4 DSP) to LPO/OBO (board-level optics) to CPO (co-packaged optics) to all-optical networks. Marvell has positioned across each phase.

Today, in Scale-Out deployments, pluggable optics remain the dominant form factor. Hyperscalers like NVIDIA and Google rely mainly on pluggables, while CPO is expected to enter small-batch use only in H2 2026.

Marvell’s PAM4 DSP holds a relative lead in pluggables. Management noted the 1.6T solution has begun volume production and should ramp quickly in FY2027, which will be a major near-term growth driver for connectivity. Beyond DSPs, Marvell has active roadmaps in LPO/OBO, CPO and all-optical domains.

At MRVL’s current market cap (~$173.7bn), implied FY2028 Adj. net income multiple is ~37x, assuming ~+45% revenue CAGR, 57.5% Adj. GPM and 11.5% Adj. tax rate.

On this quarter’s print alone, results were ‘okay’. With full-year guidance provided, quarterly moves matter less. FY2027/FY2028 revenue outlooks broadly match the Street after recent upward revisions.

The rich multiple embeds high expectations, and the annual guide did not deliver incremental upside, which could prompt some near-term disappointment. That said, under the new guide, growth above 40% for the next two years remains intact, underscoring sustained high growth.

Beyond connectivity, ASICs, CXL memory expansion/pooling (Structera) and rack-scale solutions add medium-to-long term optionality. Even if the annual guide felt ‘plain’, the high-growth narrative remains in place.

Longbridge Dolphin Research’s detailed analysis of MRVL’s results follows below:

I. Marvell’s biz.

Founded on storage, Marvell expanded via a series of bolt-on M&A. Data center has become the largest revenue stream.

By segment:

1) Data center (~75%): high-growth, propelled by data center and ASIC demand and the main investor focus. It includes optical/electrical interconnect, SSD controllers and custom ASICs (e.g., Amazon AWS, Google Axion CPU), serving cloud servers and edge computing.

2) Others (~25%): the legacy enterprise networking, carrier infrastructure, consumer, and auto/industrial have been consolidated into ‘communications & others’.

II. Key metrics: connectivity heating up, revenue growth re-accelerates QoQ

2.1 Revenue

FY2027 Q1 revenue was $2.42bn, +9% QoQ and in line with the $2.41bn consensus. Growth was mainly driven by optical/electrical interconnects.

2.2 Gross profit

Gross profit was $1.26bn, up ~$110mn QoQ. GAAP GPM was 52.1% this quarter.

GAAP GPM is affected by acquisition-related amortization, so it does not fully reflect operations. On an adjusted basis we reference below.

Adj. GPM was 58.3%, down 20bps QoQ. The mix impact from lower-margin custom ASICs kept margins trending lower.

2.3 Opex and profitability

Net income was $35mn in FY2027 Q1. Ex-nonrecurring items, EBITDA was $735mn with EBITDA margin at 30.4%, pressured by lower Adj. GPM and a notable increase in opex.

III. Segment detail: connectivity drives high growth; ‘growth angles’ well stocked

Since 2018 Marvell has acquired Cavium, Innovium and others, bolstering connectivity and ASIC capabilities. Demand for interconnect and custom ASIC has lifted data center, now the biggest swing factor for results.

From last quarter, legacy lines were grouped into communications & others. Enterprise networking, carrier, consumer, and auto/industrial each fell to around or below a 10% revenue share.

3.1 Data center

FY2027 Q1 data center revenue was $1.83bn, +11% QoQ and in line with the $1.82bn consensus, driven mainly by stronger optical/electrical interconnect demand.

The focus within data center is interconnect and custom ASIC.

1) Custom ASIC: the largest program is Amazon’s Trainium.

Marvell continues to supply Trainium 2.5, while Trainium 3 is primarily provided by Alchip. This is a key reason for recent softness in ASIC.

For FY2027, the ASIC outlook is modest with ~20% YoY growth, well below CSP capex growth and reflecting share loss at a major customer (Amazon). For FY2028–FY2029, management is more constructive. They expect custom revenue to double in FY2028 and exceed $10bn in FY2029, roughly consistent with a 20% long-term market share target and including Microsoft’s next-gen Maia, plus CXL and NIC attach products.

2) Interconnect: the largest data center contributor

With ASIC softer, interconnect is delivering the bulk of incremental growth. We estimate interconnect revenue at ~$900mn this quarter, roughly half of data center.

As inference scales, the importance of ‘AI infra = storage + XPU + connectivity’ is clearer. Pluggable optics remain the mainstream Scale-Out method, and Marvell’s PAM4 DSP is a relative leader, while the company is also invested across copper, LPO/OBO, CPO and all-optical. Management raised FY2027/FY2028 revenue guides largely on interconnect strength. They now expect interconnect revenue to grow >70% YoY in FY2027 (prior 50%+), and remain fast-growing in FY2028.

3) Other incremental areas:

i) DC switch silicon: after buying Innovium, Marvell strengthened its Ethernet switch position. Related revenue was about $300mn in FY2026 and is expected to reach ~$600mn/$1bn+ in FY2027/FY2028, supported by AI demand.

ii) Full network-layer capability: recent deals for XConn (PCIe/CXL switch), Celestial AI (photonic interconnect) and Polariton (silicon photonics) build out Scale-Out, Scale-Up and Scale-Across layers, enabling a full high-speed interconnect portfolio from in-rack to cross-DC.

iii) CXL memory expansion & pooling: main offerings are Structera A, Structera X and Structera S for near-memory acceleration, memory expansion control, and memory pooling/switching.

Through these acquisitions, Marvell has built Scale-Out, Scale-Up and Scale-Across capabilities. The company aims to offer full solutions, not just sell chips.

Marvell also expanded collaboration with NVIDIA, including i) optics (building from DSP/TIA/driver supply into silicon photonics to enable Scale-Up networks); ii) NVLink Fusion integration (custom chips and networking silicon that interoperate with NVIDIA infra to give hyperscalers mix-and-match flexibility); and iii) AI-RAN (tighter coupling of OCTEON baseband processors with NVIDIA GPUs so operators can run 5G/6G and high-performance AI on the same software-defined platform).

In the inference phase, connectivity is increasingly critical, driving near-term results and supporting longer-term growth visibility. Near term, interconnect contributes the main data center uplift, with ASIC, CXL memory expansion/pooling (Structera) and rack-scale solutions as longer-term options.

3.2 Communications & others

Management now reports only the combined communications & others line, no longer breaking out enterprise networking, carrier, consumer and industrial.

FY2027 Q1 communications & others revenue was $585mn, +28% YoY. As data center accelerated, the segment’s mix fell to 24%.

For this segment, the annual outlook is unchanged: about +10% in FY2027 and low single-digit growth in FY2028.

<End>

Prior Longbridge Dolphin Research coverage on MRVL:

Earnings:

Mar 6, 2026 call Trans: Marvell (Trans): Hikes revenue guide again; second XPU to ramp in FY28

Mar 6, 2026 note: Marvell: ASIC card missed; connectivity powers the AI comeback

Dec 3, 2025 call Trans: Marvell (Trans): AWS becomes Celestial AI’s lead customer

Dec 3, 2025 note: Marvell: M&A fills gaps; a faster NVIDIA alternative?

Aug 29, 2025 call Trans: Marvell (Trans): Still not addressing Alchip’s challenge head-on

Aug 29, 2025 note: Marvell: Hyperscaler spend rises, but why didn’t ASICs benefit?

May 30, 2025 call Trans: Marvell (Trans): AI to reach half of revenue in the future

May 30, 2025 note: Marvell: AI sequential growth stalls; where is the next trump card?

Mar 6, 2025 call Trans: Marvell (Trans): Data center growth faltered

Mar 6, 2025 note: Marvell: Cooling expectations on AI; ASIC warning signs

Deep dives:

Jan 14, 2025: ASIC wars: Can Marvell beat Broadcom?

Jan 2, 2025: Marvell vs. the trillion-dollar Broadcom: Can ASIC ignite a comeback?

1) Growth weakened against the tide: unlike peers that showed sequential improvement, PDD’s top-line growth slowed vs. last quarter.

Specifically, total revenue rose 11% YoY, below the 13.5% Bloomberg cons. and slower than Q4’s 12%. More notably, ad revenue missed with just 2.5% YoY growth, far under the ~8% cons.

Longbridge Dolphin Research believes domestic core-site GMV growth likely slowed QoQ again. The bigger driver, however, was a meaningful decline in ad monetization, likely linked to tighter tax compliance that hurt SME merchants’ profitability and ad budgets.

2) Temu kept recovering, but was not a big upside surprise: in contrast, transaction-commission revenue was approx. RMB 63.9bn, with YoY growth ticking up to 20%, topping the 18% Bloomberg cons.

Given weak domestic core-site growth, overseas Temu (and/or Duoduo Grocery) likely outperformed expectations. By Longbridge Dolphin’s initial model, Temu’s revenue growth accelerated from ~20% in Q4 to ~25% in Q1.

That said, with Temu expanding across Europe, LatAm and SE Asia, our prior expectations were higher. The print, while better, was not especially exciting.

3) Marketing did not decline YoY: gross profit rose 8.4% YoY, slightly below expectations. GPM reached 55.9%, rebounding QoQ and above the 55.6% cons., and its YoY compression should keep narrowing ahead.

On opex, the market had expected lower marketing spend given last year’s peak phase when the company funded state subsidies itself. Instead, marketing expense rose 1% YoY, which likely reflects heavier Temu spend amid multi-market expansion, while core-site marketing probably fell.

G&A declined YoY and R&D stayed strong, underscoring continued AI feature investment.

4) Profit inflected but lacked punch: as state subsidies faded, profit bottomed and turned up, with OP up 22% YoY to RMB 19.6bn.

However, domestic core-site revenue missed and marketing did not fall meaningfully, leaving profit below expectations. We think Temu’s loss-narrowing likely undershot, while domestic core-site OP growth was limited and likely did not exceed RMB 1bn.

Longbridge Dolphin Research view:

PDD’s quarter was clearly weak. Key signals are as follows:

1) While sector peers improved sequentially, PDD’s growth weakened, marking a shift from prior outperformance to underperformance. Whether due to core-site GMV slowing or the tax normalization impact on SME merchants that drags monetization, the hoped-for post-subsidy rebound in core-site revenue and profit now looks less likely.

2) With multi-country exposure, Temu’s growth kept accelerating and beat expectations, but its loss-narrowing appears slower than hoped.

2) Outlook:

1) For PDD’s domestic core site, our stance is more cautious than last quarter. First, while the fade of state subsidies helped profit bottom, the benefit was milder than expected.

PDD was expected to relatively benefit post-subsidy given lower exposure to 3C and home categories. In reality, it has not outpaced peers.

b. As regulators push platforms to handle merchant tax filings centrally, tax evasion gets harder. This hits SMEs on Taobao and PDD most, squeezing merchant margins and ad budgets, and weighing on platform ad monetization.

This concern appears validated this quarter. c. Post the Apr-26 incident, regulators issued a formal penalty totaling nearly RMB 3.6bn across platforms including PDD, Meituan, JD and Taotian, with PDD fined the most at RMB 1.5bn.

Reasons included lax license checks allowing ‘ghost’ stores to sell food like cakes without proper qualifications, and improper order-routing to third parties whereby orders placed at well-known stores were fulfilled by unknown workshops. PDD’s higher fine also reflected obstruction of regulation related to the incident.

With fines settled, the RMB 1.5bn impact is immaterial vs. over RMB 100bn in annual profit. Regulators did not address a potentially tougher issue — bans on forcing merchants into auto price-matching to the lowest across the web — so regulatory risk appears meaningfully reduced for PDD.

Overall, we still expect domestic core-site profit to repair from the bottom and return to growth. But with sector growth slowing and monetization still pressured, the recovery strength will likely be mild.

2) The domestic business is mature, and PDD is not leaning into instant retail or AI as core growth pillars. Share-price and valuation optionality therefore hinges on Temu receiving a standalone valuation.

Given recent headlines (Temu now the No.2 e-comm site by global users) and faster growth this quarter, the odds of a separate Temu valuation are not low. The timing, however, could be pushed out as profitability breakeven may be delayed amid multi-front expansion.

3) As for the newly announced ‘Xin PimMu’ line last quarter, reports suggest a model akin to overseas grocers’ private labels, like Costco’s Kirkland.

Current pilots span apparel, home and outdoor. ‘Xin PimMu’ handles design/selection, pricing, marketing, sales and fulfillment, outsourcing only manufacturing, and targets overseas starting with the U.S. before Europe and SE Asia, aiming for cost-plus pricing with low GPM to ensure price competitiveness.

Strategically, unlike Shein’s small-batch fast-turn model, ‘Xin PimMu’ focuses on classic evergreen SKUs with steady demand. This can be viewed as Temu’s self-operated arm and an additional controllable supplier for Temu, though it will tie up more capital and management bandwidth.

3) Valuation: we continue to value the domestic core site and Temu separately.

For the core site, we lower FY26 GMV growth from 10%+ to ~8–9%. Given heavier-than-expected monetization pressure from tax normalization, we also cut FY26 OP growth for the core site to ~10%.

We now assume total core-site OP around RMB 114bn in FY26 (with interest/other roughly offsetting taxes), implying ~7x pre-mkt PE post-print. After this pullback, PDD’s prior premium vs. JD and Taotian has largely evaporated, and the recent fundamentals do not warrant a premium.

What could repair valuation and share price are two levers: a standalone Temu valuation, and shareholder returns via buybacks/dividends, unlocking over RMB 500bn of cash (vs. current mkt cap ~RMB 800bn) and a Hong Kong listing. The second path may be more straightforward.

For Temu, we model GMV at $110bn in FY26 and a 2% margin on GMV, implying steady-state profit of ~$2.2bn. With a better growth outlook, we lift the PE to 25–30x, suggesting Temu could account for ~46–55% of PDD’s current mkt cap as upside optionality.

That said, with PDD’s performance now average and few clear advantages vs. peers, if management keeps its stance of limited communication and no dividends/buybacks, it will be hard to win investors back.

Detailed read-through of this quarter:

I. Ad growth slowed against the trend — is PDD now the sector’s laggard?

This quarter, PDD reported total revenue of approx. RMB 106.2bn, up 11% YoY and below the 13.5% cons., also decelerating vs. Q4’s 12%, contrary to sector trends where Q1 typically outperforms Q4.

Worse, ad revenue that reflects the domestic business rose just 2.5% YoY, far below the ~8% market view and Q4’s 5%.

With sector GMV accelerating QoQ, even a slower core-site GMV at PDD should not translate to only ~2.5% ad growth. The more plausible explanation is a notable drop in ad monetization.

While the exact drivers need more work, the previously discussed tax normalization likely hit SME merchants’ profitability and ad spend the most.

II. Temu accelerated — solid, but not spectacular

This quarter, transaction-commission revenue was approx. RMB 63.9bn, up 20% YoY and above the 18% Bloomberg cons. Given the weak domestic core site, Temu and Duoduo Grocery likely beat.

By Longbridge Dolphin’s initial estimate, Temu’s revenue growth rose from ~20% in Q4 to ~25% in Q1.

Despite rapid expansion across Europe, LatAm and SE Asia, and reports that Temu reached 366mn monthly unique visitors globally to rank No.2 after Amazon, our prior bar was higher. The result was good, but not a big surprise.

III. GPM has likely bottomed; marketing spend did not fall

PDD’s GP was RMB 59.3bn (+8.4% YoY), slightly light. GPM was 55.9%, still down YoY but up QoQ (vs. 55.5% in Q4) and above the 55.6% cons., and should keep narrowing its YoY decline as Temu’s drag eases.

Total opex was approx. RMB 38.0bn (+3% YoY), well below revenue growth, indicating continued cost discipline. Marketing expense was nearly RMB 33.8bn, up ~1% YoY, versus expectations for a decline given last year’s subsidy peak.

By segment, we think domestic marketing likely fell YoY, while Temu’s spend rose meaningfully amid multi-market expansion. G&A fell ~5% YoY and R&D jumped ~24% on AI feature development, with PDD reportedly testing AI search.

IV. Profit did inflect, but missed expectations

With domestic revenue light and marketing not falling much, OP rebounded post-subsidy fade to RMB 19.6bn (+22% YoY) but still missed.

Given faster Temu growth and heavier marketing, Temu likely narrowed losses less than expected. Even so, domestic core-site OP growth looks limited and likely rose less than RMB 1bn YoY.

Adj. OP growth (ex-SBC) was 15.5%. Also, with ~RMB 2.0bn of other losses this quarter vs. over RMB 3.0bn of other gains a year ago (likely investment-related), net income fell ~15% YoY.

<End>

Longbridge Dolphin Research coverage on PDD:

Earnings season

Nov 18, 2025 review: PDD’s awkward turn: aging at light speed, yet no buybacks

Nov 18, 2025 call notes: PDD (Trans): the classic self-inflicted cut

Aug 25, 2025 review: PDD: cash machine again? Can’t beat management’s strong hand

Aug 25, 2025 call notes: PDD (Trans): yet another self-swing of the knife!

May 27, 2025 review: Self-inflicted and aggressive! PDD’s big spending with a clear agenda?

May 27, 2025 call notes: PDD (Trans): investing in consumers and merchants creates long-term value

Mar 20, 2025 review: PDD: fallen from the pedestal, how long can pride hold?

Mar 20, 2025 call notes: PDD (Trans): do not judge the company by short-term financials!

In detail: Details as follows.

1) Kling is the only bright spot: Q1 Kling revenue reached RMB 650 mn (+330% YoY), ahead of the RMB 500 mn guide. The monetization ramp was led by Dec’s all-modal instruction editing on Kling O1 and the Feb launch of Kling 3.0, an all-in-one multimodal I/O model.

ARR surpassed $500 mn by Mar and kept growing in Apr–May. The spin-off/IPO trajectory of Kling matters most for Kuaishou’s valuation, so watch for disclosures on the call.

2) User surprise: Q1 user metrics looked solid, likely boosted by Chinese New Year campaign programming that lifted engagement in the short term. DAU rose 1.2% to 413 mn, with avg. time spent said to be stable, while MAU accelerated 5% (+31 mn) to 770 mn.

However, reported trends differ from some third-party data. Management’s color on the call will be key to explain the user growth drivers.

3) Legacy core under pressure as expected: Growth decelerated in ads and commissions, tied to slower e-comm momentum. Ads saw some lift from external channels (longform/short dramas, AI app ad spend) and AIGC creative assets, while live-streaming is under healthy reset with revenue -14% YoY.

GMV is no longer disclosed this year. Based on e-comm-related revenue trends (ex-Kling, ‘others’ led by e-comm +5.7% with a clear QoQ slowdown), changes in take rate (more influencer rev-share, take rate down) and street models, we estimate GMV growth near ~10%, slower vs. last year.

4) AI spend starts to weigh on GPM: Q1 GPM fell 4pct QoQ, driven by AI compute costs and ad rev-share for IAA short dramas. The company guided full-year Capex of RMB 26 bn last quarter (incl. PPE and right-of-use), +RMB 11 bn YoY, largely for AI, with Kling’s compute the largest piece.

Longbridge Dolphin Research estimates Capex alone could trim full-year GPM by roughly 1–2pct under a 5-year depreciation schedule. This is a directional impact from the stepped-up AI investment.

5) Spend where needed, tighten where possible: On opex, Q1 promotion spend increased while headcount costs were further controlled. Selling expenses rose 4% YoY on marketing, but sales headcount costs fell 8%.

Meanwhile, G&A decreased 7.5%, reflecting team optimization. R&D grew 10%, a slower pace, mainly pulled by an 18% increase in bandwidth/server costs.

6) More room to lift shareholder returns: The company repurchased 11.4 mn shares in Q1 for HK$562 mn, about half of Q4’s pace, with a lower return rate. The slowdown was largely due to the CNY holiday and the earnings blackout period.

Short-term net cash stood at RMB 65 bn (cash + short-term investments – short-term borrowings), and available liquidity was RMB 117.7 bn. New long-term borrowings of RMB 13.7 bn were added in Q1 via $600 mn 5Y + $900 mn 10Y senior notes and RMB 3.5 bn 5Y senior notes.

Kuaishou spent RMB 12 bn in Q1 on PPE depreciation, right-of-use assets and intangibles, or 47% of the full-year Capex budget. Management reiterated the RMB 26 bn Capex target on the call, and current liquidity still allows headroom for more buybacks.

8. Financials at a glance

Longbridge Dolphin Research View

Q1 was in line, but the core businesses outside Kling showed weak underlying trends. The strategy is to drive efficiency and tilt resources toward Kling.

After last quarter’s surprisingly weak Q1 outlook, Kuaishou fell 15% that day, signaling a value reset. Shares rebounded in mid-May on reports that Kling sought a spin-off at a $20 bn valuation. But as headlines faded and Hong Kong markets softened, the stock gave back gains.

At Kuaishou’s current ~$25 bn valuation, using an SOTP framework and a 50% holdco discount (for reduced ownership and internal cross-selling), the implied value of legacy core is under $15 bn (Kling’s fair value post-IPO should exceed $20 bn). That leaves scope for positioning and upside debates.

(1) Kling valuation: A $20 bn mark on >$500 mn ARR implies ~40x P/S, well below leading foundation-model comps, though a discount is reasonable as Kling is not a base model. The multiple looks conservative vs. peers.

If, per media, ARR reaches $1.3 bn by Mar-27, forward implies ~15x P/S, which looks attractive. That said, it requires a high execution bar for Kling’s own growth path.

Kling sits in the global top tier of video models, with domestic rivals mainly ByteDance’s Seedance and Alibaba’s HappyHorse, and global peer Google Veo. Despite strong peers, no price war has emerged amid demand spikes for AI short dramas, and competition intensity has even eased at the margin recently.

Under cost pressure, Seedance raised prices multiple times since Mar (removing low-price options, cutting discounts, upping credits required for premium, and hiking annual fees). This shift makes Kling’s monetization and scaling more straightforward near term.

(2) Core biz. valuation: Excluding the incremental RMB 11 bn compute-related Capex tied to Kling implies core OP would be ~RMB 2 bn higher this year (via higher depreciation). We also trim our e-comm revenue forecasts and now expect ~RMB 15.5 bn in GAAP core OP (ex ‘other gains’).

Applying a 15% effective tax rate yields net profit of ~RMB 13.2 bn, and a $15 bn implied value would be under 8x P/E. We have not carved out Kling-specific opex or included ‘other gains’, making this a conservative take; hence, despite near-term growth headwinds, absolute valuation looks undemanding.

Bottom line: Kling’s spin-off is a rerating catalyst for Kuaishou. Just like BIDU’s Kunlun Chip spin-off plan last year, once the listing process progressed, AI revenue materialized and shareholder returns stepped up, its valuation rose nearly 50%.

Charts below

<End here>

1) Revenue still strong, China beat. MINISO delivered revenue of RMB 5.69bn in 1Q26, up 28.5% YoY, topping the upper end of its prior ‘no less than 25%’ growth guide. By brand, the core MINISO label rose 30% YoY to RMB 3.22bn, with sequential acceleration. Driven by larger-format and themed stores, IP-led products, and same-store recovery, domestic retail outperformed prior market expectations.

TOP TOY grew 51% YoY. The pace moderated from the 80%+ rates of prior quarters, but remains robust, likely reflecting a high base effect kicking in.

Overseas revenue rose 22% YoY to RMB 1.94bn, a clear deceleration. Southeast Asia is in a channel and merchandise reset with softer comps, and based on channel checks, North America saw only low single-digit same-store growth in the quarter, a weak print.

2) Store openings slower, mix improving. The company added a net 80 stores QoQ to 8,565 globally, only ~15% of the full-year target (510–550), implying a slower start and likely acceleration over the next three quarters. Overseas directly operated stores increased by 45 to 745, taking the direct-op mix up to 20.6%. In China, the MINISO brand added a net 25 stores, still skewed to large and ‘amusement park’ formats, while TOP TOY added 21, including 9 overseas, signaling faster international rollout.

3) GPM edged down. To support rapid expansion in markets such as the U.S., the company increased local direct sourcing and emergency replenishment, lifting unit procurement costs. In addition, a lower contribution from high-margin overseas regions weighed on mix. As a result, GPM fell 90bps YoY to 43.3%.

4) AI gains propped up optics; core profit still pressured. Reported operating profit surged 114% YoY, primarily driven by a fair-value gain of approx. RMB 880mn from the MiniMax AI investment. Excluding this boost, adj. OP was RMB 840mn (+14.3% YoY), well below revenue growth. The main drag was sales and marketing: heavy IP licensing fees and front-loaded costs from overseas direct expansion pushed the sales expense ratio up 280bps to 25.9%.

On the positive side, Yonghui contributed RMB 78mn of investment income in the quarter, a notable improvement from a loss YoY and a marginal tailwind.

5) Detailed financials at a glance:

Longbridge Dolphin Research view:

The apparent doubling of operating profit this quarter looks dramatic, but it does not mean the retail core suddenly turned into a cash machine. Rather, prior financial investments caught a favorable updraft, lifting asset values on the balance sheet. Excluding that profit, the core narrative repeats recent quarters: faster top-line growth paid for by higher opex, more direct stores, and heavier IP operations.

On revenue, the market’s premium multiple for MINISO has been largely a bet on overseas expansion—especially North America—rather than the more mature domestic base. This quarter, China growth outpaced overseas as large-format stores and IP products lifted comps, showing domestic operations still have room for optimization rather than being in a pure rent-collection phase. However, the slowdown overseas is a new concern, and as the share of direct-operated stores abroad rises, organizational complexity will increase.

On profit, our prior take stands: MINISO is not a pure self-owned IP model but closer to an ‘Android-like’ open ecosystem, using a long tail of co-brands and licenses to drive traffic. The upside is rapid SKU refresh and hit efficiency, but the downside is that IP licensing fees remain a major line item in sales expenses and are hard to cut meaningfully in the short term. That keeps margin leverage constrained.

On valuation, management guided that adj. net profit growth in 2026 will accelerate vs. 2025. Under a base case of 8% growth (vs. 6.5% in 2025), shares trade at ~9x, which is not demanding and arguably already prices in margin concerns and slower growth. Even so, we see MINISO in a ‘low expectations, wait-for-proof’ zone.

In other words, it is not simply ‘cheap so it rallies.’ Until we see a) the sales expense ratio peaking and b) self-owned IP and overseas direct operations translating into higher-quality earnings, we remain cautious and prefer to wait and see.

Detailed earnings analysis follows:

I. Revenue beat the top end of guidance

1Q26 revenue was RMB 5.69bn, up 28.5% YoY, extending high growth and beating the prior ‘no less than 25%’ guide. At minimum, this indicates front-end demand held up under store expansion, same-store recovery, and IP-driven retail. By brand, MINISO core revenue reached RMB 5.17bn, up 29.6% YoY. Channel checks suggest that large-format stores, themed concepts, IP products, and improving comps drove high single-digit same-store growth in China in Jan–Feb, better than the market had expected.

As the trendy-toy label under MINISO, TOP TOY delivered RMB 600mn in Q4 revenue, up 112% YoY and a quarterly record, a standout performance. In 1Q, TOP TOY revenue was RMB 520mn, up 51% YoY; while still solid, growth slowed significantly from the 80%+ pace in prior quarters, signaling a high base effect.

With domestic growth normalizing, MINISO has staked its ‘second growth curve’ on overseas—especially North America—making offshore trends a key investor focus. Overall, overseas revenue was RMB 1.94bn, up 21.9% YoY, with a sequential slowdown. Based on HF data from foreign brokers, U.S. comps were up 20%+ in Jan–Feb, but credit card data shows growth eased to 19.7% in Mar; a continued deceleration in North America, MINISO’s key overseas engine, is disappointing.

Southeast Asia remains in ‘recovery loading’ mode, and we stay cautious on this region. These markets often face two risks: slower-than-expected consumption recovery, and organizational frictions during the shift from distributor to direct operations. If Southeast Asia underperforms in 2H, overseas growth and margins could face further pressure.

II. Store pace slower, quality first

By quarter-end, global stores reached 8,565, a net add of 80 from end-2025. Against the full-year opening plan of 510–550, this is only ~15% complete, clearly slow and implying a catch-up pace in the next three quarters. Mix-wise, overseas direct stores rose by 45 to 745, lifting direct-op to 20.6%. Domestically, MINISO added a net 25 stores, still skewed to large and ‘park’ formats, while TOP TOY added 21, including 9 overseas, pointing to faster international build-out.

III. Same-store growth eased sequentially

Domestic comps were up high single digits in 1Q26, a slight step down from double-digit growth in the prior quarter, which is reasonable for a chain with 4,000+ stores in China. Overseas comps were up low single digits; drivers were discussed above. Management’s FY26 goal is positive global comps, with low single-digit, healthy growth in China and North America. This implies the 1Q strength was partly transitional, and if store openings do not accelerate, overall revenue growth may face pressure in subsequent quarters.

IV. Lower GPM + higher opex: profit below expectations

GPM declined as the company increased local sourcing and urgent replenishment to support rapid overseas expansion, raising unit costs. A lower mix of high-margin overseas markets also weighed on margins. GPM fell 90bps YoY to 43.3%.

Sales and distribution expenses rose to RMB 1.47bn, up 44.0% YoY, far above revenue growth; the sales expense ratio climbed to 25.9%, up 280bps YoY. In our view, two factors dominate: licensing fees and direct operations. Sustaining product novelty and youth appeal in an IP retail model requires ongoing licensing, marketing, and fresh merchandise; overseas direct expansion also lifts front-end operating costs. Without relief on both, the sales expense ratio is unlikely to fall quickly.

G&A was better controlled: G&A expenses were RMB 300mn, up 22.8% YoY; the G&A ratio was 5.2%, down 30bps from 5.5% a year ago, signaling continued back-office efficiency. That improvement, however, was more than offset by higher front-end spend. On the financing side, interest expense rose to RMB 120mn, up ~40% YoY from RMB 86mn, mainly tied to the Yonghui acquisition loan. Adj. NP was RMB 550mn (+6.7% YoY), well below revenue growth.

<End of text>

Related reading:

Commentary:

Nov 21, 2025 earnings take: MINISO: Revenue ‘printing press’, spend ‘shredder’

Aug 21, 2025 earnings take: MINISO: Big-box to the rescue—has IP retail ‘recharged’?

May 23, 2025 earnings take: MINISO selloff? Without an IP ‘soul’, no next ‘Pop Mart’

Mar 21, 2025 earnings take: MINISO: Profitability up another notch—is IP retail a ‘cash machine’?

Deep dives

MINISO: From ‘10-yuan store’ roots to the endgame of IP-led hits?

MINISO: Has Shein lost its shine? An offline ‘daily goods Shein’ stands out

1. Headline: revenue RMB 99.1bn, -11% YoY. The decline was driven mainly by smartphones and IoT, with legacy biz. (Smartphone x AIoT) revenue down 14.5% YoY.

GPM was 22%, down 80bps YoY, pressured by lower margins in smartphones and autos. The mix and pricing effects weighed on overall profitability.

2. Auto:$XIAOMI-W.HK auto-related revenue was RMB 19.86bn, broadly in line. Shipments were 81k units, with ASP down to RMB 235k.

The volume drop reflected the phase-out of the prior SU7 and production reallocation. The ASP decline stemmed from purchase tax subsidies and a higher mix of lower-ASP ready-to-deliver units.

Auto GPM fell to 20.1%, close to the street at 20.5%, mainly on ASP pressure. The impact included Xiaomi subsidizing purchase tax and selling some lower-priced inventory cars this quarter. Given the margin retreat, Dolphin Research estimates core OP for the auto unit slipped back to a loss of RMB 3.1bn in Q1.

Purchase tax subsidy: for orders locked by 24:00 on Nov 30, 2025 but invoiced and delivered in 2026 due to Xiaomi Auto production or logistics, the company will compensate the difference by deducting from final payment. For the YU7, the purchase tax subsidy impact on ASP is approx. RMB 12k per unit.

3. Smartphones: RMB 44.3bn, -12.5% YoY, in line with street at RMB 44.5bn. Unit shipments fell 19% YoY, while ASP rose 8% YoY.

By market: China shipments fell 34.6% YoY, while overseas shipments fell 12% YoY. Memory shortages directly constrained shipments. With smartphone GPM up QoQ, Dolphin Research believes memory was prioritized to higher-end models, lifting ASP.

4. IoT: RMB 24.7bn, -24% YoY, near street at RMB 25.0bn, hurt by fading state subsidies and memory constraints. Large home appliances were more exposed to policy normalization, with some products previously enjoying RMB 1-2k per-unit subsidies.

5. Internet services RMB 9.5bn, +4% YoY, in line, driven by ads. MIUI users grew 4% YoY, with ARPU up 0.5% YoY.

By region: overseas internet revenue was RMB 2.97bn, while China internet revenue was ~RMB 6.5bn. MIUI users in China continued to grow, while overseas MIUI users declined.

6. Profit:$Xiaomi Corporation.US core OP RMB 2.9bn, adj. NP RMB 6.1bn. Legacy biz. core OP was ~RMB 6.0bn, while autos lost RMB 3.1bn this quarter.

Amid tight memory supply, Xiaomi prioritized higher-ASP models, supporting GPM and legacy core OP QoQ. The auto unit returned to losses on purchase tax subsidies and lower volume.

Dolphin view: share price halved, but the hand is not as weak as feared

Xiaomi’s quarter largely tracked expectations. The YoY revenue decline was concentrated in legacy smartphones and IoT, consistent with prior headwinds.

Operational pressures remain elevated, with double-digit YoY declines in smartphones and IoT. Auto growth and GPM also cooled visibly.

The stock fell from ~HKD 60 to ~HKD 30, already reflecting memory shortages, weak smartphones, and cooling auto momentum. A sustained rebound requires operational improvement, with focus on autos, smartphones, and IoT.

1) Autos: 550k FY target

Q1 auto sales were 81k units, down 44% QoQ. The company discontinued the old SU7 and launched the new SU7, disrupting capacity. With the new SU7, Apr sales rebounded above 30k units, indicating a short-term lift.

Although volumes improved post launch, delivery cycle trends suggest the model is more of a bridge offering than a blockbuster like the YU7. Demand appears more normalized vs. prior peaks.

YU7 delivery cycle has fallen below 10 weeks, implying prior backlogs are largely cleared. The new SU7 delivery cycle lengthened post May Day to about 3 months, which is still within a normal range.

Mgmt. guided to 550k auto units for 2026, but with only 81k in Q1, the remaining three quarters need 470k (i.e., 155k+ per quarter). That is challenging given current SU7/YU7 trajectories.

b) Legacy biz. (Smartphone x AIoT): memory pressure persists, GPM holding at a low base.

i) Smartphones and IoT: both saw sharp declines, driven by memory price hikes and tighter state subsidies. These factors pressured sell-through and margins.

China smartphone shipments fell 34% YoY, as Apple’s iPhone 17 spec bump at same price and memory shortages weighed. Apple shipments in China rose 34% YoY vs. market -3.6% YoY, intensifying competition.

Qualcomm mgmt. noted channel inventory digestion should ease, with China Android likely to bottom next quarter and return to sequential growth in 2H.

End-demand has yet to recover, but the memory shock is priced into the stock, in our view. With smartphone and IoT GPM stabilizing, further operating deterioration looks less likely even if memory stays tight.

All in, legacy segments face notable pressure with GPM anchored at a low level, though hardware margins improved QoQ this quarter. Autos remain the key swing factor with a 550k FY target, and new model performance bears watching.

With multiple headwinds, the share price has halved from ~HKD 60, making bottom-up valuation work more relevant at this stage.

Under a cautious case (smartphone revenue -9% YoY, IoT slightly down), legacy declines remain single-digit. Autos meet the 550k target but see lower ASP and GPM. We estimate 2026 legacy core OP after tax at ~RMB 20.0bn (-16% YoY), and auto revenue at ~RMB 140.0bn (+32% YoY).

Applying SOTP: 15-20x PE for legacy and 1.5x PS for autos (based on 2026 unit target, +34% YoY), implies HKD 600-700bn mkt. cap (HKD/CNY=0.87). That equates to ~HKD 23-27 per share as a cautious-case reference.

Execution against the 550k auto guide is key to this framework. If delivery targets look at risk or get cut, the stock could find a lower trough. For legacy, much of the downside is in the price already, awaiting demand recovery to lift earnings.

Dolphin Research maintains the HKD 23-27 range from last quarter. While shares are now below HKD 30 and downside is narrowing, operating traction has yet to turn. A pullback toward HKD 25 would offer a better margin of safety, in our view.

Below is Dolphin Research’s detailed read of Xiaomi’s print:

I. Group: revenue down again, but GPM has stabilized

With autos added, Xiaomi now reports beyond the prior ‘Smartphone x AIoT’ to include ‘Auto and innovation’. The separate disclosure highlights management focus on autos.

Autos were the main driver that once pushed Xiaomi’s valuation through the HKD 1tn ceiling. Investor expectations remain anchored there.

1.1 Revenue

Q1 2026 revenue was RMB 99.1bn, -11% YoY, broadly in line with the street at RMB 99.6bn. The decline was mainly from smartphones and IoT.

1) Legacy Smartphone x AIoT revenue was RMB 79.3bn, -14.5% YoY. Hardware stayed weak, with smartphones -12.5% YoY and IoT -24% YoY.

2) New auto and related revenue was RMB 19.86bn, +7% YoY, shaped by YU7 deliveries, SU7 discontinuation, and the transition to the new SU7.

1.2 GPM

Group GPM was 22% vs. street 21.3%. Smartphone and IoT margins dipped QoQ, and auto margins continued to trend down, reflecting price-mix and subsidy impacts.

a) Legacy GPM was 22.5%, up 250bps QoQ, as memory was prioritized to higher-ASP products; smartphone GPM rose QoQ to 10.1%, and IoT GPM to 25.2%.

Other legacy lines still posted a small GP loss of RMB 80mn, including AC installation services. If reallocated to IoT, the implied true IoT GPM would be ~24.9%.

2) Auto and new biz. GPM was 20.1%, near street 20.5%. The QoQ drop reflected Xiaomi’s purchase tax subsidies and the sale of some lower-priced inventory cars.

Purchase tax subsidy: orders locked by 24:00 on Nov 30, 2025 but invoiced and delivered in 2026 due to production/logistics will receive compensation via final payment deduction. For YU7, the subsidy impact on ASP is approx. RMB 12k per unit.

II. Autos: maintaining the 550k FY target

Auto revenue was RMB 19.0bn, and including adjacent auto revenue, total reached RMB 19.86bn, broadly in line with the RMB 20bn street view.

Shipments were 81k units as guided, with ASP at RMB 235k, down RMB 15k QoQ, driven by purchase tax subsidies and some low-priced inventory sales. Dolphin Research estimates the per-unit subsidy around RMB 12k.

Auto GPM was 20.1%, down 260bps QoQ. We attribute the decline mainly to the ASP drop, as the subsidy directly compressed unit margins.

Based on shipment trends and delivery cycles, the prior YU7 backlog has largely cleared. The new SU7 lifted volumes near term, but delivery cycles do not suggest runaway demand. The auto biz. has shifted from supply-constrained to demand-driven, making order intake the key KPI.

Current delivery times are under 10 weeks for YU7 and ~13 weeks for the new SU7, both reasonable without large order queues.

The company still targets 550k units in 2026. With only 81k in Q1, the remaining 470k implies 155k+ per quarter, which is demanding. Stronger new models will be needed.

III. Smartphones: protecting price over volume

Q1 2026 smartphone revenue was RMB 44.3bn, -12.5% YoY, reflecting memory tightness and tougher competition.

Dolphin Research breaks down volume and price for smartphones.

Volume: 33.8mn units, -19% YoY, with China down 34.6% YoY in particular. Market share in China fell to 12.6% (down 8ppt YoY), due to memory shortages and heightened competition. Overseas shipments fell 12% YoY, with share down 0.8ppt YoY.

Price: ASP was RMB 1,310, +8% YoY. We see ASP strength as allocation-driven rather than demand-led, with memory prioritized to higher-end models.

Smartphone GPM was 10.1%, up 180bps QoQ. Margin expansion was ASP-led, while shipments remained weak. Faced with memory constraints, the strategic sequence appears to be ASP > GPM > volume.

IV. IoT: clearly hit by tighter state subsidies

Q1 2026 IoT revenue was RMB 24.7bn, -23.7% YoY, due to subsidy roll-off and memory issues. Large appliances were more affected by policy normalization, with prior per-unit subsidies of RMB 1-2k for some SKUs.

IoT GPM was 25.2%, up 510bps QoQ, helped by higher margins and mix in certain overseas lifestyle products, and margin recovery in China large appliances and tablets.

V. Internet services: smartphone shipment drag slowed growth

Q1 2026 internet services revenue was RMB 9.5bn, +4% YoY. The main driver remained advertising.

a) Ads: RMB 7.1bn, +7% YoY, a slowdown from prior double-digit growth. The core ad surfaces are app distribution and pre-installs, which function like a distribution tax on app publishers.

As pre-installs track smartphone shipments, the sharp shipment decline directly hit that revenue line. This diluted the ad growth rate.

b) Value-added services: mainly game distribution, Youpin e-commerce, and Xiaomi Finance. Revenue was ~RMB 2.4bn, roughly flat YoY, indicating stability.

Structurally, internet services are still tied to hardware volume. Xiaomi has regrouped this under Legacy in its revised disclosure. Sustainable internet monetization hinges on the device base, and with shipments down, growth pressure remains.

VI. Overseas: growth across both hardware and internet

Q1 2026 overseas revenue was RMB 39.6bn, +3.5% YoY. With domestic demand soft, the overseas mix rose back to ~40%.

Specifically, overseas internet revenue rose 10% to RMB 3.0bn, while overseas hardware revenue grew 3% YoY. This points to a recovering demand backdrop for overseas IoT and related categories.

VII. Profit: legacy improved, autos back to loss

Q1 2026 total opex was RMB 18.9bn, with opex ratio up to 19%. Auto and AI innovation opex was RMB 7.1bn, stable around that level.

Ex-auto, legacy opex was ~RMB 11.76bn, +10.5% YoY. Legacy opex ratio fell to 14.8%, with higher R&D spend YoY, and R&D headcount reached 26k, up for 8 straight quarters.

Adj. NP was RMB 6.1bn, but Dolphin Research does not support Xiaomi’s adjustment approach, which keeps financial income and investee dividends in adj. NP. Even if sustainable, these are not core operations and do not evidence long-run earnings power.

We focus on core OP (revenue - COGS - opex), which better reflects recurring operating profitability. It is a more consistent gauge of operating health over cycles.

Q1 core OP was RMB 2.9bn with a 3% core OPM. The decline YoY was driven by lower revenue, while QoQ margin improvement came from hardware GPM recovery. By segment, legacy core OP was ~RMB 6.05bn (up RMB 3.9bn QoQ), while autos posted a RMB 3.1bn core OP loss (down RMB 4.1bn QoQ).

1) Walmart U.S.: the e-comm flywheel keeps accelerating. Core U.S. comps rose 4.1%, steady overall. Mix-wise, transactions accelerated to +3.0% (vs. +1.6% YoY), while average ticket slowed to +1.1% (vs. +2.8%), signaling a shift from inflation-led price lift to value-led traffic gains and improving growth quality.

E-commerce grew 26%. Within that, Walmart Connect (ads) rose 44% and 3P Marketplace jumped 50% (fastest in ~10 quarters), underscoring a broad-based acceleration in high-margin flywheels.

2) Sam’s Club (U.S.): higher volume, lower ticket. Ex-fuel comps rose 3.9% in North America, with transactions up 6.2% and accelerating QoQ, while average ticket fell 2.2%. The ticket pressure mainly reflects a higher mix of lower-ticket, high-frequency grocery and trade-down to Member’s Mark private label replacing some higher-priced brands.

E-comm grew 23%. Reserved delivery surged over 90%, contributing roughly half of e-comm growth, suggesting Sam’s is expanding into higher-frequency, more immediate-use cases.

3) Walmart International: China Sam’s remains the brightest star. By region, China grew 22% YoY, with Sam’s still the No.1 growth engine and transactions sustaining double-digit growth, and an exceptionally strong Chinese New Year. This indicates China Sam’s is still in a three-fold resonance of store expansion, strong same-store growth, and rising digital penetration.

Walmex (Mexico & C. America) grew 4.1% YoY with some deceleration QoQ, which we attribute to 3P drag. Canada grew 7.4% YoY, benefiting from higher e-comm penetration and a modest QoQ acceleration.

4) Higher fuel costs constrained margin release. Consolidated GPM expanded 20bps YoY to 25.1%. Despite upstream cost inflation and largely stable shelf prices, the modest GPM improvement points to a higher mix of high-margin advertising.

On opex, surging fuel prices lifted fulfillment costs, pushing SG&A up 30bps YoY to 20.9%. Operating income reached $7.5bn (+5% YoY), lagging revenue growth.

5) Key financials at a glance:

Dolphin Research view:

On the positives first — the variables anchoring Walmart’s valuation have broadly continued to improve. These are the true drivers of the multiple.

a) E-comm remains on a high-growth track: global e-comm rose 26%, with U.S. e-comm contributing 530bps to comps (vs. 520bps in Q4), shifting from an incremental channel to a core engine.

b) Ads, the highest-margin business, kept accelerating: global ads rose 37%, with Walmart Connect +44% and faster QoQ, implying traffic monetization is far from a ceiling. This strengthens the structural margin story.

c) Marketplace sales up nearly 50%, a 10-quarter high — more 3P sellers broaden SKUs, lift search and conversion, and reinforce ad monetization. E-comm, ads, and Marketplace are three blades of the same flywheel, and all three are spinning faster.

The sticking point is guidance not moving higher and cautious commentary on the call:

Management flagged that elevated fuel is squeezing consumer budgets, that the Q1 tax refund tailwind will fade from Q2, and that retail price inflation could tick higher in H2 if the high-cost backdrop persists. The market reads this as a tougher macro and heightened consumer strain, which goes straight to the valuation debate.

Before 2022, Walmart traded at a 20–30x retail multiple; in less than two years since H2 2024, it rerated to 45x+. The reset rests on Walmart’s shift from a traditional retailer to an omnichannel fulfillment plus high-margin tech-enabled platform — namely 3P Marketplace, ads, and membership, all high-margin, high-growth, asset-light, driving OP growth to outpace revenue and loosening the low-margin, low-beta retail multiple constraints. Meanwhile, 4,600+ stores have evolved into fulfillment hubs, enhancing resilience and cyclicality management.

But a tech-like multiple assumes sustained OP growth outpacing revenue and repeated guidance raises. With Q1 OP up 5% vs. revenue up 7.3% and no full-year guide-up, both pillars were absent this quarter, prompting a valuation check.

On valuation, based on full-year OP growth guidance of 6–8%, FY27 implies ~42x, down from 45x+, partly digested but still not cheap.

Given high fuel as a likely transitory headwind, and management indicating Q1 is the low point for OP growth, profitability should improve sequentially. With the core flywheel variables — e-comm, ads, Marketplace, membership — all accelerating and still early in profit mix expansion, the story remains intact. If Q2 beats and the FY guide is raised, the multiple could revisit ~45x; in a downside case with prolonged high fuel and weaker consumption, the multiple may compress toward 38–40x. Detailed modeling to follow.

Detailed read-through below:

I. Investment framework

Per disclosures, revenue growth splits into three segments: Walmart U.S., Walmart International, and Sam’s Club U.S. Each segment breaks down further into comps (traffic × ticket), new/closed stores, and e-comm contribution.

(1) Walmart U.S. is the core franchise, with FY26 revenue of ~$483bn (~68% of total). The format is led by Supercenters, complemented by Neighborhood Market and Discount Stores. Comps split into two drivers: transactions and average ticket.

E-comm is the primary growth engine for this segment, now ~20–23% of segment revenue. Internally, it includes 1P first-party sales, 3P Marketplace, Walmart Connect (retail media), and Walmart+ (paid membership), with ads and membership the core profit engines.

(2) Walmart International spans Walmex (Mexico & C. America), China, and Canada, with both Walmart stores and Sam’s Clubs overseas. FY26 revenue was ~$130bn (~18% of total). E-comm penetration varies: China is over 50% (largely Sam’s online), Flipkart in India operates independently, remains loss-making but narrowing losses, while Walmex penetration is ~8–9% and is early in rolling out the U.S.-style ads + Marketplace playbook.

(3) Sam’s Club U.S. is a membership warehouse model, with FY26 revenue of ~$93bn (~13% of total). Revenue relies on steady comps from clubs, while profits leverage membership fee scale and high renewal rates, with Scan & Go and Club Pickup driving digital penetration. In competition with Costco, Sam’s pricing is more aggressive, but product curation and in-club member services still trail somewhat.

II. Walmart U.S.: the e-comm flywheel keeps accelerating

2.1 Volume is the core driver of comps

North America delivered $117.2bn in Q1 FY27 revenue, up 4.5% YoY, steady overall.

Ex-fuel comps rose 4.1%. Transactions rose 3.0% and average ticket rose 1.1%, pointing to a volume-led recovery.

Transactions: we believe the key incremental driver remains ongoing downtrading by higher-income households (>$100k). With sustained upgrades in fresh and private label quality and 3-hour express delivery building out omnichannel fulfillment, convenience and experience keep improving, drawing more high-income households to Walmart.

By category, low-single-digit growth in grocery aligns with resilient staples. General merchandise posted mid-single-digit growth, accelerating from low-single digits last quarter and delivering its strongest share gains in nearly five years.

This matters because only share gains in general merchandise signal rising platform attractiveness. It is a stronger proof-point than staples alone.

Average ticket: food inflation is easing, and Walmart stepped up Rollbacks to drive value for core middle- and lower-income shoppers and take share. Both factors weighed on ticket growth but strengthened wallet share.

Store openings/closures: Walmart added three Supercenters in Q1 FY27, reflecting disciplined expansion. Management indicated the focus is on remodels, digital upgrades, and fulfillment enhancements to lift unit productivity.

2.2 E-comm still in high-growth mode

North America e-comm rose 26% YoY, contributing ~530bps to comps. With U.S. comps at 4.1%, ex-e-comm, physical comps would be negative, reaffirming that the U.S. growth engine has fully shifted to digital.

Within 1P, orders fulfilled within three hours now account for 36% of store-fulfilled orders. This directly validates rising demand for convenience and immediacy in shopping.

Marketplace grew ~50%, a 10-quarter high, indicating the supply side of the ecosystem is improving rapidly. Greater SKU breadth boosts onsite traffic value and ad pricing power.

Ads: as the pivotal lever in Walmart’s shift from low-margin retail to a higher-margin platform, Walmart Connect grew 44% YoY with faster QoQ momentum. Beyond search/display, Walmart began piloting Sponsored Prompt within its in-house AI shopping assistant Sparky, embedding paid product recommendations in natural language answers.

This is an incremental ad surface independent of existing search inventory, avoiding cannibalization. While still early with limited near-term revenue, we believe Sparky’s high-intent conversational context can become a new growth driver for Walmart Connect over time.

Membership: Walmart+ membership fee revenue rose double digits, with net adds hitting a record high in Q1. Benefits like free delivery, fuel discounts, and convenience tie users to Walmart, driving higher frequency and stability and improving ad and recommendation efficiency.

III. Sam’s Club: a ‘more-for-less’ volume story

Sam’s Club U.S. net sales were $23.4bn, up 6.1% YoY and accelerating QoQ. Ex-fuel comps rose 3.9%, with transactions +6.2% and average ticket -2.2%.

Transactions +6.2%, up from +4.8% last quarter, driven by two factors. First, a fuel-station traffic halo as oil spiked post Iran-related tensions: Placer.ai shows Sam’s fuel stations saw weekly traffic up +10–13% YoY in Mar and over +16% in mid-Apr, lifting overall club traffic.

Second, express e-comm fulfillment kept surging. On Apr 22, Sam’s launched an enhanced 1-hour Express Delivery (from 3 hours), covering 600+ clubs, with ~65k early orders averaging ~55 minutes and some in just minutes. Members who frequently use rapid delivery spend ~2x more annually than average members.

Price/ticket: the ticket decline reflects faster growth in low-ticket, high-frequency grocery vs. high-ticket general merchandise and substitution toward Member’s Mark private label. This creates structural pressure on average ticket.

But a lower ticket is not bad — it means members buy more for less, reinforcing value perception and supporting renewals and member growth. This strengthens the long-term unit economics.

E-comm rose 23%, with reserved delivery up over 90% and contributing about half of e-comm growth. This shows Sam’s is moving into more frequent, more immediate demand scenarios.

No new clubs opened in Q1. With recent fee hikes to $60 for base and $120 for Plus, we believe the focus is on protecting renewal rates through the price change.

IV. Walmart International: China Sam’s still shines brightest

Walmart International net sales were $35.1bn, up 18.0% YoY (+10.1% at cc) and accelerating QoQ. Importantly, e-comm rose 27% and membership revenue rose 31%, suggesting the U.S. formula of e-comm + ads + membership is being replicated internationally.

Walmex (Mexico & C. America): 3P became a drag. Walmex revenue was $12.2bn, up 4.1% YoY.

Walmex pre-released Q1: comps +3.1% split into +4.0% ticket and -0.9% traffic. We think high inflation reduced non-essential trips among lower-income shoppers, and the rapid rise of hard discounters siphoned off highly price-sensitive long-tail customers, though Walmart’s traffic decline was milder than the industry (-2% to -3%).

In e-comm, 1P GMV stayed strong at ~20% growth on instant delivery, but 3P Marketplace GMV collapsed from +15% in Q4 to -14%. Management cited delistings of some large 3C sellers due to compliance/credential issues as the core driver.

Beyond the short-term factor, Walmex 3P relies heavily on a few mega sellers (notably large electronics distributors), with thin long-tail SKU breadth and slower merchant onboarding. This has led to share loss in all-category retail traffic and mindshare, making 3P ecosystem build-out a key focus.

Ads did well, with Walmart Connect revenue accelerating to +33% (vs. +5% last quarter) on World Cup advertising brought forward. This is a positive signal for retail media scaling.

China: Sam’s remains the primary growth engine

China revenue rose 22.3% YoY at cc to $8.0bn, with e-comm +31% and digital sales penetration at 50% (+335bps YoY). These are standout metrics within Walmart and across global retail.

Sam’s drove the bulk of growth: transactions stayed in double digits, CNY was very strong, and nine new clubs opened in the last 12 months. This confirms a threefold resonance of store additions, strong comps, and deepening digital penetration in China Sam’s.

At 50% digital mix, China Sam’s has embedded dark stores, on-demand delivery, app ordering, and membership deeply into the model. That boosts club catchment radius, purchase frequency, and inventory efficiency vs. pure offline warehouses.

Canada revenue rose 7.4% YoY at cc to $5.5bn, with e-comm +38% (vs. +31% in Q4). All categories grew, indicating a phase of rapid e-comm penetration gains.

V. Gross margin: best in U.S., stable Intl., near-term pressure at Sam’s

Consolidated GPM expanded 20bps YoY to 25.1%. Despite higher input costs and largely stable shelf prices, the GPM uptick indicates a rising mix of high-margin ads.

By segment, Walmart U.S. GPM rose 29bps YoY, Intl. was flat, and Sam’s fell 26bps. The higher the mix of ads, membership, and digital monetization, the greater the margin improvement.

VI. Higher fuel costs constrained profit release

Opex rose as higher fuel drove up fulfillment costs, lifting SG&A by 30bps YoY to 20.9%. By region, Intl. saw the biggest OP margin pressure due to strategic upfront omnichannel investments in Canada and Mexico and longer logistics chains, making it more exposed to fuel spikes.

Operating income was $7.5bn (+5% YoY), slower than revenue growth. This mismatch explains the market’s focus on cost headwinds this quarter.

<End>

Dolphin Research past coverage on ‘Walmart’:

Deep Dive

Apr 21, 2026: ‘Walmart: What makes the trillion-dollar retail juggernaut tick?’

In detail: See below.

1. Games revenue beat ‘conservative’ street expectations: Q1 games revenue (ex-CC Live) was RMB 25.1bn, up 7% YoY and above consensus. The market had expected a softer print given no new titles this quarter and a tough base last year, with many modeling sub-5% growth.

In practice, NetEase delivered 7% growth on strength from ‘Fantasy Westward Journey’ PC ‘Changyou’ servers, the overseas roll-out of ‘Where Winds Meet’, and operations recovery for ‘Identity V’ and ‘Eggy Party’. Also, last quarter’s 22% YoY growth in deferred revenue already pointed to support for Q1 revenue.

2. Deferred revenue stayed robust: deferred revenue rose 22% YoY at quarter-end, beating the 17% expected by institutions. Longbridge Dolphin Research estimates gross billings rose 18% QoQ, not as strong as the exceptional Q1 last year but ahead of typical seasonal patterns.

Longbridge Dolphin Research expects high-base pressure to persist into Q2–Q3, but heavy mobile launches in 2H should help, including Q3’s ‘Sea of Oblivion’ (street est. full-year billings across platforms of RMB 3–5bn) and ‘Infinity’ in late Q4 or early next year (street est. RMB 10bn). These could support sequential acceleration in game revenue.

3. GPM saw a sharp uplift: the biggest highlight, in our view. Q1 GPM reached 69.4%, up over 500bps YoY and QoQ, largely driven by games and live streaming.

Games-related GPM improved to 75%, up 6ppt YoY. Drivers included a higher mix from self-developed, high-margin PC titles and lower channel costs as the company steers users to official web top-ups and benefits from lower app store take rates, a long-anticipated margin lever.

4. Staff costs were materially optimized: opex was tightly managed in Q1; R&D grew slowly, and S&M—given YoY base effects—declined QoQ in line with seasonality, while G&A fell 33% YoY.

G&A has been shrinking for two years from a low base, and the further drop suggests significant headcount optimization or pay adjustments in Q1. SBC also points to this: despite NetEase’s share price being ~10% higher YoY, SBC nearly halved YoY. Media reported sizable cuts to outsourced teams in Mar, and this optimization may not be fully reflected in Q1; watch the call for details.

Core operating profit was RMB 12.7bn, up 21% YoY, with margin up 5ppt YoY. Adj. net profit (adding investment gains/losses and FX, and primarily excluding SBC) was roughly flat YoY, as lower investment gains (PDD and Alibaba declined in Q1) and FX losses on overseas revenue offset operating upside.

5. Buybacks increased, but overall shareholder returns remain average: the company plans a cash dividend of $0.144 per share ($0.72 per ADS), totaling $460mn with a ~30% payout, broadly stable. Buybacks were $127mn in Q1, already above all of last year; management typically paces repurchases with market cap moves, and $2.9bn remains authorized (three-year window).

Annualizing Q1 dividends and buybacks implies ~$2.35bn of shareholder returns for the year, about 3.3% of the ~$70bn market cap intraday, which is not high.

6. Key financial metrics at a glance

Longbridge Dolphin Research View

Despite a lull in new launches, this print stood out not only because expectations were conservative but also due to the marked improvement in games GPM. This showcases the earnings power that leading studios with multiple evergreen IPs can unlock.

The implications are: (1) amid an industry-wide push for lower channel fees, top studios have greater bargaining room. (2) With a portfolio built over years, user stickiness is stronger, making it easier to steer users to official downloads and payments, leaving less revenue to external channels.

(3) Self-developed evergreen IPs also save on external licensing costs; for example, titles based on Marvel or the prior Harry Potter franchise required large royalties. In Q1, with the resurgence of NetEase’s top evergreen IP ‘Fantasy Westward Journey’ on PC, the profit potential materially surfaced.

For NetEase, the PC side carries no heavy channel costs and relies on self-owned IP, keeping marginal costs very low. Pure development costs are mainly staff compensation booked in R&D, front-loaded with lower ongoing needs thereafter, so a large share of billings drops to profit.

That said, GPM will not extrapolate linearly. As new mobile titles scale in 2H, channel shares will eat into margins, though over a longer cycle, the trend should be upward with volatility.

NetEase’s 520 game showcase yesterday featured fewer new titles vs. prior years and focused more on secondary development of evergreen franchises, which may disappoint some. But the ‘Changyou’ servers of Fantasy Westward Journey validate the playbook: keep the core gameplay, layer in new content styles and monetization aligned with current user tastes, and potentially reignite an evergreen title’s second spring.

That does not mean new titles do not matter. New genres and themes help reach new cohorts and fill category gaps. Since the lull began last 2H, the stock has struggled to re-rate, and delays to ‘Sea of Oblivion’ and ‘Infinity’ last quarter further tested market patience.

At this point, ‘Sea of Oblivion’ is likely still set for Q3 (summer break, pre–back-to-school), but whether ‘Infinity’ launches this year is uncertain; the 520 event gave no firm update. Conservatively, Q2 still faces pressure, Q3 eases as the base normalizes, and the true new-release cycle approaches from Q4 onward.

From now to Q4, evergreen titles provide a floor, but the valuation re-rating may take time. Typically, the market turns constructive 1–2 months ahead of a major public beta. At a ~$70bn market cap vs. the street’s FY2026 adj. profit estimate of ~RMB 40bn, ~12x P/E looks near a sentiment trough, suggesting a window to build into the next product cycle.

Separately, the process for a dual-primary listing adjustment and stepwise Southbound inclusion is underway; any progress this year could serve as a near-term catalyst.

Detailed Analysis

1. Games beat conservative expectations

Q1 games and VAS revenue was RMB 25.7bn, with core games at RMB 25.1bn, up 7% YoY. With no new launches in the quarter, growth was driven by legacy titles.

Deferred revenue rose 22% YoY in Q1 but slowed QoQ. Gross billings are estimated up 18% QoQ, less than last year’s 28% but better than historical seasonal averages; YoY growth slowed to 2% on the tough base. To return to healthy growth, new titles are needed.

(1) Mobile: Q2 remains a lull; heavy launches arrive in 2H.

The lack of tentpoles keeps mobile billings under pressure. Q1 saw incremental contribution from three Q4 launches, but none were priority projects.

Among evergreen titles, ‘Eggy Party’ rebounded late last year on content updates and events, driving slight YoY billings growth. ‘Identity V’ stabilized in Q1 after prior pressure from competitors like ‘Paranormal Squad’, though still below last year’s high base.

‘Fantasy Westward Journey’ benefited from renewed PC ‘Changyou’ server buzz, while ‘Where Winds Meet’ maintained high stickiness on product quality. Other evergreens such as ‘Justice’ Mobile continued to trend down.

Looking to Q2 and the full year, Longbridge Dolphin Research expects: (1) Q2 still faces a tough base with no major new launches; only the PC early access of life-sim ‘Star Painter: Sunny Day’ went live in late Apr, so evergreen titles must carry the load.

(2) 2H should improve sequentially, with ‘Sea of Oblivion’ in Q3 and ‘Infinity’ by late Q4 or next year. The company places high emphasis on both; street FY1 billings expectations are RMB 3–5bn and RMB 10bn, respectively.

For ‘Infinity’, a modern-city anime open-world game, expectations have tempered after subpar debuts from ‘Honor of Kings: World’ and ‘Different Circle’. Given likely late-2026 timing, its full-year revenue contribution could be limited; near term, focus remains on ‘Sea of Oblivion’ and globalizing legacy titles.

‘Sea of Oblivion’ mixes Age-of-Sail naval combat, open-world exploration, and extractions, developed by the ‘Identity V’ team Joker Studio. Early-year tests met expectations; watch the call for more updates.

(2) PC/console: strength in evergreen and recent titles: performance likely remained solid in Q1, though growth moderated on the rising base. ‘Fantasy Westward Journey’ (concurrent users hit a record 3.9mn in Mar) and ‘Where Winds Meet’ should have offset softness in ‘NARAKA: Bladepoint’.

For other Marvel titles, ‘Marvel Rivals’ saw a clear rebound in S6 early this year vs. the prior two seasons, but S7 from Mar was average, while the broader Marvel slate was stable.

2. Non-core segments face varying pressures

Non-gaming segments underperformed and missed expectations. Structurally, both industry shifts and competitiveness pose medium- to long-term challenges.

Youdao’s education revenue grew only 4% despite a low base, still dragged by a faster decline in learning devices. Ads and online courses (Youdao Lingshi for K-12, Little Turing for quality education, and AI tools under the Ziyue LLM) also slowed notably.

Cloud Music growth recovered from a low base with GPM improving to 37%. Other innovations fell 5% as Yanxuan continued to contract amid e-commerce competition.

Note: Non-core segments are not Longbridge Dolphin Research’s primary coverage; please refer to the charts for details.

3. GPM is the highlight; opex belt tightened again

GAAP operating profit (ex investment gains, interest, FX and other non-core items) was RMB 12.6bn, up 21% despite a tough base. OP margin rose 5ppt QoQ to 41% as both COGS and opex shrank materially.

By line item, total COGS and G&A fell YoY, while S&M rose sharply YoY on a very low base from last year’s anti-corruption phase and no new launches then. Although there were no new titles this quarter, evergreen operations still required normal promotion.

Q1 GPM reached 69.4%, up over 500bps YoY and QoQ, driven mainly by games and live streaming. Games-related GPM rose to 75%, up 6ppt YoY, helped by a higher mix of self-developed, high-margin PC titles and lower channel costs (more web top-ups; reduced app store take rates).

Non-GAAP net profit was RMB 7.2bn. Because only SBC is excluded from GAAP net income, investment losses from holdings in PDD and Alibaba and FX losses on offshore revenue were included; even so, stronger cost/expense discipline drove an above-consensus adj. net result.

4. Bigger buybacks, but shareholder returns still average

Net cash was RMB 167.5bn ($23.4bn) at Q1-end, up RMB 4bn vs. Q4. The company plans a $0.144 per-share dividend ($0.72 per ADS), totaling $460mn with a ~30% payout, broadly steady; buybacks were $127mn in Q1 (chart shows cash outflow), exceeding all of last year, with $2.9bn remaining authorized over three years.

Annualizing Q1 distributions implies ~$2.35bn in full-year shareholder returns, ~3% of the $74bn market cap at yesterday’s close, which is not high.

<End here>

Longbridge Dolphin Research on NetEase: archive

Earnings season (past year)

Feb 11, 2026 earnings call notes: NetEase (Trans): ‘Sea of Oblivion’ shifted to Q3; ‘Infinity’ still in polishing

Feb 11, 2026 earnings quick take: NetEase: the ‘pig farm’ still needs patience

Nov 20, 2025 earnings call notes: NetEase (Trans): focusing resources on high-conviction products

Nov 20, 2025 earnings quick take: NetEase: the last kneel before the ‘pig cycle’?

Aug 14, 2025 earnings call notes: NetEase (Trans): rich pipeline; will keep optimizing marketing allocation

Aug 14, 2025 earnings quick take: NetEase ‘bouncing around’? Veterans still have hard skills

1. Vehicle revenue slightly missed but remained up sharply YoY: In Q1, vehicle sales revenue was RMB 25.5bn, just below the market’s RMB 26.0bn estimate but up 112% YoY.The miss was mainly due to ASP of RMB 273k vs. RMB 277k expected, though ASP rose by RMB 20k QoQ as the high-priced ES8 accounted for over half of the mix, lifting overall pricing.

2. Vehicle GPM continued to improve, bucking seasonality: Q1 vehicle GPM was 18.8%. Despite weaker scale benefits and higher procurement costs, Nio maintained a high margin level and even improved 70bps QoQ, above the market’s 18.2% and prior guidance for flat QoQ.Mix shift toward higher-end models offset headwinds from softer scale and rising raw materials.

3. Opex disciplined; cost-down and efficiency gains ongoing: With Q1 deliveries down 33% QoQ to ~83.5k units, management had expected a small operating loss.However, better-than-expected GPM and lower R&D spend (RMB 1.9bn, below consensus and guidance) led to a narrower loss: OP was -RMB 0.3bn vs. -RMB 1.0bn expected, and OPM was -1.2%, down 3.5ppt QoQ but ahead of the -3.8% consensus.

Overall, Nio demonstrated resilient profitability through the off-season. Strong volumes of the high-priced ES8 supported margins, while ongoing cost and expense controls created more room at the bottom line.

All in, Q1 was solid: revenue was a slight miss, but margins held at a high level through seasonality.The core driver was the ES8’s strong ramp, which shored up GPM and offset headwinds from weaker scale and higher procurement costs, while tighter cost control widened profit headroom and kept losses slightly better than expected.

More importantly than the print is Q2 guidance. With two major models — ES9 and ONVO L80 — both starting deliveries in May, Nio’s upbeat volume and revenue guide implies healthy order intake for these launches.

① Q2 deliveries are guided at 110k–115k, above the market’s ~106k. With only ~29.4k delivered in Mar, the guide implies average monthly deliveries of 40k–43k in May/Jun, a very strong pace.

Notably, ES8 lead time shortened from 4–5 weeks in Mar to 2–4 weeks in Apr, indicating rapid digestion of backlog. Thus, Q2 upside should mainly come from ES9 and L80 (and, to a lesser extent, refreshed 5566 models), corroborating ample order reserves and delivery cadence aligned with demand.

② Q2 total revenue is guided at RMB 32.9bn–34.4bn, well above the ~RMB 30bn consensus, with both volume and ASP beating expectations. The guide implies Q2 vehicle ASP of ~RMB 270k, roughly flat vs. Q1’s RMB 273k.

Given ES8’s Q1 mix share reached 54% but its backlog and deliveries are now declining QoQ, maintaining a high ASP in Q2 hinges on the higher-priced ES9 stepping in. Management confidence in ES9’s monthly run-rate appears strong; if ES9 (pre-sale RMB 528k–658k) sustains 5k+ units per month, Nio’s GPM should remain elevated, with mix upgrade offsetting BOM inflation and potentially pushing margins higher.

For full-year 2026:

Management previously guided for 40%–50% YoY delivery growth (approx. 456k–490k units), backed by a heavy product cycle starting in Q2.

Nio ES9 (on sale and delivering from May 27): a 5.4m all-new flagship carrying ET9 core tech, with pre-sale pricing of RMB 528k–658k. Dealer checks indicate refundable RMB 5,000 deposits could total ~40k units nationwide.ES9 emphasizes spaciousness and first-class ride experience, with premium interior and design that reflect Nio’s brand aesthetics and flagship positioning; dealers expect a steady 5k+ units per month, potentially repeating ES8’s success.

ONVO L80 (on sale and delivering from May 15): the 5-seat version of L90, priced at RMB 242.8k–279.8k. It targets incremental demand in the RMB 200k–250k segment with a lower entry point.

Nio ES7 (Q3 launch): a 5-seat version of ES8 to further consolidate the high-end base.

Among existing models, the previously popular 5566 series has been refreshed. ES8, successful in 2025, will contribute a full 12 months in 2026, and ONVO L60 and L90 are also slated for facelifts.

As such, Nio’s 2026 pipeline looks robust, underpinning confidence in high delivery growth guidance. The strategic focus is clearly pivoting toward ‘big SUVs’ and further premiumization.

Against a slower NEV market in 2026 due to purchase tax roll-off (industry growth expected at only 5%–15%), Nio’s BaaS model stands out. Because batteries are excluded from the vehicle invoice, the taxable base is much lower.This means Nio buyers pay less purchase tax vs. peers in the same segment, helping cushion macro headwinds from the tax rollback.

Guidance on margins and profitability is also constructive:

Nio expects vehicle GPM to improve from 18% in Q4 2025 to around 20% in 2026. Key drivers include the launch of higher-margin ES9, scale-driven cost reductions (further magnified by ONVO L80, Nio ES7, etc.), and margin uplift from the refreshed ONVO L60 SUV.

On expenses, Nio continues to tighten the belt: 2026 full-year R&D is expected to be flat vs. Q4 2025 annualized levels, and SG&A ratio is guided to around 10% of revenue. With higher volumes, better vehicle margins, and strict control of R&D and SG&A, Nio aims to achieve full-year profitability in 2026.

In sum, Longbridge Dolphin Research believes that, while Q1 volumes were pressured by seasonality, pricing and margin resilience from an improved mix have become more evident.

For 2026 full-year deliveries, Nio benefits from some buffer against the tax rollback via BaaS and a strong pipeline (ES9, L80, ES7). However, with the broader NEV industry expected to grow only 5%–10% YoY, we prefer to stay cautious versus management’s 40%–50% guide to 456k–490k units and will monitor execution.

Base-to-neutral case: factoring in tougher competition, price-war risk, and ramp uncertainties, we assume 20%–30% YoY growth to 390k–430k units in 2026. With ASP rising from RMB 236k in 2025 to ~RMB 248k in 2026 on greater mix of higher-margin large vehicles, and continued growth in services revenue, we estimate total revenue of RMB 110bn–119bn (+26%–36% YoY).On a neutral 1.0x 2026 P/S, implied market cap is RMB 110bn–119bn, offering ~18%–27% upside vs. the current HK valuation of ~RMB 93.5bn.

Upside case: if the high-priced ES9 sustains 5k+ units per month (lifting margins) and the volume driver ES7 reaches 10k–12k per month, deliveries could rise ~40% YoY to 456k (near the low-end of guidance). Under this scenario, at 1.2x 2026 P/S, implied market cap would be ~RMB 150bn, or ~60% upside vs. current levels.

Thus, whether Nio can break through its valuation ceiling hinges on the ES9 and ES7 achieving volume scale — the former sets margin height, the latter drives volume and revenue. If both deliver, the profit inflection and re-rating would have a solid foundation.

I. Vehicle GPM ‘strong despite off-season’

As the most critical metric each quarter, we first look at vehicle profitability. Nio had guided that Q1 vehicle GPM would be flat QoQ vs. Q4 2025’s 18% as ES8 (GPM >20%) gained mix, offsetting cost inflation and smaller scale, aligning market expectations at 18.2%.

Actual Q1 vehicle GPM was 18.8%. Despite weaker scale and higher procurement costs, margin stayed high and rose 70bps QoQ, beating expectations and prior guide, as higher mix lifted ASP and offset cost drag.

1) ASP: continued QoQ improvement, driven by high-priced ES8

Q1 ASP was RMB 273k, slightly below the RMB 277k consensus, but up a notable RMB 20k QoQ from RMB 253k. The key driver was continued scale-up of the high-priced ES8 SUV (RMB 406.8k–456.8k list).Although ES8 orders in Q4 2025/Q1 2026 carried RMB 15k/10k discounts, ES8’s mix jumped 22ppt QoQ to 54%, lifting overall ASP and offsetting promotional pressure.

Meanwhile, the premium Nio brand mix rose 16ppt QoQ to 70%, reinforcing the upward price structure.

2) Unit cost: up QoQ on weaker scale and raw material inflation

In Q1, unit cost was RMB 222k, up ~RMB 14k QoQ, driven by higher ES8 production cost and headwinds from weaker scale and raw materials. Even so, vehicle margins remained stable.

① Off-season scale effect: Q1 sales were ~83.5k units, slightly above the guided range but still down 33% QoQ amid seasonality and tax headwinds, diluting scale and lifting per-unit overhead.② Procurement costs rose, but near-term impact was limited: pressures came from memory, raw materials, and batteries, but were contained because: a) long-term contracts delay major cost swings, likely showing up in Q2; b) the swap model’s low-cost battery inventory buffered the Q1 battery price spike.

③ Platform and supply-chain cost-down continued to offset: the NT3.0 platform with 900V architecture and zonal controllers drove high hardware integration, lowering materials and manufacturing costs. Coupled with supplier savings and in-house tech (replacing Nvidia Orin-X with the self-developed NX9031, cutting ~RMB 10k per vehicle), ES8 maintained 20%+ GPM even with a RMB 10k discount in Q1.

3) Unit GP: up QoQ, margins ‘strong despite off-season’

Q1 unit GP was RMB 51k, up RMB 5k QoQ from RMB 46k, and vehicle GPM reached 18.8%, up 70bps QoQ. Again, ES8’s strong ramp offset weaker scale and higher procurement costs.

II. Q2 volume and revenue guides ‘beat’, implying healthy new-order intake

1) Q2 deliveries guided at 110k–115k, above market

With two major launches — ES9 and ONVO L80 — entering sales and delivery in Q2, Nio guided 110k–115k units, above the ~106k consensus. Given only ~29.4k were delivered in Mar, the guide implies 40k–43k average monthly deliveries for May/Jun, a robust cadence.

ES8 lead times shortened from 4–5 weeks in Mar to 2–4 weeks in Apr, signaling fast backlog digestion. Hence, most incremental beats in Q2 should be driven by ES9 and L80 (and refreshed 5566), confirming strong order books and aligned delivery pace.

2) Q2 revenue and ASP both above expectations

Total revenue guided at RMB 32.9bn–34.4bn vs. ~RMB 30bn consensus, with both deliveries and ASP ahead. Implied vehicle ASP is ~RMB 270k in Q2, roughly flat vs. Q1’s RMB 273k.

As ES8 accounted for 54% of the Q1 mix but its backlog and deliveries are now trending lower QoQ, ES9 must backfill to sustain high ASP. If ES9 holds at 5k+ per month, mix benefits should counter BOM inflation and may drive further GPM expansion.

Looking at the overall picture:

III. GPM continued to climb QoQ, reaching a record high

Total Q1 revenue was RMB 25.5bn, up 112% YoY but slightly below the RMB 26.0bn consensus, mainly on ASP.

① Vehicle revenue was RMB 22.8bn vs. RMB 23.1bn expected as ASP of RMB 273k came in below the RMB 277k estimate, though, as noted, ASP rose by RMB 20k QoQ on a higher ES8 mix.② Other revenue was RMB 2.75bn, down ~RMB 0.3bn QoQ, driven by lower tech services and used-car sales.

Overall GPM was 19% in Q1, up 1.5ppt QoQ to a record high, with both vehicle and other segments beating.

① Vehicle GPM was 18.8%, up 8.6ppt YoY as ES8 mix rose and prior cost-down delivered, offsetting weaker scale and raw material inflation.② Other-segment GPM was 20.6%, up 8.8ppt QoQ from 12% even as high-margin tech services declined QoQ, mainly due to: a) efficiency gains in energy services and faster loss reduction at swap stations; b) growth in higher-margin parts, accessories, and after-sales.

IV. Continued R&D cuts; cost-down and efficiency push sustained

Nio historically spent heavily to build a luxury brand, but since Q2 2025 it has undertaken sweeping reforms. Q1 again showed cost-down and efficiency gains beating expectations, supporting profit leverage.

1) SG&A: flat QoQ, down meaningfully YoY

SG&A had been elevated due to the Nio House footprint, service system, staffing, and separate channels for the Nio and ONVO brands. Now, ONVO has merged into the main Nio channel to share resources, and ongoing optimization of sales and service teams has lowered payroll and marketing costs.Q1 SG&A was ~RMB 3.5bn, down ~RMB 0.9bn YoY from the ~RMB 4.4bn peak in Q1 2025 and roughly flat vs. Q4 2025’s RMB 3.54bn.

Despite seasonality, SG&A did not drop QoQ because the company invested to prepare for the Q2 launch cycle — maintaining channel exposure and staffing for ES9 and L80 — rather than relaxing discipline. Management has guided 2026 SG&A ratio to below 10% of revenue, indicating orderly expense control.

2) R&D: further reduced to RMB 1.9bn

Q1 R&D was RMB 1.9bn, down RMB 0.13bn QoQ from RMB 2.03bn, below the RMB 2.2bn consensus and the prior RMB 2.0bn–2.5bn quarterly guide. The decline reflects the passing of prior peaks (e.g., NT3.0 development largely complete), team optimization, and efficiency gains in design and development.

Importantly, R&D efficiency improved even as spend fell, evidenced by two aspects:

① AD stack upgraded with world-model approach, boosting all-scenario performance: Nio rolled out NWM 2.0 early in the year, integrating a ‘world model’ and closed-loop reinforcement learning. Through massive imitation learning, long-horizon digital-world decision rollouts, and high-frequency closed-loop RL, it delivers an end-to-end loop from core tech to user experience, significantly improving highway/city pilot, auto-parking, and active safety; user AD usage hours rose over 80% MoM in Feb, with two major updates planned this year.

② In-house chip business secured external funding, advancing commercialization: the entity for the self-developed NX9031 (Shenji) completed a first-round external raise of RMB 2.257bn via new shares; Nio holds 62.7% of the chip subsidiary. Built on 5nm with 1,000+ TOPS, performance rivals multiple Nvidia Orin-X chips.Beyond strengthening the cash position, this marks a shift from internal use to external supply, opening new monetization pathways.

Meanwhile, the second-gen 5nm AD chip has taped out, with better performance and much lower cost, and a mid-range product line is planned to cover broader customers. This signals a move from single-point R&D to a product matrix and readies offerings for external commercialization; the market expects continued independent financing with a long-term goal of a standalone listing.

V. Losses slightly better than expected

With Q1 deliveries down 33% QoQ to ~83.5k, management had anticipated a small operating loss. Yet better-than-expected GPM and further R&D cuts to RMB 1.9bn (below consensus and guidance) led to narrower losses vs. expectations.

Q1 OP was -RMB 0.3bn vs. -RMB 1.0bn expected; OPM was -1.2%, down 3.5ppt QoQ, but ahead of the -3.8% consensus. Overall, Nio showed strong profitability resilience during the off-season — ES8 mix supported margins, and ongoing cost controls expanded bottom-line flexibility.

<End here>

1) Core ops — GMV grew nearly 9% YoY, a sharp acceleration from 0.6% in 4Q, and ahead of the ~5% market view. The whole sector saw a QoQ recovery from late last year, but VIPS improved slightly more than expected. This is the key beat on the top-line proxy.

That said, active users barely grew (+1%) and orders rose 3%, not a clear beat. In other words, ecosystem activity and order volume were not as strong as GMV suggests. Growth was largely driven by an ~5% uplift in ASP, which benefited from robust CNY sales of higher-priced, higher-margin winter apparel.

2) Despite faster GMV, total revenue rose only 1.2%, in line with conservative expectations. By mix, self-operated retail revenue grew just 0.2%, while ads/commission/fulfillment rose 14%, widening the spread. In our view, a higher 3P mix (incl. outlet business) and more returns likely drove the divergence.

3) Reversal after a reversal: while revenue growth was muted, gross profit tracked GMV and grew strongly. GPM reached 24.4%, up 120bps YoY, beating estimates, and GP was RMB 6.5bn (+~7%), about RMB 330mn above the Street. The drivers were higher ASP and a greater mix of higher-margin 3P.

4) As growth recovered, opex ticked up modestly. Total opex was about RMB 4.16bn, up 3.6% YoY and faster than revenue, slightly diluting margins. Spending discipline was mostly intact despite the volume recovery.

By line, only fulfillment, tied to volumes, rose 8% YoY, lifting total opex, while other items remained restrained, generally at -2% to 0%. Stepping back, VIPS’ beat came from stronger GP on solid ASP and GMV, and although operating leverage softened, OPM still rose 70bps YoY (GPM +120bps, opex rate -40bps drag). Operating profit reached RMB 2.5bn, up ~10% YoY and above the RMB 2.29bn consensus, delivering profit growth without much revenue growth.

Dolphin Research view:

With only 1% revenue growth and low-single-digit user and order growth, VIPS’ results are not absolutely strong. But with expectations subdued, relative performance this quarter was respectable. The delivery was better than feared.

One tailwind was CNY timing, which shifted some demand from 4Q into 1Q. Even so, VIPS showed an ‘important capability’ — even at a ‘small yet refined’ scale with limited growth in active users and orders, it can lift profits by improving GPM through high-repeat, sticky users and mix upgrades toward higher-ticket apparel and luxury. This mix-led model can still drive earnings.

Of course, this assumes sector conditions do not deteriorate materially, and the biz does not shrink. Without scale stability, asking for profit growth is unrealistic. Scale effects matter for margins.

The bigger issue now is the new consensus that e-commerce conditions are worsening rapidly in 2Q, as seen in Apr retail sales and company guidances. The key is whether sector sentiment improves in 2H as state subsidies fade and the tough base rolls off. A stabilization would help VIPS sustain earnings momentum.

On valuation, with revenue growth guided to roughly 0%–5%, and despite limited visibility into 2H, the lower base should make mid-single-digit profit growth achievable. That implies FY26 net profit of about RMB 8bn. This supports a still-attractive earnings yield.

Shares have pulled back ~27% from the peak, implying only ~5.8x FY26 PE, the low end of the sector’s ~6–8x range. The company did not repurchase stock this quarter, reserving cash for the Apr full-year dividend (~US$300mn). Management has reiterated a 75% payout of FY25 adj. profit via dividends and buybacks, roughly RMB 6.5–6.6bn, implying a near-14% total shareholder return on the current market cap for FY26 and offering strong downside support.

We maintain our view: upside may be limited, but for low-risk investors prioritizing capital preservation over outsized gains, VIPS is worth considering when the multiple is ~6x or below. Risk-reward skews better near the lower end of the range. Selective entry makes sense.

Core charts and details below:

I. GMV has bottomed, with some mix blemishes

GMV rose nearly 9% YoY, a sharp improvement from 0.6% in 4Q, and beat Bloomberg’s ~5% view. Consistent with industry prints and peers, VIPS showed a clear recovery in 1Q vs. 4Q. This confirms a bottoming in growth momentum.

There are mix issues. Active buyers were ~41.7mn, up only 0.9% YoY, and orders grew 3%, slightly below Bloomberg consensus. Volume indicators were therefore lukewarm.

Thus, growth relied more on an ~5% YoY increase in ASP. Per management, this reflected strong CNY sales of high-priced, higher-margin winter wear. The mix tailwind was notable.

II. Revenue–GMV gap widened again, but GPM moved higher

Although GMV recovered to nearly 9% YoY, revenue grew only 1.2%, near the low end of guidance and in line with conservative expectations. This underscores the growing gap between flow and recognized revenue. It also reflects returns and mix headwinds.

By mix, self-operated retail grew just 0.2%, while ads/commission/fulfillment rose 14%, a clear scissors spread. Historically, a large gap between revenue and GMV often coincides with higher return rates. Hence, a higher 3P contribution (incl. Shanshan outlets) plus more returns likely drove revenue growth well below GMV.

In a ‘re-reversal’, while revenue lagged, gross profit growth was stronger and tracked GMV. Specifically, GPM was 24.4% (+120bps YoY), ahead of expectations, and GP reached RMB 6.5bn (+~7%), about RMB 330mn above the Street. Margin quality improved with mix and pricing.

As for the GPM uplift, it reflects higher ASP and a modest increase in higher-margin 3P. Both factors supported profitability. The margin trajectory looks healthier than revenue trends suggest.

III. Flat revenue, growing profits

With GMV and biz growth rebounding, opex ticked up slightly. Total opex was ~RMB 4.16bn (+3.6% YoY), above revenue growth, modestly diluting margins. Expense discipline remains broadly intact.

Specifically, fulfillment costs, mechanically tied to volume, rose 8% YoY, lifting total opex, while other expenses stayed restrained. Marketing fell 2%, and G&A and R&D were roughly flat YoY. Cost control outside logistics remains tight.

Thus, the beat was driven by stronger GP on better ASP and GMV, and despite slightly lower operating leverage, OPM rose 70bps YoY (GPM +120bps, opex rate -40bps drag). OP reached RMB 2.5bn (+~10% YoY), above the RMB 2.29bn consensus. Profit growth broadly matched, if not exceeded, underlying GMV growth.

Revenue barely grew, but profit delivery was solid and aligned with GMV momentum, even a touch higher. This underscores mix-driven profitability. The earnings profile improves even with subdued top-line.

<End>

Past Dolphin Research on VIPS:

Feb 26, 2026 note: VIPS: Back to negative growth again? Fear not, 10% shareholder returns to the rescue

Nov 20, 2025 note: VIPS: After finally bottoming and warming, what is next?

Nov 20, 2025 call Trans: VIPS (Trans): Expect 4Q user growth to top 3Q

Aug 15, 2025 note: VIPS: Small yet refined out of the trough, watching giants’ life-or-death battle

Aug 15, 2025 call Trans: VIPS (Trans): Expect revenue growth to be positive in both 3Q and 4Q

May 20, 2025 call Trans: VIPS (Trans): Targeting a 75% payout of 2025 profit to shareholders, hoping to return to positive growth in 2H

May 20, 2025 note: VIPS: Sliding toward the bottom again — will buybacks provide a floor?

Feb 21, 2025 note: VIPS: Despite volatile results, shareholder returns provide support

Feb 21, 2025 call Trans: VIPS (Trans): Striving for profit growth in 2025

1. Core operating metrics: $NVIDIA.US reported total revenue of $81.6bn, beating the raised buyside range ($78-80bn). Quarterly revenue rose by $13.5bn QoQ, driven almost entirely by the data center business as Blackwell ramped.

GPM came in at 74.9%, down 10bps QoQ and broadly in line with the Street (75%). With B300 ramping, margins have recovered to around 75%.

2. Data center: Revenue was $75.2bn, up $12.9bn QoQ. The increase mainly reflects scaled deliveries of B300 series chips, with Blackwell now the dominant architecture across customer types.

The company changed disclosure from ‘Compute and Networking’ to ‘Hyperscale customers and Other cloud customers’. Specifically, hyperscale revenue was $37.9bn (+$4.0bn QoQ), while other cloud revenue was $37.4bn (+$8.9bn QoQ), the largest incremental driver.

With large cloud vendors pushing in-house silicon, Longbridge Dolphin Research believes the customer-based split aims to highlight NVIDIA’s sustained growth in industrial and enterprise cloud. The intent is to underscore momentum outside the mega-cap hyperscalers.

3. Edge computing: All non-data center businesses have been consolidated into Edge, including PCs, consoles, workstations, AI-RAN base stations, robotics and autos. These sub-segments are no longer disclosed separately.

Edge revenue was $6.4bn, up 29% YoY. Gaming is the largest piece within Edge. Growth this quarter was mainly driven by gaming.

4. Profit: Core OP was $53.5bn, up 147% YoY, helped by rapid top-line growth and GPM back near 75% (the prior-year quarter saw a margin air-pocket due to the H20 ban). Core OPM reached 65%.

5. Next-quarter guide: The company guides FY2027 Q2 (2Q26) revenue to $91.0bn, up $9.4bn QoQ, which excludes China data center compute revenue and tops the raised buyside view (~$89bn). GAAP GPM is guided to 74.9% for next quarter, flat QoQ, essentially in line with the Street (74.8%).

Longbridge Dolphin take: The compute king is being hunted on ‘value’

At GTC, Jensen Huang raised the AI outlook again, expecting data center revenue to total $1tn across 2025-2027 (vs. $500bn guided at last year’s GTC). The market is not particularly worried about FY2027-28, so a small beat this quarter is unlikely to meaningfully lift the stock.

Capex at the tech majors is still moving higher, NVIDIA’s AI chip share continues to rise, and the company offered a Vera Rubin plan tailored for agentic workflows. This includes more suitable system-level compute for agentic inference.

a. It developed an in-memory inference context platform (ICMS). The goal is to ease the ‘memory wall’.

b. It introduced Groq 3 LPU to accelerate decode. This targets the output stage.

c. It uses its Vera CPU to take over task scheduling. This aims to improve efficiency.

However, NVIDIA’s monopoly premium is fading. Training and inference should be viewed separately: NVIDIA still has clear advantages in training, but post-training inference repeatedly runs identical compute and is more sensitive to token throughput, latency and cost, making it a prime target for in-house chips to replace GPUs.

As inference shifts from chatbots to agentic reasoning, the raw compute bottleneck is easing. Memory, CPUs and adjacent assets are becoming the more urgent constraints.

NVIDIA’s products and system-level solutions still lead by a wide margin. Yet as AI hits a phase where ‘memory, CPU, interconnect and value’ matter more, and cloud vendors have fallback options for inference, the moat and pricing power that were unique in the training era have weakened. This is a structural shift in where bottlenecks and value accrue.

That said, with industry beta plus Blackwell volume, NVIDIA is squarely in an earnings release phase. Shares should track Blackwell shipments and the Vera Rubin rollout in 2H, which is better tuned for inference.

NVIDIA’s current market cap (~$5.4tn) implies about 18x PE on FY2028 net income. This assumes a two-year revenue CAGR of ~60%, ~74% GPM and a ~17% tax rate, with high growth through FY2026-28 guided by management’s data center outlook.

Longbridge Dolphin Research set a $175-240 target range after the last print, and the recent high near $236 is close to the top end. If Vera Rubin ships ahead of plan, the stock could clear $250, but a ‘double play’ of multiple expansion and earnings like the past two years will be hard; 2026 may rely on EPS alone to support the stock.

Longbridge Dolphin Research’s detailed analysis of NVIDIA’s results follows below. Further detail is provided in the sections that follow.

I. Business overview

With continued data center expansion, it now exceeds 90% of revenue and is the largest contributor. The company reclassified gaming, autos and other lines into a new ‘Edge computing’ bucket and stopped separate disclosure.

By segment, the details are as follows. Key lines are outlined below.

1) Data center: This is the main focus area, with core products including Blackwell compute and InfiniBand networking and core customers across Amazon, Microsoft and Google.

The business is in the Blackwell cycle with B300/GB300 as the workhorses. As Rubin ramps in 2H, the cycle will transition from Blackwell to Rubin.

2) Edge computing: This includes PCs, consoles, workstations, AI-RAN base stations, robotics and autos. Sub-segment performance is no longer broken out.

Within Edge, gaming is the largest sub-segment. Current products are the RTX40 and RTX50 series, with gamers and PC OEMs as primary customers.

II. Core metrics: A modest beat

2.1 Revenue: FY2027 Q1 (1Q26) revenue was $81.6bn, +85% YoY and above the raised buyside range ($78-80bn). The $13.5bn QoQ step-up came almost entirely from data center as Blackwell scaled.

For next quarter, guidance is $91.0bn, up $9.4bn QoQ and above the raised buyside view ($89bn). Growth next quarter should still be led by B300/GB300. Rubin begins mass production in 3Q26.

2.2 GAAP GPM: FY2027 Q1 (1Q26) GAAP GPM was 74.9%, broadly in line with the Street (75%). The sharp dip a year ago was linked to the H20 restrictions.

Management guides next quarter GAAP GPM to 74.9%, flat QoQ and in line with expectations (74.8%). With Blackwell ramping, margins have moved back toward ~75%.

Management previously indicated a 75% target GPM for FY2027. This provides some confidence, though investors still worry about post-FY2027 downside risk to margins. That risk remains a debate into FY2028 and beyond.

III. Core business trends: Non-megacap cloud drove the bulk of the increment

On the back of AI capex, data center (compute + networking) now exceeds 90% of revenue. Other businesses have been squeezed to under 10% and are now grouped under Edge without separate disclosure.

3.1 Data center: FY2027 Q1 (1Q26) data center revenue was $75.2bn, +92% YoY. It remains the key focus, with growth this quarter primarily from increased Blackwell production, driven by accelerated computing and AI.

By line item: ① Compute revenue was around $60bn, up $8.7bn QoQ, with B300 shipments contributing most of the increase. ② Networking revenue was around $15.1bn, up $4.2bn QoQ.

Cloud players remain the largest buyers of AI chips, so their capex underpins NVIDIA’s data center growth. Based on commentary from Google, Meta, Microsoft and Amazon, Longbridge Dolphin Research expects combined 2026 capex to exceed $700bn, up ~80% YoY, providing a strong backdrop for FY2027 growth.

Beyond downstream capex, competition from in-house ASICs is top of mind. The AI chip market is concentrated among NVIDIA, ‘Google + Broadcom’, and AMD, with these three taking over 90% share. Since model development splits into training and inference, NVIDIA’s GPUs still dominate training, but on lower-spec inference, Google TPUs and other in-house parts offer better value.

Upstream, NVIDIA’s high margins extract a large share of industry profit, pressuring downstream economics. As workloads move into inference, dependence on NVIDIA falls sharply, increasing incentives to build in-house or switch to ASIC alternatives.

Against this backdrop of in-house silicon across the majors, NVIDIA’s disclosure change highlights growth in industrial and enterprise cloud. This reframes the mix by customer type.

By customer: hyperscale revenue was $37.9bn, up $4.0bn QoQ, while other cloud was $37.4bn, up $8.9bn QoQ. The latter was the largest source of incremental revenue.

Over the past two quarters, despite in-house efforts at the megacaps, growth in enterprise/industrial cloud outside the big four has clearly accelerated. Management appears keen to demonstrate sustained growth capability.

With GPM back near 75%, any future share loss could present a ‘potential headwind’ to margins. Management has set a 75% GPM target for FY2027, but mainstream estimates still model margin compression into FY2028 and later. This remains a key watch point.

3.2 Edge computing: FY2027 Q1 (1Q26) Edge revenue was $6.4bn, +29% YoY, mainly on RTX50 and other gaming shipments.

With data center now over 90% of mix, other lines are less material, so they have been grouped under ‘Edge’. Separate disclosure has ceased.

Within Edge, gaming is the largest sub-segment at roughly 60-70%. Despite a soft PC and console market, NVIDIA retains a clear lead in gaming GPUs versus AMD’s ~$0.72bn quarterly gaming revenue.

IV. Key financials: Rapid growth supports a steady climb in margins

4.1 Core OPM

FY2027 Q1 (1Q26) core OPM was 65.6%, continuing to trend up, driven by lower opex ratio.

The build-up is as follows. We break down margin drivers below.

‘Core OPM = GPM - R&D ratio - S&A ratio’, by definition.

1) GPM: 74.9% this quarter, down 10bps QoQ and hovering near 75%. It remained close to the target level.

2) R&D ratio: 7.7% this quarter. While absolute R&D spend rose by ~$0.8bn QoQ, the ratio continued to fall on rapid revenue growth.

3) S&A ratio: Through the Blackwell cycle, sales and G&A stayed relatively stable. The ratio fell to 1.6% of revenue this quarter.

Management guides next-quarter opex to $8.5bn. Based on the revenue guide, the opex ratio should hold around ~9.3% next quarter.

4.2 Core OP

FY2027 Q1 (1Q26) net income was $58.3bn, up 71% YoY.

Given net income includes non-operating items, we focus on core OP (GP - R&D - S&A). Core OP was $53.5bn, up 147% YoY. Core OPM rose to 65.6%, mainly on revenue growth and margin recovery (vs. H20 restrictions a year ago).

Current growth is primarily driven by the Blackwell cycle. With management’s data center outlook and Rubin mass production starting in 3Q, high growth should extend through FY2027-28.

<Ends here>

Back catalog of Longbridge Dolphin Research on NVIDIA:

Hot topics

Mar 18, 2026 post-GTC analyst huddle: NVIDIA (Analyst Huddle): IT to become a token distributor; 50% of cash flow to shareholders.

Mar 17, 2026 GTC minutes: NVIDIA (GTC Minutes): LPU reshapes AI inference; compute factories head to space.

Mar 17, 2026 GTC conference: NVIDIA GTC: AI’s gala night — hype in, mixed feelings out?.

In detail:

1. Multiple factors slowed Billings: Q1 Billings grew 10% YoY, within guidance but decelerating vs. last quarter. Macro headwinds were one factor; a higher base, a later and longer CNY holiday, and the ensuing lag in hiring demand also weighed. Sensor Tower shows Billings rebounded quickly after workers returned from CNY.

2. Revenue guidance in line: Q2 revenue guidance implies 13%–15% YoY growth, an uptick QoQ and in line with expectations, reflecting lagged confirmation of post-CNY demand. Combining Q1 and Q2, 1H revenue growth is about 11%, roughly flat vs. last year’s 1H. Given the low base last year, momentum is best described as steady.

3. B-side client count kept rising: Billings growth was driven by steady net adds in paying clients. Paying enterprise accounts rose by 3 mn QoQ to 7.1 mn, a new high. Implied ARPPU was roughly flat YoY; Sensor Tower’s iOS Billings grew 14% YoY in Q1, outpacing total Billings, and Q2 to date is up 29%, indicating further acceleration. This points to demand led mainly by SMEs.

4. C-side MAU growth slowed: Q1 MAU reached 60.9 mn, a net add of 2.9 mn QoQ, softer than prior years and a clear deceleration. Near term, timing of CNY likely played a role. Longer term, white-collar penetration may be nearing a local peak, while blue-collar user acquisition must compete with traffic-heavy, tech-forward general platforms like Douyin/Kuashou and offline hiring channels.

QM data show Kanzhun remains dominant within verticals, with MAU far ahead and user time-share above 60% and rising.

5. Operating leverage continues to release: Q1 core OP was RMB 620 mn, a modest beat, with a 30% OPM, down 400bps QoQ and up 750bps YoY. The upside was mainly from tighter G&A. Self-help remains the fastest way to sustain earnings. R&D is small in absolute terms, and trimming S&M and G&A has been more impactful.

This year includes a co-sponsor fee for the World Cup (Approx. RMB 200 mn), and S&M had already trended lower last year. In the near term, G&A offers the most room to tighten.

6. Shareholder returns average: Last quarter, management raised the buyback authorization, lifting the 2-year $250 mn program (announced Sept 2025) to $400 mn, and pledged from 2026 onward to return 50% of Adj. net income via buybacks and dividends. Under market and top-tier house assumptions, 2026 Adj. net income is Estimated at RMB 3.8 bn, implying nearly RMB 2.0 bn (~$280 mn) for returns. Against a current RMB 6.8 bn market cap, that is about a 4% yield, decent but not standout.

7. Core metrics vs. consensus

Longbridge Dolphin Research View

The setup is complex: consumption remains fragile, and white-collar tech workers face AI risk, worsening labor market imbalances. While blue-collar online penetration is rising and Kanzhun continues to take share, SME hiring demand is volatile with uneven monetization. That shows up as strong online paid Billings but potentially smaller annual framework contracts at large enterprises.

Hence, more top-line optionality will come from geographic expansion. In Hong Kong, penetration is progressing well, with the OfferToday platform now No.1 by DAU and No.2 by MAU among mainstream job portals. At the current trajectory, it is close to overtaking JobsDB.

Hong Kong is small, though, so until expansion scales, self-help remains the fastest way to sustain earnings. This has been Kanzhun’s main response to early growth deceleration over the past year. Despite World Cup sponsorship, more AI and new product spend this year, management still guides Adj. OPM higher in 2026, albeit at a slower pace (vs. last year’s +900bps).

Assuming 2026 revenue growth of 11% and Adj. OPM rising 100bps from 40.8%, profit would reach RMB 3.8 bn, up 13%. At a 15% tax rate and a current RMB 6.8 bn market cap, 2026 P/E is about 14x, roughly in line with this year’s growth and with most of the vertical-leader premium gone.

Valuation has compressed, but not absolutely cheap. On one hand, sentiment is weak and internet platform multiples have drifted below 15x. On the other, AI and macro headwinds imply slower growth ahead, and management already guided a 10% 2-year CAGR last quarter, limiting near-term upside to expectations.

In short, Kanzhun’s management is credible and execution is solid, and it has emphasized shareholder returns early. But with growth slowing and a 2-year lower-growth path signaled, a clear inflection for a trend reversal is hard to foresee near term. Unless it further steps up capital returns or monetizes expansion to lift the market view, a strong share-price recovery still looks challenging given current sentiment.

Details below

1. Users: expansion slowed, vertical share kept rising

Kanzhun added a net 2.9 mn MAU in Q1 to 60.9 mn, with growth decelerating. Near term, CNY timing likely matters, but longer-term shifts suggest white-collar penetration is nearing a phase peak, while blue-collar acquisition must compete with traffic-rich general platforms like Douyin/Kuashou and offline channels. QM data show Kanzhun’s vertical scale advantage remains overwhelming, with MAU far ahead and total time-share above 60% and rising.

2. Enterprise-side Billings kept recovering

Q1 revenue was RMB 2.07 bn (+7.6% YoY). Within that, B2B online recruitment services revenue was RMB 2.06 bn (+8.2% YoY). Other revenue was RMB 11.01 mn (-50% YoY), as value-added services for job seekers were optimized by offering more for free to sustain ecosystem stickiness. On this view, enterprise recruitment revenue slightly beat expectations.

Management guided Q2 revenue of RMB 2.38–2.42 bn, implying 13%–15% YoY growth, consistent with lagged demand confirmation from the shifted CNY.

(1) Company view: SMEs drove growth

Paying enterprise accounts reached 7.1 mn, up 0.3 mn QoQ, while implied ARPPU was flat YoY, indicating stronger lift from SMEs. Calculated Q1 Billings growth was 9.7%, in line with guidance. Sensor Tower shows iOS Billings up 14% YoY in Q1, and Q2-to-date up 29%, with volatility mainly from the later and longer CNY holiday.

(2) Industry view: headline pressure, mixed by segment

Across labor market indicators, early-2026 employment looked softer than a year ago, with higher unemployment, lower YoY job additions, and worse household unemployment expectations. But supply-demand mismatches vary by sector. Emerging areas like AI and blue-collar manufacturing saw some pickup, with enterprise job postings (Citic-scraped data, excluding Kanzhun) showing modest YoY growth.

3. Continued cost tightening

Q1 GPM was steady QoQ at 85.6%, up 200bps YoY. Core OP (revenue minus COGS, S&M, R&D, G&A) came in at RMB 620 mn, slightly above expectations, with a 30% OPM vs. 22.5% last year, up 750bps YoY. G&A was the main area of optimization. SBC fell 24% YoY to 9.2% of revenue, now near mature-company levels. Ex-SBC, Non-GAAP OP was RMB 800 mn, for a 39.4% OPM, down 400bps QoQ on seasonality.

<End here>

Longbridge Dolphin Research on ‘Kanzhun’

Earnings season (latest quick takes only)

Nov 19, 2025 call notes: Kanzhun (Trans): Renewal improving, growth rebounding from the bottom

Nov 19, 2025 earnings take: Kanzhun: Fewer surprises, fundamentals intact

Aug 21, 2025 call notes: Kanzhun (Trans): Hiring demand on the platform has recovered

Aug 21, 2025 earnings take: Kanzhun: Environment remains choppy, but the business stays disciplined

1) Revenue: more units, lower price, a slight beat. In Q1, HSAI delivered revenue of RMB 680 mn (+29.6% YoY), near the upper end of its RMB 650–700 mn guidance. Growth was entirely driven by stronger-than-expected shipments.

2) Shipments beat; seasonally weak quarter proved resilient, with ADAS and robotics as twin engines. Despite seasonal weakness and the step-down of state subsidies on NEV purchases in China (domestic NEV sales -4% YoY), total shipments reached 477k units, above the 400–450k guide. This again underscores rapid lidar penetration into mass-market models.

ADAS shipments were 353k (+142% YoY). Growth was led by the RMB 1k-class ATX ramping in RMB 100k–200k vehicles (e.g., BYD, Geely), with low-cost blind-spot FTX also starting to scale.

Robotics shipments were 124k (+138% YoY), topping the 100k guide. The JT series for lawn-mower and similar use cases did most of the heavy lifting.

3) ASP still deflating: mix downshift and tougher competition are the key drivers. Both product structure and a more crowded field continue to weigh on pricing.

Q1 blended ASP was ~RMB 1,441, down 45% YoY and a further 7% QoQ. Even with higher-priced robotics share rising 13ppt QoQ to 25%, overall ASP was dragged by mix downshift and competition, reflecting a deliberate price-for-volume strategy.

a. Low-priced ATX (custom SKUs for BYD and Geely) took a larger share. These custom SKUs are estimated at ~RMB 800 per unit, well below the US$150 price point for the standard ATX in 2026.

b. Low-cost FTX blind-spot lidar (ASP ~US$100) began to ship. This further pushed blended ASP lower.

4) GPM under pressure, slipping below the 40% threshold. Q1 GPM came in at 39.1%, down 260bps YoY, below the 39.9% street view and below the company’s usual 40%+ level.

Dolphin Research believes the decline mainly reflects the following. Two factors stand out.

a. ASP deflation (−45% YoY) outpaced cost improvements (e.g., in-house chips), and seasonal scale benefits did not kick in. This weighed on gross profit.

b. Higher-margin service revenue share fell 2.8ppt YoY to just 0.1%. The mix shift further dragged the overall GPM.

5) Opex tightly managed, but core OP remained slightly in the red. Total opex was RMB 300 mn; against ~30% revenue growth, opex rose only 4% YoY. This reflects internal efficiency gains and a more disciplined operating approach.

Core operating profit was RMB -30 mn, a small loss. The shortfall was mainly due to ASP deflation lowering gross profit and lack of seasonal scale. Even so, results were slightly better than expected.

Dolphin Research key takeaways: Overall, the quarter landed broadly in line: revenue slightly above the midpoint of guidance, and GAAP net profit modestly above breakeven guidance (helped by non-recurring gains). The core growth driver remained a shipment beat.

However, the print also highlights the key tension: aggressive price deflation pulled GPM further down to 39.7%, breaking the 40% line. Beyond tech-driven cost downs, intensified competition is pushing the company to trade price for volume to defend share against products like RoboSense EMX.

That sets up 2026 as a full-on year of price-for-volume. With the higher-priced ETX (L3 core lidar) not ramping until year-end, competition will center on lower-priced L2 units (ATX/FTX), while humanoid robotics remains an out-of-the-money option with limited near-term revenue.

Looking into the mid-to-long term, while near-term competition and GPM pressure are real, HSAI remains a high-certainty beneficiary as the leader under the dual pillars of rising ADAS penetration and a second growth curve in robotics. The long-term setup stays intact.

But for 2026 specifically, this is an unavoidable pain point of navigating a price war and positioning for the next-gen sensor ecosystem. The year will be about holding share and building scale.

① Full-year shipments to remain strong, but ASP set to contract on intensified competition. Stronger competition will be the key headwind for pricing.

a. Shipments to grow rapidly: HSAI raised its 2026 shipment guide from 2–3 mn to 3–3.5 mn units, implying +85% to +116% YoY. To meet demand, annual capacity is planned to more than double from 2.0 mn units in 2025 to 4.0+ mn in 2026, with both ADAS and robotics expected to double.

ADAS: ADAS shipments are guided at 2.5–3.0 mn units (with ~200–300k FTX blind-spot units, and the rest RMB 1k-class ATX). Growth drivers include two vectors.

① Further ATX penetration into L2+ mass-market segments (lidar has reached the RMB 80k–100k vehicle band). ② L3 adoption will multiply per-vehicle lidar counts, though the ramp may slip to late 2026 or even 2027, taking system value from ~US$200 at L2 to US$500–1,000, with higher customer tolerance for pricing.

By customer, 2026 incremental volume is expected to be anchored by Xiaomi (~550k, sole supplier), Leapmotor (~600k, near sole supplier), Li Auto (~450–500k, sole supplier), BYD (~300–350k, roughly half of orders), Geely (~500k), and Great Wall (~200k). These accounts represent the bulk of ADAS growth.

Robotics: Shipment guidance is ~500k units (+109% YoY), still led by the JT series at ~RMB 1,000 ASP (~450k units) to brands such as Dreame and Mova for lawn-mower robots. The share of higher-value robotics lidar remains low.

Given the Q1 shipment beat, Dolphin Research expects full-year shipments to reach near the top end at ~3.4 mn units. The bias is to the high side of the guide.

b. But ASP will keep compressing: As competitors (e.g., RoboSense EMX) narrow the product gap and began large-scale shipments in Q4 2025, price competition has intensified. As the leader, HSAI is giving up its historical 10–20% premium to defend share via price-for-volume.

Specifically: The following dynamics will pressure blended pricing.

a. While the standard ATX is expected to stay around US$150 in 2026, custom SKUs for high-volume customers like BYD and Geely are ~RMB 800. As their mix rises, the ATX line average will fall.

b. FTX blind-spot radar ASP is guided at ~US$100. This caps blended pricing.

c. The higher-value ETX will not mass-produce until late 2026. Mix uplift from ETX will be limited before then.

As a result, blended lidar ASP will keep deflating in 2026. Based on guidance (revenue RMB 4.2–4.6 bn; shipments 3.0–3.5 mn), implied blended ASP is RMB 1,300–1,380, down 25–29% vs. the 2025 ASP of RMB 1,844, a steeper decline than Dolphin Research expected.

Accordingly, Dolphin Research now expects 2026 blended ASP to fall 28% YoY to RMB 1,320, especially after a weaker-than-expected Q1 price level. We expect 2026 revenue of RMB 4.58 bn (+51% YoY), near the top end of the company’s range.

c. Despite ASP pressure, HSAI previously expected FY26 GPM to stay resilient on cost-downs and scale benefits, mainly from: Technology integration and manufacturing scale are the primary levers.

a. Localization and high integration of the main SoC (in-house FMC500 based on RISC-V, integrating MCU, FPGA, and ADC), moving to a single-chip solution that slashes core chip costs, historically ~40% of BOM. This materially lowers unit costs.

b. In-house SPAD integration (mass production in 2026). Internalizing key components should support margin.

c. Scale benefits from 3.0–3.5 mn units and highly automated manufacturing. Fixed-cost absorption should improve as volume ramps.

That said, the sharp Q1 ASP decline (blended ASP −45% YoY) clearly dented GPM, which dipped below 40% to 39.1%. Considering full-year mix (higher share of low-price ATX/FTX with ETX only ramping at year-end), Dolphin Research conservatively expects FY26 GPM to fall 180bps YoY to ~40%.

d. On expenses: Total opex is expected to rise ~15% YoY, with roughly RMB 200 mn of forward-looking investment into the new robotics business (the ‘physical AI eye’ for integrated sensing and the ‘physical AI muscle’ for micro-motors). This spend is separate from the external Sharpa vehicle and is HSAI’s investment in its second growth curve.

As the new business contributes limited near-term revenue, the incremental RMB 200 mn will constrain FY26 profit delivery in the short term. Excluding the new-business spend, core opex is expected to be flat to down low-single digits, reflecting solid cost control.

At the current share price, Dolphin Research models 2026 revenue of RMB 4.58 bn (+51% YoY) and net profit of RMB 600 mn (midpoint of the RMB 500–700 mn guide), +38% YoY. On that growth, we assign 35–40x PE for 2026, implying fair value of RMB 21–24 bn, broadly in line with the current market cap.

Near term: with competition centered on low-cost L2+ lidars, rival tech routes catching up, and share pressure ongoing, upside looks limited. ETX and high-margin overseas orders will not contribute meaningfully until late 2026 or 2027, and robotics remains an unpriced option for now.

Longer term: lidar exhibits a Moore-like curve (exponential point-cloud density, exponential cost declines) and high moats from ecosystem lock-in and scale. As the leader under the dual logics of rising ADAS penetration and a second-curve robotics expansion, HSAI remains a high-certainty long-term beneficiary.

① Share may be pressured near term, but tech coverage is broad; share should stabilize mid-to-long term. Despite share volatility since H2 last year due to Huawei’s blind-spot lidar ramp, Huawei primarily ships as an integrated HW+SW stack into its own ecosystems (Aito and HI models) at higher prices.

Thus, among independent Tier-1 suppliers, the real contest remains between HSAI, RoboSense, and Tudatong. HSAI still holds the No.1 spot on strong low-cost ATX shipments, but RoboSense EMX has been ramping since Q4 last year, implying continued share pressure in 2026.

② L3 will multiply per-vehicle lidar counts. As noted, L3 demand upgrades from a single ATX to ‘1 main unit (ATX/ETX) + multiple FTX blind-spot units,’ lifting per-vehicle value from ~US$200 to US$500–1,000. HSAI’s ETX is slated for mass production from late 2026 into 2027, positioning it to benefit directly.

③ Overseas poised for a breakthrough, with NVIDIA ecosystem leverage. HSAI has completed C-sample development of a long-range lidar for a leading European OEM and expects overseas mass production by end-2026. It has also been selected for NVIDIA Drive Hyperion, giving it preferred partner status when OEMs adopt NVIDIA solutions and materially improving product integration and scalability.

Given overseas customers’ lower price sensitivity and preference for higher-end products, international expansion will help offset domestic ASP pressure. Profit elasticity should start to release by late 2026 or 2027.

④ Robotics as the second growth curve. HSAI is pursuing a dual-track approach.

In-house: a general-purpose sensor path with the ‘physical AI eye’ (fully integrated sensor) and ‘physical AI muscle’ (micro-motors), targeting a 50/50 revenue split with lidar in five years. External (Sharpa): via founder-backed Sharpa for dexterous hands, where HSAI can supply micro-motors or act as a contract manufacturer, with higher per-unit value.

Commercialization here will be gradual and remains in the investment phase. This is a long-dated, high-value option that could unlock significant valuation upside once humanoid robots scale.

From a long-term TAM perspective, the current ~RMB 25.6 bn market cap largely reflects the domestic passenger-car ADAS base (a conservative RMB 40 bn market by 2030). With overseas breakthroughs, the most comprehensive next-gen lidar roadmap, and a sizable robotics option, medium-to-long-term upside remains solid.

See Hesai: A Tesla castoff? Lidar’s stylish comeback for detailed valuation work. More details are available via the link.

Past articles: See below. Additional references follow.

Hesai deep dives:

‘4x’ Hesai: Why a Tesla castoff found new light. See link for the full analysis.

Hesai: A Tesla castoff? Lidar’s stylish comeback. Full details in the article.

Hesai earnings.

Hesai: Shipments surging for the ‘eye of intelligent driving’—can it soar again?. Click through for the report.

In detail:

1. Ads remain in a clear upswing: Q1 ad revenue rose 30% YoY, a small beat even against a weak industry backdrop. Benefiting from strength in AI and gaming verticals and robust competition, Bilibili captured notable tailwinds.

This high-growth phase should persist for a while as sector momentum and competition stay elevated. At the same time, the platform still has room to raise ad load.

2. User growth slowed but engagement improved: MAUs rose by 10 mn QoQ to 376 mn, basically back to last year’s Q3 level rather than setting a new seasonal high, which is a blemish. The timing of the Lunar New Year and winter break may have played a role, to be clarified on the call.

Engagement kept edging higher YoY, with DAU/MAU at 30.6% and avg. daily time hitting a record 119 minutes. QM data suggest total time spent growth in Jan. rebounded slightly vs. Q4, albeit off a not-high base last year, pointing to steady momentum overall.

Compared with social and long-form video peers, Bilibili looks decent aside from Douyin’s relentless growth and a mild rebound at WeChat. Xiaohongshu has notably slowed, and Kuaishou has deteriorated slightly.

2. New game cycle drawing near: Q1 was still under tough comps, with game revenue down 12% YoY. The quarter mainly saw the Hong Kong/Macau/Taiwan rollout of‘Three Kingdoms: Strategist’, a smaller market with stable performance.

Deferred revenue was up 19% YoY at Q1-end but dipped slightly QoQ, likely due to last quarter’s boost from ‘Escape from Karkov’. As ‘Three Kingdoms: Hundred Generals’ targets a late-Q2 launch and two more titles are slated for 2H, the new game cycle is approaching.

3. Live streaming under pressure; long-form video also weakened: VAS growth slowed further to +3.7% YoY in Q1, with both live-streaming and premium memberships softer, and members down by nearly 5 mn.

QM data show total time on iQIYI/Youku/Tencent Video continued to fall in Q1. Beyond long-form drama losing share to AI comic-dramas, short live-action series and short video, the Spring Festival box office was also bleak.

That said, Bilibili’s premium content logic differs from iQIYI/Youku/Tencent Video. Supply is primarily sustained by creators, while licensed films/TV focus on global classics where user comments and bullet chats help surface preferences, enabling lower-cost acquisitions of older hits.

4. Heavier AI spend starts to show: OP came in at RMB 170 mn with a 2.2% margin. Q1 typically sees lower margins QoQ due to ad seasonality from e-commerce, higher marketing, and annual bonuses, but the QoQ decline was steeper than in prior years.

The key difference is stepped-up AI investment flagged last quarter, reflected in R&D reversing three years of tightening (mainly from game R&D cuts) to +9% YoY this quarter. Adj. net profit was RMB 590 mn (adding back SBC at 4% of revenue), a 7.8% margin, down nearly 3ppt QoQ.

Management previously guided that incremental AI spend this year, partially offset by cost controls elsewhere, would still reduce profit by RMB 500 mn–1 bn. Q1 trends appear consistent with that framework.

5. Buyback finished; watching for a new plan: In Q1 the company repurchased 2.5 mn shares for $60 mn, at an avg. price of $24. The two-year $200 mn authorization approved in 2024 is now completed.

Net cash stood at RMB 19.3 bn (~$2.8 bn), providing capacity for further buybacks. Investors will watch the call for plans on shareholder returns.

Tencent’s recent comments about selling high-valuation investment assets to fund its own buybacks spooked the market, hitting Bilibili given Tencent’s ~10% stake. The stock fell notably.

6. KPI snapshot

Longbridge Dolphin Research view

Ads were again the standout in Q1, but profit is feeling the bite from AI spend before the new game cycle arrives. Management guided last quarter that this year’s incremental AI investment would cut profit by RMB 500 mn–1 bn. Based on Q1 trajectory, we estimate a roughly RMB 50–100 mn impact, broadly tracking guidance.

While management also said they would invest with discipline and calibrate as they go, the market still de-risked first, which is understandable. At an equity value near $11 bn at the time, the stock traded at ~25x on previously expected adjusted profit (adding back SBC), well above many China internet peers.

Investors had been pricing to a 15%–20% long-term OP margin guide over the next three years. New AI spend disrupts that path since it is not a one-off, and its revenue lift is not as visible as cloud or model subscriptions.

That raises questions about ROI, and near-term profit drag inevitably extends the timeline to target margins. Risk premia have to be adjusted.

From today’s vantage point, we see some changes to how AI spend pressures profit, and further near-term pullbacks could present opportunities:

(1) New game cycle approaching: Top-line acceleration is the best antidote to concerns about investment pressure on earnings. Bilibili currently relies on games to drive growth.

The timeline is drawing closer for ‘三国：百将牌’ around mid-year, ‘闪耀吧噜咪’ in Q3, and ‘三国志：王道天下’ possibly by year-end. While these may lack the breakout potential of ‘三谋’, the lower 2H base and incremental launches should ease AI spend pressure on profitability.

(2) Some valuation premium partly digested: On Tencent’s earnings call last week, management said higher capex and shareholder returns would be balanced by portfolio adjustments, i.e., monetizing certain investments to fund buybacks. They noted they would compare asset valuations with Tencent’s own (ex-investments, core PE ~10x) and sell where valuations are significantly higher.

Given Bilibili’s premium valuation among China internet names, the stock dropped 15% over Thu–Fri on the concern it might be a candidate. Tencent’s holdings fell from 13.5% to 10.5% during 2020–2022 and have been broadly stable since 2023.

Longbridge Dolphin Research believes Bilibili is unlikely to be a key divestment target for Tencent:

First, Bilibili’s market cap has mostly been in the $8–10 bn range over the past year. A neutral near-term valuation is just above $10 bn, with upside to $15 bn in a risk-on tape (cf. the valuation ranges we outlined last quarter).

Even selling the full ~10% stake would raise less than HKD 10 bn, which covers less than one month of Tencent’s current buyback pace of ~HKD 500 mn/day.

Second, as a key pan-entertainment channel and now a primary channel for AI investment, strategic considerations argue against large-scale selling. From a partnership perspective, heavy reduction would be suboptimal.

We thus think near- to mid-term concerns are somewhat overdone. At the latest close, market cap was $8.1 bn.

Assuming RMB 800 mn profit impact from AI, Bilibili trades at ~21x 2026e adjusted PE (est. adj. net profit RMB 2.6 bn). While not cheap vs. peers, 2026 is a tough year for games, and a new game cycle in 2H with a growth rebound should help further digest the multiple.

Detailed analysis below

I. User growth slowed; engagement strengthened

Bilibili added 10 mn MAUs QoQ on seasonality in Q1 but did not set a new peak. Engagement kept improving, with DAU/MAU at 30.6% and avg. daily time at 119 minutes, up 11 minutes YoY, a new record.

QuestMobile data show Bilibili’s time-spent growth ranked mid-to-high among social peers in Q1. The trend remains constructive.

II. Ads are the standout again

Q1 ad revenue was RMB 2.6 bn, +30% YoY, slightly ahead of the acceleration trend. Growth was driven by rising total user time on platform (+19% YoY, a touch faster than Q4’s +18.6%) and higher ad load.

Ad load is still only ~7%–8%, leaving headroom vs. peers. In early Apr., the company added a ‘pause ad’ slot on the video playback page that auto-plays when users pause, though both users and creators can opt out.

III. New game cycle approaching

Game revenue remained under high comps and a thin pipeline in Q1, down 12% YoY, in line with expectations. The quarter was mainly the HMT launch of ‘三谋’, a limited market, with steady performance.

Deferred revenue declined QoQ against the trend, likely reflecting softer VAS and a lackluster games slate. The pressure should ease as new titles come online.

On the forward pipeline: Beyond titles mentioned last quarter — ‘三国：百将牌’ expected mid-year and ‘闪耀吧！噜哩’ expected in 2H — a co-published title ‘三国志：王道天下’ has been added, likely for Q4.

‘三国志：王道天下’ was first revealed on Mar 18 with all-platform reservations, and began the ‘桃园首测’ on Mar 28. Positioned as a 3D sandbox SLG under the classic ‘三国志’ IP that pioneered SLG gameplay, it is worth attention.

Beyond core strategy, the game emphasizes immersive role-play and development, skewing more single-player. It also follows a ‘less grind, less pay-to-win’ approach, simplifying interactions and focusing on core strategic depth.

Some institutions forecast first-12-month revenue of RMB 1.7 bn, about 40% of ‘三谋’. We think that may be a bit aggressive and cannibalization with ‘三谋’ should be considered, but as new titles approach, the inflection for Bilibili’s games is getting closer.

IV. VAS under heavy pressure

VAS revenue, driven by live streaming and premium memberships, grew 3.7% YoY in Q1, with growth slowing further. Premium members fell by 5 mn QoQ to 24.77 mn.

Given structural headwinds in long-form video, Bilibili is not immune. Outside of premium, fan economy and live streaming likely saw the bigger drag from live.

V. AI spend starts to surface

While profit grew strongly YoY, the impact of AI investment began to show in Q1. On core operations (GP minus opex), OP was RMB 170 mn with a 2.2% margin.

Adj. net profit was RMB 590 mn (primarily adding back SBC at 4% of revenue), a margin of 8%. The QoQ margin drop was notable.

1) GPM continued a slow grind higher

GPM expansion was driven by continued high growth in high-margin ads. By cost line, revenue-sharing remained the largest at 38% of revenue, tied to games, live streaming, and Huahuo ads.

Revenue-sharing costs rose to RMB 2.85 bn in Q1, +7% YoY. The mix shift supports margin over time.

2) R&D posted its first growth in three years

On opex, selling was flat and G&A ticked up slightly, so the increase was concentrated in R&D, which rose 9% YoY — the first clear positive growth since 2023, driven by AI investment.

While profits were somewhat diluted, monetization efficiency continued to improve in our long-tracked ‘traffic monetization vs. cost’ relationship, reflecting a stable ecosystem and endogenous monetization capacity.

Whether AI spend is justified will require management to deliver faster growth and better monetization efficiency. Execution will be key.

<End here>

Longbridge Dolphin Research ‘Bilibili’ archive:

Earnings season (recent)

Mar 5, 2026 call notes ‘Bilibili (Trans): Reinvesting part of incremental profit into AI this year’

Mar 5, 2026 earnings take ‘Bilibili: Ads resilient; a tough ‘one-legged’ stretch for the platform’

Nov 15, 2025 call notes ‘Bilibili (Trans): Content paywalls as a key growth engine’

Nov 15, 2025 earnings take ‘Bilibili: A second growth spurt?’

Aug 21, 2025 call notes ‘Bilibili (Trans): Reaffirming 15%–20% long-term OP margin target’

1) Existing-home brokerage is basing: aided by top-tier cities loosening purchase caps and lifting housing fund loan limits in Feb–Mar, the core existing-home business showed signs of bottoming and stabilizing this quarter. GTV was still down about 8% YoY, but rose roughly 11% QoQ. Within that, franchise channels outgrew the Lianjia self-operated channel. As the company improved platform monetization and a resale recovery reduced the need for commission discounts to spur deals, the overall take rate for existing homes rose by 1.9bps QoQ, pointing to improving margins.

2) New-home pressure remains heavy: the new-home GTV decline improved slightly vs. last quarter but was still about -37% YoY, broadly in line with weak sales of higher-ticket new builds even after city-level easing. The new-home blended take rate dipped by ~2bps QoQ from last quarter’s high, a small fluctuation rather than a trend, but revenue still fell about 37% YoY. Bloomberg consensus was looking for a 42% revenue decline, so new homes beat on expectations.

3) Track 2 is also in adjustment: growth looked soft across Track 2, with combined revenue down 8% YoY. The key drag was home improvement, where revenue fell 21% YoY, missing expectations, as weak new-home sales damped demand and the company proactively shut lower-quality businesses and pulled back in some cities. This effectively marks a temporary setback for the ambition to rebuild BEKE via home improvement. Rental revenue dipped 1% YoY due to shifting ‘Hassle-free Rent’ to a net revenue basis. With units under management up ~47% YoY, underlying growth remained strong after adjusting for the accounting change.

4) The bright spot is profit: total revenue still fell about 19% YoY, roughly in line with expectations. But the story this quarter was not growth, it was solid profit delivery. Adj. net profit was about RMB 1.6bn (+~16% YoY), beating the market’s ~RMB 1.15bn view. Once again, the company demonstrated an ability to release earnings in a downturn.

5) Margins improved across segments: despite contracting scale and weaker operating leverage, streamlining stores and headcount lifted productivity, and tight opex control supported margin gains across the board. In Track 1, existing homes’ contribution margin rose 0.9ppt QoQ and 3ppt+ YoY, while new homes still saw a ~2.3ppt YoY margin uplift despite steep GTV declines. In Track 2, home improvement’s contribution margin jumped ~5.6ppt YoY on shedding low-quality business and optimizing the upstream supply chain. Rental’s contribution margin rose ~4.4ppt QoQ to 14.8% as net-revenue ‘Hassle-free Rent’ grew rapidly, making it the biggest upside surprise.

6) GPM expansion and opex discipline drove profits: total GPM reached 24%, up from 21.4% in Q4, limiting GP decline to ~5% YoY. Beyond the higher GPM in home improvement and rental, Track 1 commission sharing fell 29% YoY, outpacing revenue declines. Strict opex control also helped, with total operating expenses down ~22% YoY vs. revenue down 19%. Marketing fell ~39% YoY as home improvement was intentionally scaled back, reducing promotions and customer acquisition. G&A and R&D dropped ~9% and ~16% YoY, respectively, with G&A benefiting from lower SBC. The improved biz mix plus tight opex delivered the earnings beat.

Longbridge Dolphin Research view:

In absolute terms, BEKE’s quarter was not strong — even with policy support, existing-home transactions in top-tier cities only showed initial repair. New-home sales stayed soft, and the once high-hope home improvement business effectively entered a phase of retrenchment. A high base from last year also pressured growth, keeping revenue deeply underwater YoY.

From an expectations and marginal-changes perspective, Track 1 brokerage declines narrowed and came in a touch better than bearish views. Track 2 revenue, especially home improvement, missed more than expected, but the proactive pullback delivered far better-than-expected margin gains, largely offsetting the shortfall. Rental continued to perform well with counter-cyclical growth.

More importantly, the company has historically expanded profits in downcycles through strict cost control and efficiency, while in macro upcycles incremental investments dilute operating leverage. Today’s cost/efficiency drive coincides with policy tailwinds, raising the chance of a period of dual recovery in revenue and profit if the company avoids unnecessary expansion. Since late Feb, supportive measures across top-tier cities like Shanghai and Shenzhen have been decisive in helping the sector and BEKE bottom and warm up. The stock has rallied ~30% from early Apr lows, pricing in much of this reversal.

Whether there is still upside now depends on whether the recent rebound is a true bottom and trend reversal or just another policy-driven bounce. Recent channel checks and company disclosures suggest April existing-home transactions in major cities pulled back from Mar but remained elevated, with volumes still up YoY. That implies manageable pressure for Q2.

The real question is H2: the base eases, but can the market stay in positive growth? Management argues this is a trend reversal rather than a short-lived rebound and sees demand spreading from low-price rigid demand to higher-ticket upgrade demand in existing homes. Based on current information, we are not yet convinced the sector has definitively bottomed for a sustained recovery.

We therefore still view BEKE as a trading-range opportunity — buy when policy, sentiment, and the share price are at lows, and take profit when policy, sentiment, and results are relatively elevated.

On valuation, current policies should support at least H1. The company expects next-quarter existing-home GTV to turn to low single-digit growth, with new-home GTV roughly flat. H2 is harder to call, but with an easier base, we now look for FY26 revenue to fall ~5% YoY vs. nearly -10% previously. For the year’s adj. net margin, guidance is 8% (vs. ~5% in FY25) and could end up higher.

At 8.5%, FY26 adj. net profit would be around RMB 7.5bn, implying roughly 19–20x PE at the current mkt cap. After the recent rally, valuation again reflects BEKE’s market position with some faith and premium. Further multiple expansion likely requires clear evidence of a trend inflection in housing and earnings rather than a transient rebound.

Detailed takeaways from this quarter:

I. Existing homes: GTV growth and take rate are bottoming

With a series of policy backstops in Feb–Mar, the core existing-home business showed signs of bottoming and stabilization. Due to a high base, GTV fell about 8% YoY but rose ~11% QoQ. Franchises again outperformed Lianjia self-operated stores. Lianjia-led GTV fell nearly 15% YoY, while franchise-led GTV declined less than 4%, likely reflecting store and headcount rationalization and a tilt toward a more resilient platform model.

Aligned with GTV trends, self-operated commission revenue was ~RMB 4.8bn, down ~14% YoY. Platform service revenue rose ~3% YoY vs. franchise GTV down ~4%, indicating a higher platform take rate. We estimate the blended take rate across self-operated and platform at ~1.15% this quarter, up 1.9bps QoQ. As resale activity firms up, BEKE likely offered fewer commission discounts to drive transactions.

II. New homes: weaker in absolute terms, better vs. expectations

Unlike existing homes, new-home sales remain under pressure, consistent with resilience in lower-ticket existing homes post-easing while higher-ticket new builds lag. New-home GTV fell 37% YoY, a slight improvement vs. last quarter and vs. expectations but not by much. Third-party data show top-100 developers’ sales down ~25% YoY in Q1, reflecting both sector headwinds and BEKE’s deliberate pullback in new homes.

The new-home blended take rate fell ~2bps QoQ from a recent high but stayed above last year, so revenue also fell ~37% YoY, broadly in line with GTV. Still, versus a -42% revenue decline in Bloomberg consensus, new homes delivered an expectations beat.

III. Track 2 also posted negative revenue growth

Track 2 revenue was ~RMB 7.67bn, down ~8% YoY. Home improvement revenue fell 21% YoY vs. market expectations of -11%, as weak new-home sales curbed demand and management closed low-quality businesses and retrenched in select cities. This was partially by design but signals a temporary failure to scale BEKE via home improvement.

Rental revenue slipped 1% YoY mainly because ‘Hassle-free Rent’ shifted to net revenue recognition. With managed units up ~47% YoY, underlying growth remained strong. Net-net, Track 1 shows a QoQ bounce but is still down YoY, and Track 2 is also negative for varying reasons, leaving total revenue down about 19% YoY, in line with expectations.

IV. Breaking diseconomies of scale, margins rose against the tide

While revenue was soft across businesses, profit delivery was the highlight. Despite weaker scale, store and headcount optimization (total stores in the BEKE ecosystem fell by ~760 QoQ) lifted efficiency (Lianjia deals per capita +26% YoY; commission per capita +20%), and tight opex control drove broad-based margin gains. Specifically: 1) In Track 1, existing-home margin improvement outpaced new homes on a higher platform mix and better productivity. Contribution margin reached 41.3%, up ~0.9ppt QoQ and 3ppt+ YoY. New homes still improved ~2.3ppt YoY on contribution margin despite falling QoQ.

2) In Track 2, home improvement saw a ~5.6ppt YoY jump in contribution margin after shedding low-quality business and optimizing procurement. Thus, despite revenue missing, contribution profit slightly beat. Rental benefited from a higher mix of net-revenue ‘Hassle-free Rent’, lifting contribution margin by ~4.4ppt QoQ to 14.8%, a new high, with profit ~34% above expectations.

With margin gains across segments, overall contribution profit fell only ~7% YoY (down ~RMB 430mn) despite revenue down ~19%.

V. Efficiency gains and tight opex unlocked profits

Total GPM reached 24% vs. 21.4% in Q4, limiting GP decline to ~5% YoY. Besides stronger GPM in home improvement and rental, Track 1 commission sharing fell 29% YoY, outpacing revenue declines.

Opex discipline also played a major role: total opex fell ~22% YoY vs. revenue -19%. Marketing dropped ~39% YoY as home improvement scaled back and reduced CAC. G&A and R&D fell ~9% and ~16% YoY, respectively, with lower SBC reducing G&A. As a result, adj. net profit was about RMB 1.6bn (+~16% YoY), vs. ~RMB 1.15bn expected, with GPM beating by ~RMB 240mn and opex ~RMB 160mn lower than expected.

<End of text>

Longbridge Dolphin Research on BEKE:

Earnings reviews

Nov 10, 2025 Trans: BEKE (Trans): Cost cuts and efficiency are the absolute priority

Nov 10, 2025 Review: BEKE: Is there still hope?

Aug 26, 2025 Trans: BEKE (Trans): Pruning low-efficiency stores and agents in Beijing/Shanghai

Aug 26, 2025 Review: BEKE: With the market cooling again, can looser purchase caps in BJ/SH turn the tide?

May 16, 2025 Trans: BEKE (Trans): If housing worsens in Q3, more support may come

May 16, 2025 Review: Survive the housing roller coaster, and BEKE will be fine

Mar 19, 2025 Review: Market looks hot, but BEKE just ‘shouts’ without making money?

Mar 19, 2025 Trans: BEKE (Trans): Existing homes keep repairing in 2025, new homes may keep adjusting

Last quarter the company switched to a new disclosure framework, splitting into ‘AI biz.’ and ‘Legacy biz.’. AI includes AI Infra, AI Application and AI-native Marketing Services. Since most Street models still follow the old framework, Longbridge Dolphin Research maps the two to show both the expectation gap and growth across BIDU’s new AI lines.

Specifically (Baidu Core only, ex-iQIYI): The focus below is on Baidu Core and excludes iQIYI. We highlight the key moving parts for revenue and profit.

1) AI now accounts for over half of revenue: AI-related revenue reached RMB 13.6bn in Q1, or 52% of total, up from 43% in Q4. The mix shift toward AI continued to accelerate.

(1) AI Infra (cloud, LLM API, compute leasing) is the real backbone: Revenue came in at RMB 8.8bn, roughly two-thirds of AI revenue, up 79% YoY and ahead of consensus. GPU cloud rose 183% and was the core growth engine.

(2) AI applications (Baidu Wenku, Netdisk and digital employees) weakened further QoQ. Stripping out drag from Netdisk, pure AI apps like digital employees also slowed, which looks out of sync with sector sentiment and likely reflects intensifying competition.

(3) AI-native marketing (agents and digital humans) also softened QoQ. This likely faces the same competitive pressure as the digital employee product line.

2) Legacy ads remain on the floor, as expected: Traditional ads (search, feed ads, etc.) fell 28% YoY, worsening from -26% in Q4. Holiday timing played a role, but sector weakness and BIDU’s eroding competitiveness also hurt, with Mobile Baidu MAU down another 22mn QoQ in Q1, making a near-term inflection hard to envision.

3) Autonomous driving, smart devices and others: This bucket delivered RMB 2.1bn in Q1, down QoQ. Robotaxi completed 3.2mn orders in Q1, up 120% YoY.

Due to Mainland regulatory constraints, Apollo has focused on overseas markets since H2 last year. After tests in the Middle East, the U.K. and Korea, highway testing is now progressing in Switzerland, with Dubai the fastest where operations have started and the Robobus app launched locally in Mar.

Apollo now covers 27 cities globally. This asset could also be carved out and capitalized in the future.

4) Costs down on restructuring: Last quarter already showed sizable workforce optimization (c.RMB 700mn in severance). In Q1, opex fell 10%, taking opex/revenue down to 31% from 45% in Q4.

SBC also fell 24% YoY in Q1. With the market cap higher vs. last year, this implies a sharp reduction in total award recipients.

Despite a sharp decline in high-margin ads, OP improved QoQ but still fell YoY. Relative to expectations, the scale of cost tightening surprised to the upside.

5) Higher investment, buyback on plan: Capex nearly doubled YoY to RMB 5.8bn in Q1. Even so, it remains low vs. other LLM/cloud leaders, and AI Infra revenue of RMB 8.8bn fully covers capex.

Under the 2-yr $5bn buyback announced in Feb, BIDU repurchased $172mn in Q1 (roughly one month), a pace that looks a bit light. We will watch management’s plans for the rest of the year on the call.

Management appears more focused on shareholder returns than before, and BIDU holds nearly RMB 100bn of net cash (ex short-term borrowings). Cash flow is ample to fund buybacks and capex over the next two years.

The implied shareholder yield is still around 5–6% before dividends. If the buyback runs at the cap, $2.5bn per year compares with a ~$46bn market cap.

6) Kunlun Chip IPO approaching: Kunlun filed for IPO guidance in early May. Based on typical HK procedures, an early listing window could be Jul–Aug, and as domestic compute is a hot theme this year, Kunlun’s IPO should offer a near-term catalyst for BIDU’s valuation.

7) Detailed financials at a glance

Longbridge Dolphin Research View

AI Cloud is what can carry BIDU near term, including LLM API, compute leasing and the Kunlun chip stack. The sector backdrop remains strong, lifting most vendors regardless of product gaps.Among these, Kunlun looks the most competitive and offers the most upside optionality.

What could Kunlun be worth? Longbridge Dolphin Research ran the math in the early-year note ‘Kunlun IPO Accelerates: BIDU’s Google Moment?‘. The framework remains broadly valid.

(1) There is not much change in the operating outlook so far, though Street revenue expectations are ~RMB 6–7bn, making our prior RMB 8.4bn look optimistic. Upside risks include Samsung capacity constraints and potential pricing pressure in H2 as rival shipments bunch up amid performance gaps.

(2) Versus listed domestic GPU peers, a 15x P/S looks conservative. Kunlun is ASIC while the ‘Big Four’ use GPGPU architectures, yet the end goal for customers is similar.Clients ultimately benchmark chips on performance, stability and price.

Kunlun’s R&D and supply-chain execution are not behind peers, and it has ample HBM reserved to support shipments. Current orders are mainly from internal use at BIDU, plus Tencent, China Mobile, Geely, China Southern Power Grid and China Merchants Bank. With BIDU pivoting harder to AI Cloud this year, Kunlun is well placed to capture the current upcycle.

The more the BIDU story leans on Kunlun, the more the valuation needs Kunlun’s catalysts. Without progress and fresh positive news, the stock will likely drift with the market and get pulled back by legacy-biz logic.

Opportunity still clusters around the Kunlun IPO window, i.e., Q3. On valuation, the current ~$46bn market cap implies ~21x 2026E P/E, which screens rich vs. China ADR peers, but as the IPO nears, SOTP becomes more appropriate.

(1) Ads (incl. AI-native marketing): assign 4x P/E on RMB 56.5bn revenue (-10%), 30% OPM and 15% effective tax to get ~$8.3bn. This embeds further weakness and execution risk in the legacy engine.

(2) Kunlun: on RMB 7.5bn revenue, 40x P/S, 59% ownership and a 30% holdco discount, we get ~$13bn. This reflects scarcity and the near-term listing catalyst.

(3) Cloud ex-Kunlun: assume +20% to RMB 43.6bn revenue. At 4x P/S, that suggests ~$25.5bn.

<1–3> sum to ~$46.6bn, essentially in line with the current ~$46bn. Some institutions also value Robotaxi separately and add net cash and the iQIYI stake.

Longbridge Dolphin Research thinks that approach embeds peer-premium risk: add Robotaxi only when there is a concrete spin move. The iQIYI stake is volatile on weakened fundamentals and fits only in a bullish case.Adding all net cash is also not reasonable; at most add the $5bn buyback plan, implying a 5–6% shareholder yield.

In early trading after Kunlun’s listing, if macro sentiment holds and BIDU advances a dual-primary listing, a short-term spike is possible. That will be more about trading timing.

Deep Dive

I. Biz. structure

BIDU is unusual among internet names in giving highly granular splits: 1) Baidu Core covers legacy ads (search/feed) and new initiatives (Intelligent Cloud/DuerOS smart speakers/Apollo), plus AI revenue across AI Infra (cloud, LLM API, compute leasing), AI Application (Baidu Wenku, Netdisk, digital employees) and AI-native marketing (agents and digital humans). The framework clarifies growth and profitability drivers.

2) iQIYI: membership, ads and content sublicensing. Disclosures are detailed as it is separately listed.

The two are cleanly separated, and iQIYI has standalone data for triangulation. There is ~1% (RMB 200–400mn) intercompany offset between the two, so our Baidu Core rebuild may differ slightly from reported figures, but the trends are unaffected.

II. AI contribution rose to 52%

Baidu Core revenue was RMB 26.0bn (+2% YoY). AI growth essentially offset declines in legacy businesses.

(1) AI biz. After a year of ramp, AI Infra is the true pillar, with GPU cloud as the engine. The real demand is for compute, and BIDU’s full-stack AI with integrated soft/hard capabilities remains relatively scarce.

AI apps derived from LLMs appear to have hit a near-term ceiling, with both AI applications (Wenku, Netdisk, digital employees) and AI-native ads (agents, digital humans) flat to weakening QoQ. Momentum has clearly cooled.

(2) Legacy biz. There is little to add here, as the market does not expect a near-term turn. In Q1, Mobile Baidu lost the AI-portal traffic battle and kept shedding MAUs, with even seasonal strength failing to help.

Recasting into the old framework for comparison: (1) Ads fell 21.5% YoY, with traditional ads down 28%. Holiday timing played a part, but underlying pressure remains.

(2) Cloud and other revenue rose 42% YoY off a high base, beating expectations. AI-driven demand underpinned the strength.

III. Higher capex, leaner headcount

BIDU made sizeable org changes late last year to integrate legacy ops, drive efficiency, and dedicate owners for each AI sub-line. Last quarter was a transition; this quarter already shows tangible effects.

(1) Core GPM fell to 44% as ads weakened and AI cloud mix rose. The mix shift weighed on margins.

(2) Core OP was RMB 3.4bn with a 13% OPM, reflecting the early impact of layoffs. S&M and R&D fell 17% and 4%, respectively.

IV. Capex expanding, more room to go

Q1 FCF turned negative as capex nearly doubled YoY to RMB 5.8bn. That is still modest vs. other LLM/cloud leaders, and AI Infra revenue of RMB 8.8bn fully covers capex.

Part of this reflects cost advantages from BIDU’s full-stack AI and in-house chips. Given the cash buffer and sector heat, BIDU has room to lean in more aggressively.

<End here>

Longbridge Dolphin Research on ‘Baidu’ – Past Notes:

Recent earnings season

Feb 26, 2026 Trans ‘Baidu (Trans): Sharper Focus on Shareholder Returns and Execution’

Feb 26, 2026 Results Review ‘Baidu: Ad Moat Eroded, Is Kunlun the Lifeline?’

Nov 18, 2025 Trans ‘Baidu (Trans): AI-Native Marketing as the Second Growth Curve’

Nov 18, 2025 Results Review ‘Talking Up AI Hard — Can Baidu Be Saved?’