The following is compiled by Dolphin Research for$MEITUAN(03690.HK)'s Q2 2026 earnings call Trans

I. Key Takeaways

1. Shareholder returns and investment monetization: Buybacks remain the primary way to return capital, with pacing tied to the competitive landscape, cash flow and offshore liquidity. Management noted its holdings in high-quality companies are worth over RMB 70bn on current marks; when timing is right, it will consider exits or monetization to recycle capital into core ops.

2. Q3 and full-year guidance

a. Food delivery UE: Significant YoY improvement, but seasonality will weigh QoQ and UE should remain positive in Q3. Industry subsidies are still far above 2024 levels and will take several quarters to normalize.

b. Per-order delivery cost: Higher in Q3 vs. Q2 due to heavier peak-season marketing and extra heat-season and peak subsidies for riders. In addition, nationwide roll-out of occupational injury insurance from Jul 1 adds another cost layer.

c. In-store, hotel & travel: OPM could decline vs. Q2 given heavier investments in Q3 and Q4.

d. New initiatives: 2026 segment OP loss will not exceed 2025, and this includes Keeta and fresh grocery retail under the same disclosure.

3. Key metrics this quarter

a. Aggregate: Revenue RMB 104.6bn (+14.4% YoY) with net profit turning positive. Segment OP totaled RMB 3.9bn, and adj. net profit was RMB 2.5bn.

b. Opex ratios: COGS ratio fell to 66.5% and S&M ratio to 23.6%, both driven by better marketing efficiency. R&D ratio rose to 7.3% on heavier AI spend, while G&A held flat at 3.1%.

c. Segments: Core local commerce revenue RMB 71.5bn (+10.1% YoY, accelerating), with segment OP turning positive to RMB 5.7bn. New initiatives revenue RMB 33.1bn (+25% YoY) with segment loss narrowing QoQ to RMB 1.7bn.

d. Mix: Merchandise sales revenue grew 78.9% YoY. Delivery services and merchant services growth both re-accelerated, and food delivery revenue returned to YoY growth.

4. Cash and investments

a. Cash, cash equivalents and short-term investments totaled RMB 168.3bn at end-Jun, with the investment portfolio near RMB 77.3bn. Liquidity remains solid.

b. FV changes on some investments generated RMB 22.2bn this quarter, booked in OCI rather than the P&L and thus not impacting current profit. This accounting view differs from the earlier reference to over RMB 70bn of equity value.

II. Detail from the earnings call

2.1 Management commentary

1. Food delivery and Meituan Flash (on-demand retail) a. Core user stickiness strengthened, with frequency, retention and basket size all moving up.

b. 'Pin Haofan' enhanced demand recognition and accelerated rollout of chain best-sellers. 'Shen Qiangshou' kept raising supply-quality standards for mid-to-high-end consumers, and innovative formats like brand satellite stores expanded steadily.

c. Flash continued healthy new-user acquisition, with especially fast growth among post-2005 cohorts, while order frequency among existing users rose. Lightning warehouses grew rapidly and remained a key growth engine.

d. 'Waimai Wine Delivery by Waimai' expanded to most provinces and added more private labels, launching an anti-counterfeit verification system for premium liquor in Jun. 'Squirrel Convenience' accelerated penetration via customized development and factory partnerships, focusing on night-time and travel scenarios.

2. Platform ecosystem and merchant enablementa. In Q2, ten food-safety measures were rolled out comprehensively, forming an end-to-end framework spanning pre-onboarding checks, in-process monitoring, anti-fraud enforcement and social co-governance.

b. Launched the industry's first integrated AI solution for on-demand retail. Around holiday peaks, provided 220k SMB restaurants with funding, supplies, equipment, AI tools and store renovation support.

c. Initiated a special program in Q2 to boost lower-tier markets and county economies, expanding to nearly 400 counties nationwide in H2. The plan aims to help over 500k SMBs go digital.

d. Rider occupational injury coverage has been rolled out nationwide on a per-order, per-rider basis. The 'red-light stop the clock' initiative launched first in Beijing, and together with pension subsidies and critical-illness care will form the rider benefits framework.

3. In-store, hotel & travela. Local services consumption is shifting from standardized spend to personalized, experiential and emotion-driven demand. The company is steering the sector from 'pick the right merchant' toward 'pick the right artisan and experience'.

b. The new Must-Eat List is generated from 1.5bn authentic user reviews, now covering 264 cities/regions worldwide with 120 newly added. The Must-Play List is expanding into more immersive, interactive and interest-driven scenarios.

c. AI assistant 'Xiaotuan' saw rising usage in complex local-decision scenarios, boosting user engagement. Features launched include Pick-up Now, online reservations, order-ahead, smart queuing and in-store smart ordering.

d. Through the CatPaw platform, dedicated AI agents are provided to F&B, local services, retail, healthcare and travel to help merchants improve day-to-day ops efficiency.

4. New initiativesa. Xiaoxiang Supermarket accelerated expansion with private-label share of GTV continuing to rise. The 'Happy Monkey' community format had 40 stores operating by end-Q2.

b. Keeta improved growth and efficiency concurrently, with Hong Kong achieving stable profitability. The Middle East saw further QoQ efficiency gains, and Brazil is focused on São Paulo.

2.2 Q&A

Q: How is the competitive landscape evolving in food delivery and on-demand retail? What is Meituan's share trend in mid-to-high basket orders?

A: We widened our lead in orders and GTV QoQ, especially in mid-to-high baskets. Management sees the sector pivoting toward marketing and operating efficiency, with competition gradually returning to quality, service and innovation. Further regulatory guidance on subsidies should foster healthier development and create a level playing field for companies with real core capabilities.Over recent months, our advantages in user mix, order mix and operating efficiency have strengthened. Stronger member benefits, expanded high-quality supply and better service have driven deeper engagement from premium users. Flash remains the industry leader, and management called the shift in expectations for convenience and reliability irreversible, with penetration still early across cohorts, while acknowledging pressure from last year's high base and the macro environment.

Q: As subsidies normalize, how will food-delivery UE trend QoQ in Q3?

A: Q3 UE will improve significantly YoY but be seasonally softer QoQ, and should stay positive. Industry subsidies remain far above 2024 levels and will need several quarters to normalize.Q3 is the peak season for instant delivery driven by summer demand. We will step up marketing QoQ to capture the year's highest demand window and add heat and peak-season subsidies for riders to ensure service quality, lifting per-order delivery costs vs. Q2. The nationwide roll-out of occupational injury insurance from Jul 1 adds another cost layer.Management emphasized recent volatility mainly reflects seasonality and deliberate trade-offs among scale, profitability and ecosystem. UE recovery has clear visibility, and over the medium term should normalize to reasonable levels.

Q: What role does LongCat-2.0 play in the AI strategy post open-sourcing? Is there a commercialization path?

A: It is positioned as internal infrastructure, with no intent to compete in token factory businesses. The AI strategy rests on three pillars: building models, using AI to work, and embedding AI in products. For Meituan, AI is not about model showdowns but about using AI to remake the organization, products and workflows.LongCat-2.0 is the next-gen in-house foundation model, and management said it is among the first, possibly the first, trillion-parameter model trained entirely on domestic compute infrastructure. With open weights, it has been deployed across core internal scenarios such as software development and operations, customer service and AI agents, showing clear gains in coding, reasoning, tool use and complex task execution.A fully domestic training and inference stack is seen as a structural edge in long-term cost and infrastructure control. On the product side, the in-app AI assistant 'Xiaotuan' continues to be upgraded, moving from understanding complex multi-constraint queries toward full agentic execution and ultimately closing the loop from discovery and decision to transaction and fulfillment.

Q: How should we think about AI investment impacts near term and long term?

A: No quantified guidance was given, with a reiteration of ROI discipline and capital allocation rigor. The near-term priority is to keep building capabilities and driving real AI deployment in operations, using self-developed models and AI products to strengthen the core, improve user and merchant experience and raise internal efficiency.On organization, more employees are using AI tools and the share of AI-generated code keeps rising. Aside from the R&D ratio rising to 7.3% due to AI spend, management did not disclose AI-related capex or payback cadence.

Q: How is competition in in-store services? Are peers shifting to monetization and profit? Is the macro a visible drag?

A: Subsidy-driven users acquired by peers are mostly price-sensitive with weak repeat behavior, and our core users and merchants are not materially affected. Today's in-store market is larger than a few years ago, with more and more varied players, and growing differentiation by user cohorts, scenarios and merchant tiers.In recent quarters, competitors have invested more in local services via standalone shelf-style apps, diverting content-driven traffic to new apps with heavy subsidies. This brought in many price-sensitive users, especially from lower-tier cities, but their repeat purchase is weak. Our in-store GTV quality and redemption rates continue to lead major competitors by a wide margin.The macro has indeed compressed basket sizes in some categories, but local services are more resilient than e-commerce. Service retail still has low online penetration and a long runway, and we now serve over 8mn merchants across 200+ categories, with sports/health and immersive entertainment rapidly moving online.

Q: How will you balance growth and margins in in-store for H2? Is the margin recovery path clear?

A: Q3 and Q4 OPM could decline vs. Q2, with improvement coming in the medium term after competition normalizes. We will step up investment in H2 to seize growth opportunities and reinforce our position among core categories, users and merchants.We will also drive more local merchants to go digital, aiming to evolve from a 'customer acquisition channel' to a 'merchant operating platform', and gradually become an AI partner that helps merchants operate smarter. We will keep trimming low-ROI spend, improve resource allocation and leverage synergies between in-store and on-demand retail, while giving no specific OPM range or recovery timetable.

Q: How is Xiaoxiang Supermarket performing? Where is the omni-channel strategy now?

A: We now cover 68 cities, have opened the fifth offline store, and stick to an omni-channel route combining front warehouses with offline flagship stores. Management believes online penetration in fresh retail is still too low with ample headroom, and the right approach is to run online and offline stores on a single platform.On-demand retail is changing grocery shopping: instead of morning wet-market trips or weekly stock-ups at hypermarkets and warehouse clubs, consumers now order on demand because they trust 30-minute delivery. Once the habit forms, purchase frequency rises meaningfully and per-capita spend potential is large over time.Xiaoxiang accelerated city coverage to 68 by Q2, with strong GTV growth and steady efficiency gains. Offline rollout timing: Beijing first store in Dec last year, Ningbo second in Apr, Hangzhou third in Jul, Ningbo fourth in Aug, and the fifth in Shenzhen opened on the day of the call.Front warehouses drive rapid cross-city expansion and cover most target consumers. A small number of offline flagships deliver see-smell-touch experiences to build trust in products and the brand.

Q: What is 'Happy Monkey' aiming for? How do you plan long-term investment in fresh retail?

A: 'Happy Monkey' is a community format with small footprints and high private-label mix, with 40 stores by end-Q2 and still very early. Unlike Xiaoxiang's offline flagship value proposition, Happy Monkey stores are smaller and more flexible with curated SKUs, focused on high value-for-money goods for communities; the two are positioned as differentiated and complementary, with ops and merchandising still being honed.Longer term, management sees the true moat in fresh retail as organizational and supply-chain capabilities, which take time to build but unlock a large market once in place. Confidence stems from strong demand for quality fresh goods beyond Tier-1 cities, as consumers in mid-sized cities and even affluent counties have both spending power and appetite for good products.The company will keep strengthening merchandising and deepening supply-chain synergies across fresh retail businesses, advancing in a disciplined and sustainable way.

Q: How is Keeta progressing in the entered markets? How will you approach Brazil?

A: Hong Kong turned positive in 29 months and Saudi in 22 months, and Brazil will first focus on São Paulo, which accounts for 25% of the country's market. Launched in May 2023 in Hong Kong, Keeta's UE turned positive in Oct 2025; Saudi, entered in Sep 2024, reached profitability in Jul this year, faster than Hong Kong, and Saudi is a much larger market.Management believes this validates that operating playbooks can be replicated across markets by mastering fundamentals. Consumers everywhere want better selection, better prices and more reliable, faster delivery, while merchants care about incremental orders, commission rates and dependable fulfillment.Brazil is one of the world's top five delivery markets, still growing fast with low penetration, but very different from our existing markets. We will stay flexible and iterate as we go, first nailing operations and building differentiation in São Paulo before considering broader expansion.

Q: What is the overseas investment cadence for H2? Any guidance on new-initiatives losses?

A: Overseas focus is shifting to operating optimization in current markets, and 2026 new-initiatives losses will not exceed 2025. This guidance already includes both Keeta and fresh retail. Management did not break out budgets by single market and did not give specific spend for Brazil.

Q: With competition shifting to efficiency in food delivery, how are capital allocation priorities set?

A: Core businesses come first, with others evaluated dynamically by ROI, and AI investments handled prudently. Management reiterated a consistent ROI mindset anchored on long-term value creation, seeing irrational competition in core local commerce as unsustainable and focusing on high-quality growth and operating efficiency.Overseas, Keeta in Hong Kong and Saudi has shifted to operating optimization, and expansion pacing in each country will be governed by ROI. In fresh retail, management is constructive on Xiaoxiang's long-term potential and confident in ongoing efficiency gains, while treating AI as a major strategic opportunity but investing carefully to embed AI in real business scenarios to improve user experience, merchant ops and organizational productivity.

Q: Will you monetize investment assets? Can buybacks continue?

A: Equity holdings are worth over RMB 70bn at current valuations, with exits considered when timing is right; buybacks remain the main return tool. Buybacks have been the primary way to reward shareholders, executed at meaningful scale over the past years, and will continue subject to the competitive backdrop, cash flow and offshore liquidity.On investments, the company will periodically assess monetization opportunities in quality holdings. Beyond financial returns, some positions have strategic value, offering deeper insights and proximity to frontier technologies, and the company will weigh market conditions, valuations, funding needs and overall capital allocation priorities to open exits or monetize selected stakes when timing is appropriate, without giving a timetable or naming targets.

<End of text>

Risk disclosure and statement:Dolphin Research Disclaimer and General Disclosure