---
title: "BABA (Trans): Flash Sale 2-yr target RMB 1tn; AI + Cloud 5-yr target RMB 700bn---"
type: "Topics"
locale: "zh-HK"
url: "https://longbridge.com/zh-HK/topics/39382281.md"
description: "Below is Dolphin Research's Trans of $Alibaba(BABA.US) FY26 Q3 earnings call.For our full take, see 'Alibaba: E-com Slips Again, AI Is the Lifeline'.1) Shareholder returns: as of Dec 31, 2025, the company held net cash of $42.5bn. Excluding debt maturing beyond five years, net cash exceeds $60bn, underscoring balance-sheet strength that supports continued investment.2) Performance trends: consolidated revenue was RMB 284.8bn. Excluding Sun Art Retail and Intime revenue, ..."
datetime: "2026-03-19T15:13:01.000Z"
locales:
  - [en](https://longbridge.com/en/topics/39382281.md)
  - [zh-CN](https://longbridge.com/zh-CN/topics/39382281.md)
  - [zh-HK](https://longbridge.com/zh-HK/topics/39382281.md)
author: "[Dolphin Research](https://longbridge.com/zh-HK/news/dolphin.md)"
---

> 支持的語言: [English](https://longbridge.com/en/topics/39382281.md) | [简体中文](https://longbridge.com/zh-CN/topics/39382281.md)


# BABA (Trans): Flash Sale 2-yr target RMB 1tn; AI + Cloud 5-yr target RMB 700bn---

**Below are Dolphin Research's compiled notes on**$Alibaba(BABA.US) **FY26 Q3 earnings call. For our earnings analysis, see '**[**Alibaba: E-Comm Slips Again, AI Is the Lifeline**](https://longbridge.com/en/topics/39382054)**'.**

**I. Key takeaways**

1\. **Shareholder returns**: As of Dec 31, 2025, net cash stood at $42.5 bn. Excluding debt maturing in 5+ years, net cash exceeds $60 bn, underscoring a balance sheet that supports sustained investment.

2\. **Trend**: Consolidated revenue was RMB 284.8 bn. Excluding Sun Art and Intime, like-for-like revenue rose 9%, while GAAP net profit came in at RMB 15.6 bn, down 66% YoY.

3\. **Cash flow**: Operating cash flow was RMB 36.0 bn, and FCF RMB 11.3 bn, down RMB 27.7 bn YoY, mainly due to strategic reinvestment in AI and quick commerce.

4\. **China Commerce**: CMR grew just 1%, pressured by weak macro consumption, a warm winter, and a later Spring Festival. Quick commerce revenue reached RMB 20.8 bn (+56% YoY). Adj. EBITDA for China Commerce was RMB 34.6 bn, down 43% YoY.

5\. **Cloud Intelligence**: External customer revenue rose 35% YoY (vs. 29% last quarter). AI-related product revenue delivered triple-digit growth for the tenth consecutive quarter, with adj. EBITA margin around 9%.

**II. Earnings call details**

**2.1 Management highlights**

1\. **AI strategy and full-stack build-out**

a. Two strategic pillars: 'AI+Cloud' and 'Consumption'. We are entering the Agent-driven era, requiring tight integration between models and applications. **b. 5-year target: combined external revenue from Cloud+AI (incl. MaaS) to exceed $100 bn.** c. Launched the Alibaba Token Hub Biz. Group (ATH), integrating Tongyi Lab, MaaS, Tongyi Qianwen BU, Wukong BU, and AI Innovation BU.

2\. **Cloud Intelligence**

a. Market share has risen for three consecutive quarters to 36%, extending our lead. **b. In the first 11 months of FY26 (to Feb), Alibaba Cloud's cumulative external revenue surpassed RMB 100 bn.** **c. Token consumption on Model Studio grew 6x over the past 3 months, and we expect MaaS to become Cloud Intelligence's largest revenue product.**

3\. **T-Head (AI chips)**

a. In-house GPU chips have reached scaled mass production, with cumulative shipments of 470,000 AI chips as of Feb 2026. b. Over 60% of T-Head chips serve external customers, supporting AI workloads for 400+ enterprise clients. **c. Annual revenue has reached the RMB 10+ bn range.**

4\. **Foundational models and AI apps**

a. Released Tongyi Qianwen 3.5 Plus, with standout reasoning, coding, and Agent capabilities, and materially improved reasoning efficiency. b. A next-gen model optimized for coding and Agent scenarios will launch soon. c. Tongyi Qianwen consumer MAU exceeds 300 mn, with deeper integration across Taobao, Alipay, Fliggy, and Gaode during Spring Festival.

d. Tongyi Qianwen open-source models have topped 1 bn cumulative downloads on Hugging Face. e. Rolled out 'Wukong', an enterprise AI Agent platform.

5\. **Consumption**

a. Quick commerce drove double-digit MAU growth for the Taobao app. b. Platform AAC rose by 150 mn in 2025, including 100 mn from traditional e-comm physical goods AAC, exceeding the prior three-year total. c. AIDC revenue grew 4%, while EBITDA losses narrowed meaningfully YoY.

6\. **Other biz.**

a. Other revenue was RMB 67.3 bn, down 25% YoY due to divestments of Sun Art and Intime. b. Other EBITA loss was RMB 9.8 bn, mainly reflecting higher investment in large models and consumer-facing Qianwen. c. Unallocated adj. EBITA loss was RMB 2.7 bn (vs. RMB 0.2 bn loss a year ago), reflecting talent retention incentives.

**2.2 Q&A**

**Q: How will the Token Hub Biz. Group (ATH) change Cloud and AI coordination? What are management's priorities between Cloud and AI?**

A: The rationale for ATH aligns with the current trajectory of AI. By late 2025 and early 2026, AI entered the Agent-driven era, which differs from earlier phases. In Agentic AI, tight integration between models and applications is essential. Earlier, model training leveraged largely static datasets, whereas in the Agent era, much data comes from client-side interactions, demanding deeper model-app integration.

Across layers of AI deployment — apps, models, AI infra, and chips — the most important shift in Agentic AI is the need for tight app-model integration. This is the top priority.

**On ATH's internal coordination,** industry trends are clear: AI Agents will be closely embedded with the application layer, and app diversity will be vast. **In 2C, we are scaling the Tongyi Qianwen app as a personal assistant; in 2B, Wukong is positioned as an enterprise Agent platform.** Vertical AI apps will proliferate across industries, serving diverse scenarios, **and all of this requires a strong MaaS layer.**

MaaS will support internal apps and serve a large base of external and industry-specific AI use cases, underpinning sizable value and a large TAM. From the model-app lens, our **top priority is building the smartest models.** Strong models are necessary to drive AI deployment across industries and to attract application usage of our MaaS offerings.

To build the strongest models, we must collaborate closely with industries and with our own 2C/2B apps, leveraging data flywheels to keep improving model capabilities. This is one reason we formed ATH now. In short, **the top priority is enhancing model capabilities, which requires coordinated effort across the model stack, applications, and infra.**

**Q: CMR growth slowed in the Dec qtr. How is CMR trending in the Mar qtr? Any signs of improvement?**

A: The Dec qtr was challenged by weak macro consumption, a warm winter, and a later Spring Festival. With a longer promo season, we increased consumer protection spending vs. prior years, which weighed on CMR and EBITA trends. Into the Mar qtr, we see improving consumer confidence and momentum from quick commerce, with physical goods GMV and CMR significantly recovering vs. the Dec qtr. EBITA is expected to improve accordingly.

**Q: For Flash Sale, is the priority market share or unit economics? What synergies exist between quick commerce and traditional e-comm?**

A: **We are expanding share while materially improving UE,** driven by better fulfillment logistics, higher monetization, and optimized order mix, with further improvements expected in coming quarters. Flash Sale has a strong positive impact on traditional e-comm and the broader ecosystem. Platform AAC rose by 150 mn in the past year, including 100 mn from traditional physical goods AAC, exceeding the total of the prior three years.

New users have lower ARPU and purchase frequency vs. existing users. Our **goal is to continuously lift ARPU and purchase frequency,** making this a new growth engine in the coming years. **Flash Sale has notably boosted categories like fresh food and health, accelerating growth for Freshippo and Tmall Supermarket.**

Looking ahead, we **maintain our FY28 Flash Sale GMV target of \>RMB 1 tn.** We expect to be cash-flow positive when reaching that GMV, **and expect Flash Sale to turn profitable in FY29.** Quick commerce has become the cornerstone of e-comm, strategically critical in the AI era — fueling acquisition, engagement, diverse demand, transactions, monetization, and supporting logistics infra.

We are committed to **heavy investment over the next two years to hit the RMB 1 tn GMV target as a market leader.**

**Q: On T-Head chips, is there a spin-off/IPO plan? Can you share more operating metrics, shipment-to-revenue linkage, growth outlook, and internal vs. external use and competitiveness?**

A: T-Head is integral to Alibaba's group-wide AI strategy. Within China's domestic AI chip ecosystem, we believe T-Head is at the top tier in tech and product capabilities. The product suite covers the full AI workflow from training and fine-tuning to inference, and T-Head AI chips are used at scale across training and inference via Alibaba Cloud.

**Over 60% of T-Head chips serve external commercial customers via Alibaba Cloud's public and hybrid products,** across internet, financial services, autonomous driving, and smart manufacturing. These clients use T-Head chips for both training and inference workloads. On the software stack, T-Head has strong compatibility with the Linux ecosystem, enabling easy client migration.

**T-Head's significance goes beyond narrowing process and performance gaps vs. overseas peers.** Recognizing that our chips still lag in certain aspects, we **are deepening co-design with Alibaba Cloud infra and Tongyi Qianwen models to deliver superior cost-performance.** This is T-Head's key differentiation vs. other chipmakers — **the primary goal is exceptional cost-performance, lowering inference costs and improving AI efficiency.**

Given China's AI context, **another critical advantage is compute supply assurance.** Global AI compute will be extremely scarce over the next 3–5 years, especially in China. As China's only cloud provider with in-house chip R&D, T-Head is vital to the group — expanding AI compute supply will bolster Cloud and AI (incl. MaaS) growth momentum.

Over the **past two years, T-Head successfully commercialized with cumulative shipments exceeding 470,000 and annual revenue in the RMB 10+ bn range.** Looking to 2026–2027, we expect capacity for high-quality AI chips to keep expanding, providing strong compute for the group's AI and becoming a powerful growth driver. We also expect profitability to improve further. Overall, T-Head's value lies not only in cost optimization but in ensuring supply-chain resilience in a compute-scarce era.

On IPO, we do not rule out the possibility of a future listing, but there is no set timeline currently.

**Q: On the 5-year $100 bn AI+Cloud revenue target, can you share the implied CAGR, growth drivers, and timing for sustained margin improvement in Cloud?**

A: We **do believe AI and Cloud-related revenue can exceed $100 bn in five years,** based on market momentum and product strength. The fundamental driver is continued breakthroughs in large-model capabilities. In the first two months of 2026, we already see models handling complex B2B workflows, with more enterprises deploying Agent-driven end-to-end business tasks.

This signals a structural change in how IT budgets are viewed. Many enterprises no longer treat token consumption as IT spend, but as part of overall operating or R&D costs. Tokens are becoming a factor of production rather than an IT expense — the most fundamental long-term driver we see.

The biggest growth drivers are threefold: first, MaaS-led business as the core engine. MaaS growth will be supported by diversified use cases, including our own apps and industry/customer scenarios, making MaaS the key revenue driver for AI and Cloud. Second, beyond public MaaS, many mid-to-large enterprises need internal inference and training. This market will persist — firms will choose between public MaaS APIs and private deployment based on business models, security needs, and scenarios.

These use cases represent a major incremental growth opportunity for Alibaba Cloud AI infra. Third, **a critical but overlooked driver — CPU-centric traditional cloud has huge room to expand in the AI era.** Traditional cloud serves IT engineers (in China, likely a few million to <100 mn), but there may be billions of Agents created by large models, which will require massive traditional CPU, databases, storage, and memory.

The challenge is transforming traditional cloud from a platform for human users into one adapted for Agents, which is a focus of Alibaba Cloud's upgrade this year. As revenue scales, **AI will transition from 'selling resources' to 'selling intelligence',** while integrating T-Head chips to lower costs and drive efficiency.

We believe Cloud margins will become increasingly clear and improve steadily with AI and Cloud scale. The path will not be linear — scale effects or T-Head ramp could lead to step-changes in profitability. Regarding 2026–2031 CAGR, **growth will not be linear.** Some investments made today may take 1–2 years to translate into significant growth, but we are confident in hitting the 5-year target.

**Q: You previously said e-comm is in a 3-year investment cycle. Has this view changed with Flash Sale and Agentic Commerce? Where are we in the cycle?**

A: As noted, we are investing heavily in on-demand retail this year. We see high-certainty opportunities here and will invest aggressively over the next two years to exceed RMB 1 tn GMV in quick commerce. We also believe this investment will deliver positive economic returns for the overall e-comm business two years out.

We must also consider AI. AI will profoundly impact e-comm, but at AI's pace, three years is too long — AI progress is measured in weeks or months. That is why we are investing heavily in AI, rolling out new experiences for consumers and merchants and upgrading merchant operating models. We believe AI will drive major upgrades across e-comm and huge deployment opportunities in B2B, which we will pursue actively.  

<End\>  
  
**Risk disclosure & statements:**[**Dolphin Research Disclaimer & General Disclosure**](https://support.longbridge.global/topics/misc/dolphin-disclaimer)

### 相關股票

- [BABA-W (09988.HK)](https://longbridge.com/zh-HK/quote/09988.HK.md)
- [Alibaba (BABA.US)](https://longbridge.com/zh-HK/quote/BABA.US.md)
- [BABA-WR (89988.HK)](https://longbridge.com/zh-HK/quote/89988.HK.md)
- [ANT GROUP (688688.CN)](https://longbridge.com/zh-HK/quote/688688.CN.md)
- [SUNART RETAIL (06808.HK)](https://longbridge.com/zh-HK/quote/06808.HK.md)
- [Alibaba HK SDR 5to1 (HBBD.SG)](https://longbridge.com/zh-HK/quote/HBBD.SG.md)

## 評論 (5)

- **貝多芬彈不出緬A的傷 · 2026-03-19T16:48:05.000Z · 👍 1**: Two days ago, just before the price increase, the company bought 400,000 RMB worth of computing power, thinking of supporting Alibaba. With these earnings results out, it's estimated to drop significantly tomorrow.
- **新用户_edUQJZ · 2026-03-19T15:32:53.000Z · 👍 6**: A complete mess, a bunch of amateurs. They can't even hold on to or do well in their main business, and the AI development plan is also weak and lacking in confidence.
  - **Lan&yuan** (2026-03-20T02:59:20.000Z): You go?
- **赛博摸鱼选手 · 2026-03-19T15:29:11.000Z · 👍 1**: Among the US stocks, Oracle, Microsoft, and Google have some basis for their pie-in-the-sky promises, relying on RPO. But Alibaba, like OpenAI and Anthropic, just makes empty promises with their mouths😂