Author: Hao Xin

Editor: Wu Xianzhi

Recently, Alibaba unveiled a financial report that showcases a mixed bag of performance results.

The silver lining is that Alibaba's investment streak has finally started to pay dividends.

For the quarter ended March 31, 2026, Alibaba reported interest income and net investment gains of approximately RMB 33.823 billion, a stark contrast to the RMB 7.516 billion loss incurred during the same period last year. Extending the analysis to the entire fiscal year 2026, Alibaba's investment earnings soared to RMB 87.512 billion, marking a staggering 3,015% year-over-year surge.

However, the crux of the issue lies in the fact that Alibaba's profits are predominantly sustained by investments.

A glaring piece of evidence is that in the first quarter of fiscal year 2026, Alibaba's net profit surged by 96% year-over-year. Yet, after stripping out investment gains, the actual net profit dwindled to a mere RMB 86 million, nearly negligible compared to the same period.

This predicament can be traced back to Alibaba's three major "cash drains": investments in instant retail, the construction of a super AI-to-Consumer (AI to C) entry point, and escalated cloud infrastructure expenditures.

Beyond the superficial financial figures, Alibaba is unveiling its AI commercialization strategy to the market.

On the B-side (business side), leveraging its cloud platform as the cornerstone and the Qianwen model as the nucleus, Alibaba delivers scalable intelligent capabilities to enterprise clients through Model-as-a-Service (MaaS) and AI-native software. On the C-side (consumer side), centered around the Qianwen App, Alibaba is integrating its internal ecosystem, potentially transforming into a conversion hub between B-side and C-side operations.

As Wu Yongming aptly put it, “Alibaba stands at a pivotal juncture where technological investments are beginning to bear commercial fruit.”

The 'Well-Funded' Qianwen App

In the fourth-quarter financial report for fiscal year 2026, one figure leaps out: a 100% year-over-year decline in Non-GAAP net profit. Behind this near-zero profit, one name consistently emerges—the Qianwen App.

This financial report offers the first quantitative glimpse into how Alibaba is “sparing no expense” to nurture this C-side AI product—and the hefty price tag attached.

Although the Qianwen App's profit statement is not separately disclosed in the financial report, its "cash-draining" scale can be clearly outlined through abnormal changes in several key indicators.

Alibaba's quarterly sales expenses skyrocketed from RMB 36.18 billion to RMB 53.42 billion, an increase of RMB 17.24 billion or 47.6%. This increment consists of two parts: subsidies and promotions for instant retail (Taobao Flash Sales) and user acquisition costs for the Qianwen App.

The financial report mentions that Alibaba's free cash flow turned negative during this quarter, shifting from an inflow of RMB 3.7 billion to an outflow of RMB 17.3 billion. This was primarily due to investments in instant retail, user acquisition for the Qianwen App, and increased cloud infrastructure spending, indirectly confirming the cash costs Alibaba has incurred to support the Qianwen App and its AI strategy.

Beyond user acquisition costs, expenses for the Qianwen App likely encompass product development, computational resource allocation, and more. Thus, Alibaba's total quarterly investment in the Qianwen App far exceeds the RMB 3 billion “hospitality” expenses.

For leading domestic AI model startups, a single funding round typically ranges from RMB 1-3 billion, with annual total financing amounting to approximately RMB 5-8 billion. Alibaba's quarterly investment in the Qianwen App already matches the annual financing volume of these startups, explaining why major corporations hold an absolute advantage, leaving AI startups with little room to compete. ByteDance's Doubao and Alibaba's Qianwen both leverage group-level profits and cash flow to secure a leading position in AI C-side entry points.

The "well-funded" approach extends beyond monetary investment, evident in the unconditional allocation of group-level traffic. The Qianwen App did not start from scratch but was lifted by Alibaba's ecosystem—a “traffic aircraft carrier.”

Currently, the Qianwen App is deeply integrated into ecosystems such as Taobao, Alipay, Maps, and Fliggy, covering super entry points across payment, travel, and tourism scenarios. These apps collectively boast billions of monthly active users, with significant cross-referrals among them. Positioned at their intersection, the Qianwen App gains access to multiple traffic spigots simultaneously.

Losses for the Qianwen App are inevitable, but Alibaba is calculating the broader ecological leverage.

Every yuan spent by Alibaba is amplified severalfold by ecological leverage: traffic leverage reduces customer acquisition costs, R&D leverage spreads technological investments, and scenario leverage accelerates the data flywheel effect. This is precisely why the Qianwen App, despite massive losses, remains pivotal to Alibaba's AI strategy.

Alibaba Cloud in Transition

If the Qianwen App represents Alibaba's AI “dagger” aimed at the C-side, then Alibaba Cloud is the true heartland of this transformation.

The fourth-quarter financial report for fiscal year 2026 provides the first clear evidence that Alibaba Cloud's transformation has moved beyond “storytelling” into an acceleration phase verifiable through financial data and calculable business models.

Alibaba Cloud's quarterly revenue reached RMB 41.626 billion, a 38% year-over-year increase. Revenue from external clients grew even faster, accelerating to 40%. AI-related product revenue hit RMB 8.97 billion, maintaining triple-digit growth for the 11th consecutive quarter and accounting for 30% of external client revenue. Alibaba's management predicts that within a year, AI revenue will exceed 50% of cloud business revenue, becoming the absolute driver of growth.

Alibaba Cloud's revenue structure is shifting. The 30% threshold is critical—when a new business contributes over 30% of total revenue and grows far faster than the overall portfolio, it signals a pivot from marginal innovation to a growth engine.

Meanwhile, MaaS has driven a business model upgrade. Annual Recurring Revenue (ARR) from models and applications, including the Bailian Platform, exceeded RMB 8 billion in Q4 of fiscal year 2026, potentially surpassing RMB 10 billion in Q1 of fiscal year 2027 and RMB 30 billion by the end of fiscal year 2027.

Wu Yongming explicitly stated that Alibaba Cloud's gross margin will “significantly improve” over the next 1-2 years, with visible changes expected within 1-2 quarters. Currently, the adjusted EBITA margin for the Cloud Intelligence Group stands at around 9%. As MaaS's share grows, this margin could rise to 15-20% within two years.

While financial reports hint at a commercially viable future for AI-driven cloud services, Alibaba Cloud remains in transition.

Alibaba Cloud has developed distinct advantages. Its full-stack self-developed Infra layer serves 60% of its computational power externally, lowering unit Token costs. The self-developed Qianwen model eliminates third-party revenue sharing, retaining most MaaS income. Depreciation costs for traditional cloud products have been amortized over years, keeping marginal costs for new Tokens extremely low.

The company has a strong base of enterprise clients, including governments, financial institutions, and manufacturers, who can smoothly transition from traditional cloud services to MaaS. Alibaba's e-commerce, logistics, and payment scenarios naturally generate Token consumption, allowing large-scale model testing. High-cost-efficiency inference solutions developed in-house can then be offered to external clients.

However, this creates a “sweet” burden for Alibaba Cloud's transformation. Traditional Infrastructure-as-a-Service (IaaS) is not a disposable asset but a foundation providing steady cash flow and client relationships, supporting early-stage AI transformation investments.

Yet, the core logic of this foundation is “selling resources,” while the new engine revolves around “selling intelligence.” These conflicting logics create internal friction during the transition. An inertial approach persists, where Alibaba Cloud drives MaaS consumption through IaaS resource stickiness, resulting in a dual identity of “selling resources + selling intelligence.”

In practice, conflicts may arise. If sales teams must meet both IaaS revenue/profit targets and promote MaaS services, clashes occur. Prioritizing low-priced IaaS bundles could encroach upon budgets for pure Token clients, while focusing solely on Tokens might reduce renewal rates for traditional cloud products. Teams must balance these objectives.

Alibaba Cloud's transformation is more evident in its approach to Tokens. Unlike Volcano Engine's aggressive strategy, Alibaba adopts a gradual transition. Last year, Alibaba Cloud focused externally on IaaS while internally shifting to Token-based metrics. This year, the focus has moved from pure Token consumption to Token revenue.

According to media reports, Alibaba Cloud set a short-term Token revenue growth target in early April 2026: by May 15, daily Token revenue should increase fivefold from early April levels.

Alibaba's AI Commercialization Blueprint

A persistent question in China's AI sector is whether large models can generate profits. Startups rely on financing, while giants mask losses through financial reports. The market lacks a verifiable commercialization model—until now.

Alibaba's AI commercialization strategy comprises a three-tier structure: infrastructure, model platform, and application ecosystem, integrated through organizational strategy to align middle and top layers.

The bottom Infra layer features Alibaba's mass-produced T-Head “Tianhai” self-developed chips, the full-modal and full-size Qianwen model family, and Alibaba Cloud's decade-long infrastructure development. This layer addresses cost and supply foundations for AI commercialization.

The middle MaaS layer consists of the Bailian Platform, model APIs, and AI-native software. It packages model capabilities into standardized Token services, commodifying intelligence. Enterprises invoke APIs via the Bailian Platform, paying per Token with low marginal costs, flexible pricing, and strong client retention—the current core source of AI revenue.

The top application layer features the Qianwen App, which directly engages users, accumulates interactions through ecosystems like Taobao, Alipay, and Gaode, and connects Alibaba's B-side and C-side operations, generating both traffic and value.

Between the MaaS and application layers, Alibaba established the ATH Business Group, integrating model R&D, MaaS business lines, C-side applications, B-side applications, and AI innovation divisions. While the three-tier architecture defines Alibaba's strategic scope, ATH acts as the brain, directing execution.

ATH dismantles “silos” between model and application teams by placing providers and consumers of model capabilities under one organization, directly managed by the CEO. It does not add a new layer but encapsulates MaaS and application layers, ensuring the shortest feedback loops and fastest iteration cycles organizationally.

The market is the ultimate test of business models. By making Token revenue a core metric, Alibaba Cloud essentially uses real client consumption to validate AI's value. This reflects a commercialization logic: the value created by models correlates with how much enterprises are willing to pay for Tokens.

Tokens offer a present-day solution, but profitability remains a future challenge. Alibaba must continually prove through quarterly financial reports that short-term profit sacrifices can secure long-term, irreplaceable intelligent barriers in future market competition.

WeChat ID: TMTweb

Official Account: Photon Planet