Profits Collapsed, Stock Price Wrong: Where Is Alibaba's Real Growth Engine?

The harsh reality of the capital markets is that they tend to reward certainty in the present and punish uncertainty about the future.

All “misjudgments” are essentially due to lagging valuation models. When a giant ship has already changed course, but onshore observers are still using old charts to measure its position, deviations occur.

While the market’s spotlight remains on the old framework of “e-commerce companies,” trying to value Alibaba based on traditional GMV (Gross Merchandise Volume) and profit margins, the latest financial report actually signals a clearer, yet often overlooked, message — the company is undergoing a profound identity shift: from a simple transaction matching platform to a productivity engine centered on cloud computing and artificial intelligence.

Why this is a “misjudged” financial report

— The pricing misconception behind profit decline

On the surface, this is indeed a report that, from a traditional financial perspective, “falls short of expectations”: slowing revenue growth, significant drops in net profit, and a sharp contraction in free cash flow.

If we only judge Alibaba based on these cold metrics, the problem lies in the observer’s lens — the market still interprets it through the framework of a “mature e-commerce company,” expecting stable dividends and buybacks rather than aggressive expansion.

However, the real business picture is that the company is experiencing a typical “profit-for-structure” strategic transformation.

Why are profits declining? It’s not due to collapsing demand in core business or mismanagement, but because two main initiatives are simultaneously ramping up investments: one is the fulfillment network for instant retail (flash sales), and the other is a arms race in AI and cloud infrastructure.

This is precisely the moment where capital markets are most prone to misjudgment. When a company actively compresses short-term profits to secure a second growth curve, its short-term financial statements often look the worst. In the eyes of long-term investors, this “bad appearance” actually signals a deepening moat.

Even more noteworthy is that the decline in cash flow is also being over-interpreted. Free cash flow down 71% year-over-year, which under traditional value investing frameworks is a major negative — indicating weakened cash-generating ability.

But if viewed through the current AI cycle, this is essentially the result of front-loaded CapEx (capital expenditure). In the era of large AI models, compute power equals power, and compute requires real money for hardware investments.

Compared to US tech giants, this phenomenon is not unfamiliar. Whether it’s Amazon’s early costly investments in AWS data centers or Microsoft’s large-scale GPU procurement and data center construction to embrace AI, their profit statements have faced transitional pressures. Wall Street once questioned whether Amazon’s logistics investments were excessive or whether Microsoft’s cloud services would drag down profits.

The difference is — after decades of technological iteration, the US market has learned to price “future profits” and is willing to assign higher valuation tolerances to high capital expenditures; meanwhile, Alibaba is still being priced based on “current profits,” with the market lacking patience for its investments, viewing them as costs rather than investments.

This is the essence of the misjudgment. The market’s fear isn’t the investments themselves, but the inability to confirm whether these investments can translate into future barriers. But in this AI war that must be won, current investments are precisely the tickets to the future.

Flash sales + AI are not two separate lines,

but a closed-loop experiment of “AI e-commerce”

Market analysts tend to break down Alibaba’s actions into isolated parts: on one side, the aggressive subsidies and investments in “Taobao Flash Sale” in instant retail; on the other, large-scale capital expenditure on cloud computing and AI. Investors often see these as two separate departments burning money independently.

But from the underlying logic of industry evolution, it’s actually one thing — Alibaba is trying to run the most realistic and profitable path to AI commercialization: e-commerce.

This quarter, instant retail revenue hit 20.8 billion yuan, up 56% year-over-year, far outpacing traditional e-commerce growth. This isn’t just a simple growth of a new business segment; more critically, it’s changing the fundamental structure of e-commerce — shifting from the traditional “search + transaction” model to a “dispatch + fulfillment” model.

This is precisely the scenario where AI is most suited to intervene. Traditional e-commerce is “people searching for goods,” relying on keyword search and recommendation algorithms; instant retail is “goods finding people” combined with “timely delivery,” relying on high-precision real-time dispatch of logistics, inventory, traffic conditions, and user demand.

Once the Tongyi Qianwen app connects Taobao, Flash Sale, Amap, Fliggy, and other ecosystems, a new interaction paradigm is forming: users no longer just “place orders,” but “give commands”; platforms are no longer passively “displaying products,” but actively “executing tasks.”

What does this mean?

It means e-commerce is shifting from information matching to real-world execution (Physical AI).

Within this framework, the significance of flash sales is redefined — it’s not just a tool for GMV growth, but an “execution layer interface” for AI deployment. Every instant delivery is a training opportunity for AI dispatch algorithms; every user’s immediate demand is a micro-tuning for the large model’s understanding.

Therefore, Alibaba’s AI path differs from pure model companies. Many AI startups face the biggest challenge of “models without scenarios,” only selling APIs or subscriptions, with long commercialization cycles. Alibaba, instead, trains, validates, and monetizes AI directly within high-frequency, high-cash-flow commercial scenarios.

From an investment perspective, this is crucial — because it means AI is no longer just a cost center on financial statements, but a productivity element that can directly generate revenue. When AI reduces fulfillment costs, improves delivery efficiency, and accurately predicts inventory, it directly translates into profits. This “internal cycle” AI commercialization is more resilient than merely expanding external compute power.

A moment of revaluation — Alibaba’s core assets are shifting from e-commerce to AI

If we dissect this financial report, what truly deserves revaluation isn’t the peaked e-commerce business, but three underlying capability curves. These three curves form Alibaba’s valuation foundation for the next decade.

First, the accelerated return of cloud computing.

This quarter, cloud revenue grew 36% year-over-year, faster than the previous quarter’s 34%, reaching 43.3 billion yuan, with an annualized scale exceeding 170 billion yuan. In the current global AI infrastructure cycle, this growth itself signals — Alibaba Cloud has re-entered the “main upward trend.”

More importantly, this growth isn’t driven by traditional enterprise cloud needs, but by incremental AI-related demand. AI-related products have achieved triple-digit growth for ten consecutive quarters, pointing to a conclusion: Alibaba Cloud is becoming the core infrastructure for AI. In China, most large models are trained and infer on Alibaba Cloud, making it a “water seller” role that’s often more stable than gold rush prospectors.

Second, the explosion of models and C-end entry points.

Tongyi Qianwen’s monthly active users surpassed 300 million in February, an extremely rapid pace in global AI applications. Compared to most AI products still at the tool stage, with users just trying out and leaving, Tongyi Qianwen is already entering real consumption scenarios through “intelligent agents.”

This means Alibaba not only has model capabilities but also distribution channels and a closed commercial loop. Having 300 million active users provides vast feedback data, the most valuable fuel for model optimization. When models become sufficiently intelligent, they can directly invoke Taobao payments, Amap navigation, Ele.me delivery, forming a super-intelligent agent.

Third, autonomous control of compute power and chips.

The open-source PanGu GPU has achieved mass production and supports the full chain from training to inference. This is especially critical amid current global compute shortages and geopolitical complexities — it determines whether Alibaba has long-term cost advantages and security.

When these three lines converge, Alibaba’s valuation logic must change. It’s no longer just an internet company surviving on ad and commission revenue, but a tech infrastructure company with compute base, model brain, and execution capabilities.

However, the current market pricing still heavily anchors on e-commerce, assigning a valuation similar to traditional retail.

This is very similar to Amazon’s early days — before AWS’s explosive growth, the market undervalued its cloud infrastructure, viewing it as an ancillary cost of e-commerce. Only when AWS’s profit share surpassed that of e-commerce did the market realize and revalue the tech giant.

A more conflicting comparison is with US AI giants.

On the surface, they show better profits and steadier growth, but a neglected issue is that the depreciation costs of AI infrastructure are lagging in reflection. US accounting standards allow longer depreciation periods, so when these huge capital expenditures enter depreciation, profits will also be pressured. In other words, current “high profits” are largely a timing mismatch.

Alibaba, on the other hand, entered this stage earlier. It chose to bear the upfront costs, front-loading expenses — financially aggressive, but strategically a “head start.” When US giants face depreciation cliffs in the future, Alibaba may have already passed its investment peak and begun benefiting from scale effects and declining marginal costs.

It’s also worth noting that in AI CapEx, US giants like Meta, Alphabet, and Amazon are increasingly issuing bonds to support investments, with some experiencing temporary free cash flow pressure. Alibaba, however, is one of the few companies advancing both fronts simultaneously (AI + instant retail). This isn’t risk — it’s a demonstration of capability. It shows the company’s strong cash-generating ability, supporting a dual-wheel strategic transformation.

Conclusion: The market is underestimating

not a company, but a path

After this financial report, it quickly made headlines on CNBC and other foreign financial media, sparking widespread debate. Essentially, it indicates that foreign and institutional investors are beginning to reassess this company, but disagreements remain.

The core issue isn’t whether Alibaba’s short-term profits are under pressure, nor whether its stock will fluctuate further, but whether the market is willing to admit that its core driving forces have fundamentally changed.

E-commerce is becoming a “cash flow foundation,” responsible for providing stable capital inflow;

Cloud computing and AI are becoming the “valuation core,” responsible for offering future growth imagination and resilience.

Once this cognitive shift is complete, the current valuation system will become invalid. We can no longer measure an AI infrastructure company with e-commerce P/E ratios, nor dismiss long-term technological barriers with short-term cash flow.

Until then, all volatility, doubts, and even “misjudgments” are essentially different expressions of the same thing — the market has yet to see this company through new eyes. For investors, the greatest risk isn’t buying a company investing in the future, but using an old map to seek new lands.

This is not just about Alibaba’s valuation recovery, but also about China’s tech industry shifting from model innovation to technological innovation — a revaluation of value. This path will be crowded and full of misunderstandings, but only those who pierce through the fog can see the dawn of a new cycle.