Alibaba invests 67.7 billion yuan in AI computing, pauses buybacks in Q1 2027

On August 20, Alibaba announced its Q1 fiscal year 2027 financial results, with operating profit declining 57% year-over-year to RMB 15.161 billion, and GAAP net profit plunging 75% year-over-year to RMB 10.444 billion. However, the cloud business performed strongly, with Alibaba Cloud’s external revenue increasing 45% year-over-year and AI-related products achieving an annualized revenue of RMB 49.5 billion, while capital expenditures for the quarter reached RMB 67.678 billion. The e-commerce group reported revenue of RMB 205.862 billion, up 4% year-over-year, with instant retail revenue reaching RMB 53.295 billion, up 45% year-over-year. The AI applications segment posted a loss of RMB 13.861 billion, with Qwen App users reaching 250 million. The market initially sold off on the profit collapse, then rebounded on the accelerated cloud growth, signaling that Alibaba’s valuation anchor is shifting from being perceived as an e-commerce giant to an AI infrastructure company.

Article author and source: Wall Street Journal

On the evening of August 20, Alibaba released its financial results for the first quarter of fiscal year 2027. Upon release, the market immediately offered two interpretations: U.S. stocks opened 4.2% lower pre-market and dropped as low as $121.88 during trading (a 5.4% decline from the previous day’s close of $128.90), but rallied strongly by the close to $130.53 (+1.26%). In other words, the market first sold off on the perception of a profit collapse, then bought back in on the realization that cloud computing is truly accelerating.

The surface numbers appear disappointing: operating profit of RMB 15.161 billion, down 57% year-over-year; GAAP net profit of RMB 10.444 billion, down 75% year-over-year; Non-GAAP adjusted diluted ADS earnings per share of RMB 8.52, 24% below the Bloomberg consensus estimate.

However, in the same financial report, Alibaba Cloud's external revenue increased by 45% year-over-year, AI-related product annualized recurring revenue (ARR) reached RMB 49.5 billion, and quarterly capital expenditures surged to RMB 67.678 billion.

The "fake stumble" in profits and the "real turning point" in business are packed into the same report, driving the entire tension of this quarter’s earnings.

Operating profit halved, net profit dropped by three-quarters—at first glance, it looks like a profit storm. But when you break down the numbers layer by layer, you’ll find that much of this storm is inflated.

JPMorgan's research report highlighted four non-operating noise factors suppressing EPS: First, the effective tax rate rose to approximately 40% from previous levels; second, an impairment loss of RMB 4.458 billion was recognized for goodwill (related to the "All Others" segment); third, net interest and investment income declined from the previous range of RMB 20-30 billion per quarter to RMB 9.004 billion, halving year-over-year; fourth, Alibaba was fined €550 million by the EU’s Digital Services Act, prompting the company to fully provision for the penalty.

Remove this layer, and adjusted EBITDA amounts to RMB 39.143 billion, a 14% year-over-year decline. According to J.P. Morgan data, this figure exceeds the market consensus by 6% and surpasses J.P. Morgan’s own forecast by 7%. In other words, Alibaba’s core operating foundation actually outperformed consensus—the reported -75% is an illusion driven by accounting noise.

If the income statement is a "reverse mirror," then segment data is another, forward-facing mirror. Alibaba Cloud (listed in the financial report as the "AI Cloud and Computing Services" segment) reported quarterly revenue of RMB 48.437 billion, up 45% year-over-year—the fastest growth rate in 22 quarters, surpassing the previous quarter's 40%.

More importantly, the structure: AI-related product revenue reached RMB 12.376 billion this quarter, marking the 12th consecutive quarter of triple-digit growth. Management disclosed that the annualized revenue (ARR) from AI-related products has surpassed RMB 49.5 billion, approximately USD 7.3 billion, accounting for 35% of external cloud revenue.

Management has also provided its own anchor: AI-related product annualized revenue next quarter is expected to approach $10 billion. Goldman Sachs, UBS, and Jefferies all consistently predict that cloud business revenue growth in the September quarter will accelerate further to over 50%.

The profitability of the cloud business is also improving in tandem. The segment’s adjusted EBITA reached RMB 5.628 billion, up 133% year-over-year; the EBITA margin rose to 12% (compared to 7.2% in the same period last year). In simpler terms: for every RMB 100 in revenue, the cloud business now retains RMB 12 in operating profit—nearly RMB 5 more than a year ago.

Goldman Sachs, UBS, and Jefferies all view this as a core bullish case for Alibaba. UBS maintains a Buy rating and raises its target price to $206, corresponding to an 18x FY27E P/E ratio, stating that this valuation is "not expensive for a Chinese AI supply chain intermediary."

The cloud's acceleration doesn't come for free. Alibaba's capital expenditure this quarter reached RMB 67.678 billion, a 75% year-over-year increase and more than two and a half times that of the previous quarter (RMB 26.9 billion). This caused free cash flow for the June quarter to turn negative by RMB 44.67 billion, compared to a negative RMB 18.8 billion in the same period last year.

Buybacks have also been put on hold: Only $162 million (13.4 million shares) were repurchased this quarter, compared to previous patterns of billions of dollars—capital has now been prioritized for computing power.

Will the money invested be recovered? Alibaba CEO Wu Yongming provided a concrete metric: at the current average gross margin, AI-related capital expenditures are expected to break even within approximately three years; as gross margins improve, the payback period could shorten to 2.5 years or even two years.

How is this commitment supported? During the call, Wu Yongming outlined three pillars: First, under the current gross margin levels of AI products, investment generates positive cash flow as long as growth is below 33%. Second, large-scale deployment of self-developed chips (T-Head Zhenwu M890), serving over 650 external customers across more than 20 industries, directly boosts gross margins by replacing external procurement. Third, the V100s purchased in 2018 and the A100s purchased in 2020 are still operating at nearly full capacity, with “actual service life significantly exceeding the theoretical depreciation period”—effectively shortening the payback window.

Third-party research has reached similar conclusions. Jefferies, in its research report, believes that with product portfolio optimization and the adoption of in-house developed chips, the payback period is expected to shorten to 2 to 2.5 years. Goldman Sachs has accordingly raised its capital expenditure forecast for fiscal years 2027/2028 to RMB 210 billion and RMB 240 billion, respectively.

On the other hand, the reality is more restrained: of the RMB 380 billion AI investment plan over three years, approximately RMB 190 billion has been invested by the end of this quarter—just half completed; meanwhile, AI applications (such as the Qwen app) are still burning cash. Whether next quarter’s returns are “solid enough” will be the next variable the market reprices.

Shift the focus from cloud to the e-commerce group—the largest revenue source for Alibaba (RMB 205.862 billion, accounting for 77% of total revenue). This quarter, e-commerce group revenue increased 4% year-over-year, with adjusted EBITA at RMB 39.749 billion, down just 1% year-over-year.

Behind the numbers lies an "internal migration": China’s instant retail revenue (including Taobao Flash Purchase, Hema, and instant delivery) reached RMB 53.295 billion, up 45% year-over-year, making it the fastest-growing segment within the group; customer management revenue (CMR) from China’s traditional e-commerce (Taobao and Tmall main platforms) declined 7% year-over-year, but grew approximately 1% on a comparable basis after excluding the impact of reverse revenue subsidies from the “New Business Development Plan.”

Alibaba emphasized this was a "conscious accounting adjustment"—reclassifying subsidies previously recorded as sales expenses to a contra-revenue reduction. As a result, the apparent -7% figure is inflated; even after adjusting for consistency, the core e-commerce business has merely stabilized, not reversed.

At the user level, the number of 88VIP members on Taobao and Tmall continued double-digit growth, reaching approximately 64 million as of June 30, making them the most valuable customer segment. Overall revenue for international e-commerce (including Lazada and AliExpress) declined slightly by 1%, but AliExpress achieved operating profitability for the first time. Comparatively, JD.com’s revenue for the June quarter declined by approximately 3% year-over-year, while Pinduoduo’s revenue for the March quarter increased by approximately 10% year-over-year—but Alibaba’s differentiation lies in cloud computing, a business segment that JD.com and Pinduoduo do not have. E-commerce is the foundation; cloud is the pricing power.

While cloud computing is "making money," AI applications are "burning cash."

This quarter, Alibaba integrated the Qwen series, Qwen App, and QwenWork into the "AI Lab and Applications" division, reporting an adjusted EBITA loss of RMB 13.861 billion, more than triple the loss of RMB 3.224 billion in the same period last year. The primary reasons for the loss were increased inference costs for the Qwen App and greater investments in AI capability development.

However, user metrics are growing rapidly: since the launch of the Qwen App, 250 million users have experienced AI-driven shopping through its agent features; the Qwen model series has been downloaded over 3 billion times globally, with more than 300,000 derivative models built upon it. The Qwen3.8-Max, released in August, boasts 2.4 trillion parameters and ranks among the world’s top-tier models. UBS estimates that as the marketing efficiency of the Qwen App improves and model training costs decline, this segment’s losses will stabilize in the range of RMB 11–12 billion. In other words, RMB 13.8 billion is likely the peak loss.

What to watch next quarter? Three most direct variables.

First, did the cloud revenue growth for the September quarter meet the 50% guidance—this is the first test of the AI narrative. Second, can MaaS (Model-as-a-Service) annual recurring revenue reach RMB 30 billion according to management’s口径. Third, will the loss in the AI applications segment narrow to RMB 11–12 billion, as expected by UBS?

If all three of these variables materialize, the cost of the profit fake-out will be worth paying; if any one of them fails to materialize, the market will once again reclassify Alibaba within the "e-commerce stock" valuation range.

This quarter’s earnings report serves as a watershed: Alibaba’s valuation anchor is shifting from “China’s e-commerce giant” to “China’s AI infrastructure company.” But while the anchor has changed, whether the hull beneath the water can hold up is the question the next chapter will answer.