Ahead of Consensus.
Intelligence across tech and capital markets, for investors, policymakers, and business leaders.
9 minute read
Alibaba Group Holding Limited posted revenue of RMB268.95 billion ($39.64 billion) for the quarter ended June 30, 2026, a 9 percent increase from RMB247.65 billion a year earlier and a narrow beat against the RMB268.88 billion consensus tracked by LSEG. Read on its own, that line suggests a company executing cleanly. Read alongside the rest of the income statement, it suggests something closer to a company in the middle of an expensive experiment. Net income fell 75 percent to RMB10.44 billion ($1.54 billion), and operating income dropped 57 percent to RMB15.16 billion, the combined weight of a goodwill impairment, a European regulatory fine and capital expenditure that outgrew nearly everything else on the balance sheet.
The contradiction is deliberate, or at least defensible, and management did not pretend otherwise. Chief Executive Eddie Wu described the quarter as proof that “the AI and cloud commercialization inflection point has arrived,” leaning on an AI product revenue line that has now grown by triple digits for twelve consecutive quarters. Chief Financial Officer Toby Xu offered the more disciplined counterpoint, crediting cloud margin expansion and steady e-commerce profitability with giving Alibaba room to keep spending. Headcount backs that framing: Alibaba employed 132,165 people as of June 30, barely above 131,462 three months earlier, evidence that this is a capital story rather than a headcount one, a distinction worth weighing before deciding whether the spending is temporary or structural.
If one number explains why Alibaba’s stock has behaved better than its earnings suggest it should, it is 45. That is the year-over-year growth rate of AI Cloud and Compute Services revenue, which reached RMB48.44 billion ($7.14 billion), and it is also the growth rate of revenue from external customers within that segment, confirming that the surge is not simply Alibaba shifting internal workloads onto its own infrastructure. AI-related product revenue alone reached RMB12.38 billion. Segment profitability, more tellingly, grew faster still: adjusted EBITA for AI Cloud and Compute Services more than doubled, up 133 percent to RMB5.63 billion, the clearest evidence yet that fixed infrastructure costs are beginning to be absorbed by genuine scale rather than subsidized by corporate largesse.
Underneath that acceleration sits a hardware strategy that gets less attention than it deserves. T-Head’s Zhenwu chip family, anchored now by the newly deployed Zhenwu M890 processor, has been adopted by more than 650 external customers across upwards of 20 industries, from autonomous driving to financial services, a customer base broad enough to suggest genuine product-market fit rather than a handful of flagship pilots. Omdia’s most recent China AI cloud market study placed Alibaba Cloud first with a 38.1 percent share, ahead of every domestic competitor by a wide margin. Together, the chip portfolio and the market position give Alibaba something few Chinese peers can claim: the ability to meet AI infrastructure demand largely on its own terms, insulated from the export restrictions and supply constraints that complicate rivals’ plans.
Alibaba spent the quarter reorganizing its retail businesses with a bluntness that suggests urgency, not routine housekeeping. China and international e-commerce, Freshippo and parts of Cainiao were folded into a single Alibaba E-commerce Group, now disclosed across four lines: China E-commerce, China Quick Commerce, International E-commerce and Global Wholesale. Group revenue rose 4 percent to RMB205.86 billion, a modest headline number that obscures a much sharper divergence underneath. China Quick Commerce, built on Taobao Instant Commerce and Freshippo, grew 45 percent to RMB53.30 billion, aided by higher average order values and a deliberate tilt toward higher-margin food and non-food categories, all while the business held its market share rather than buying growth at any cost.
The older core of the business did not fare as well, and Alibaba did not try to disguise it. China E-commerce revenue fell 8 percent to RMB110.90 billion, with customer management revenue down 7 percent, though stripping out the contra-revenue effect of its new merchant subsidy program would have left that figure up 1 percent on a like-for-like basis, a caveat worth taking at face value. Elsewhere the picture brightened: AliExpress turned an operating profit on the strength of logistics efficiency gains, and 88VIP, Alibaba’s premium membership tier, grew to roughly 64 million members, up by double digits year over year. Group adjusted EBITA held nearly flat at RMB39.75 billion, a result that reads less like stagnation and more like a business absorbing intense competitive pressure in quick commerce without letting it bleed into the rest of the portfolio.
Four line items explain most of the gap between Alibaba’s operating story and its reported one. Capital expenditure reached RMB67.68 billion, up 75 percent from RMB38.68 billion a year earlier, pushed higher by rising compute capacity requirements, uneven procurement timing and higher chip component prices. That spending turned free cash flow negative to the tune of RMB44.67 billion, more than double the RMB18.82 billion outflow of a year ago. Alongside it sat a RMB4.46 billion goodwill impairment tied to businesses within the All Others segment and a provision for a EUR550 million fine levied by the European Commission under the Digital Services Act, a reminder that operating at global scale now carries global regulatory exposure as a fixed cost of doing business.
The most expensive bet, in relative terms, sits inside AI Labs and Applications, home to Alibaba’s model research, the Qwen consumer app and the newly introduced QwenWork enterprise agent. The segment’s adjusted EBITA loss widened to RMB13.86 billion from RMB3.22 billion a year earlier, driven by heavier model investment and rising inference costs as usage climbs. Some of that spending surfaced in August with Qwen3.8-Max, a 2.4 trillion parameter flagship model released with open weights barely three months after its predecessor, alongside a fresh set of coding, video, audio, image and music models. Qwen app has now introduced roughly 250 million users to agentic shopping features since launch, a figure leadership treats as proof the losses are financing a genuine flywheel between AI adoption and commerce, not research for its own sake.
Investors walked into this report having already made up their minds, at least provisionally. Alibaba’s Hong Kong shares had climbed roughly 36 percent over the quarter, and the American depositary shares had recovered to around $128 after touching a 52-week low in the low $90s in late June, a decline compounded by a broader selloff across Chinese technology stocks and, separately, by Anthropic’s public allegation, reported by the Wall Street Journal, that accounts linked to Alibaba had misused its Claude models. By the eve of the results, sentiment had turned firmly bullish again: a Strong Buy consensus, and price targets from JPMorgan, Barclays and Susquehanna reaching as high as $205, $195 and $185. Morgan Stanley’s Gary Yu had specifically forecast 45 percent cloud growth with margins near 11 percent, alongside continued e-commerce softness, a split the results ultimately confirmed almost exactly.
The market’s response to the numbers themselves was more cautious than the run-up implied. Alibaba’s American shares fell as much as 4 percent in premarket trading before recovering most of that ground to close near 1.6 percent lower, as investors weighed the cloud outperformance against the scale of the profit decline. The company still holds RMB474.51 billion in cash and liquid investments and repurchased $162 million of shares during the quarter, evidence that management views the earnings hit as a matter of timing rather than deterioration. With capital expenditure unlikely to ease before the fiscal year turns, the question facing Alibaba’s shareholders is no longer whether the company can grow its AI business. It is whether the market will keep extending the patience that growth currently requires.