Alibaba's Profit Plunges 75% as AI Capex Hits $9.5 Billion a Quarter

On August 20, 2026, Alibaba delivered the clearest snapshot yet of what the AI infrastructure race actually costs: quarterly net income collapsed 75% year over year to roughly RMB 10.4 billion (about $1.6 billion), even as revenue grew 9% to RMB 268.95 billion ($39.6 billion) and its cloud business posted its fastest growth in 22 quarters. The stock fell roughly 5% following the report.

The gap between those two numbers — top-line growth and bottom-line collapse — is the entire story of this phase of the AI boom. Alibaba is not struggling to sell AI. It is spending almost everything it earns, and then some, building the capacity to sell AI.

The Numbers

Alibaba’s fiscal Q1 2027 (the June quarter) laid out the trade in stark terms:

• Revenue: RMB 268.95 billion (~$39.6B), up 9% year over year, beating analyst estimates
• Net income: RMB 10.4 billion (~$1.6B), down 75% (WSJ cites 76% to RMB 10.55B)
• Cloud & AI revenue: up 45% to RMB 48.4 billion, a 22-quarter high for external cloud revenue growth; AI-related product revenue alone reached RMB 12.4 billion with triple-digit growth
• Capex: RMB 67.7 billion (~$9.5B) in a single quarter, up 75% year over year
• Free cash flow: swung to a RMB 44.67 billion outflow
• Margin: net margin compressed to roughly 6%

CEO Eddie Wu, who has explicitly said he will prioritize AI growth over near-term profits, framed the quarter as proof of positioning rather than pain: Alibaba Cloud, he told analysts, is in a “superior position” for the AI boom. Management pointed to a three-year, RMB 380 billion (~$53B) AI infrastructure plan and suggested the spending could begin paying off within about three years.

The Circular Financing Shadow

The profit collapse arrives as a broader anxiety about AI economics crystallizes. Bloomberg’s coverage of the earnings explicitly paired “Alibaba’s AI spending spree” with “concerns of circular AI financing” — the practice, now common across the industry, in which chipmakers and hyperscalers take equity stakes in AI labs or neocloud providers, who then commit to buying compute right back from their investors.

The Bank for International Settlements flagged this dynamic in its 2026 annual report, calling circular financing the “most prominent” structural risk in the AI buildout: vendors effectively finance the purchase of their own products, which reinforces the appearance of demand. Short-seller Michael Burry has argued the arrangement could artificially prop up both demand and revenue across hyperscalers, AI labs, and chipmakers. A recent WSJ analysis estimated Big Tech’s true AI spending commitments run some $3 trillion higher than headline figures suggest once debt, leases, and off-balance-sheet vehicles are counted.

Alibaba itself is entangled in this web on multiple sides — as a cloud provider selling AI compute, as a Qwen-model developer needing compute, and reportedly as a participant in the China AI ecosystem’s vendor-financing loops. The company even sold its gaming unit for a reported $1.5 billion recently, partly to fund the bigger AI push that Qwen’s momentum demands.

Why China’s Biggest AI Bet Matters Globally

Alibaba has become the West’s reference case for Chinese AI ambition. Qwen models are among the most downloaded open-weight families in the world, and Alibaba Cloud’s 45% external revenue growth shows real, paying demand — not just model releases and benchmarks. When the CEO of China’s largest cloud operator says AI is the company’s “most certain growth engine” and backs it with $9.5 billion in a single quarter of capex, that is a signal about where the global compute market is heading.

But the numbers also expose the strain. A 6% net margin at Alibaba’s scale leaves little room for error. The RMB 44.67 billion negative free cash flow means the buildout is now debt- and reserve-funded, not cash-flow-funded. If AI demand keeps compounding at 45%, the bet looks visionary. If it stalls — or if circular deals across the industry turn out to have inflated the demand signal — Alibaba will have bought a decade of depreciation for revenue that never fully materialized.

The three-year payback framing from management is notably more cautious than the industry’s mood a year ago. Even the most aggressive spenders are now talking about timelines rather than infinities.

What to Watch

1. Cloud growth durability. 45% is a 22-quarter high partly because the base was depressed. The next two quarters reveal whether this is a compounding trend or a backlog burn-off.
2. Capex trajectory. The RMB 380 billion three-year plan implies roughly RMB 95 billion per year — the current quarterly run-rate (RMB 67.7B) is running well ahead of that plan. Either guidance rises or spending gets throttled.
3. The financing web. Watch whether BIS-style scrutiny of circular deals turns into actual regulatory constraints in either the US or China, which would hit vendor-financed demand first.
4. Margin floor. Management has promised growth over profit, but a net margin approaching mid-single digits tests investor patience. The 5% share drop suggests that patience is already thinning.

Alibaba’s quarter is not a warning that AI demand is fake — cloud revenue is real, growing, and customer-funded. It is a warning about the financing structure underneath the boom. The industry is now spending tomorrow’s profits to buy today’s GPUs, and the ledger is starting to show it.