Alibaba profit drops 75% as AI spending hits $10bn in one quarter

Alibaba reported a 75% drop in net profit to $1.54 billion for the June quarter, driven by capital spending of $9.98 billion—a 75% increase year-over-year—as the company accelerates its AI infrastructure buildout. The spending turned free cash flow negative at $6.58 billion, compared to an outflow of $2.77 billion in the same quarter last year.

Revenue rose 9% to $39.64 billion, slightly ahead of analyst expectations, according to details first reported by AI Watch and CNBC. Alibaba's US shares fell approximately 5% following the earnings announcement.

Cloud division delivers growth and profit

The AI Cloud and Compute Services division generated $7.14 billion in revenue, up 45% year-over-year—the fastest growth in 22 quarters. AI-related product revenue within the segment reached $1.82 billion and has grown in triple digits for twelve consecutive quarters.

The cloud unit also achieved profitability with adjusted EBITA of $830 million, a 133% increase. Chief financial officer Toby Xu reported the cloud segment's EBITA margin at 12%. Alibaba Cloud held a 38.1% share of China's AI cloud market, according to research firm Omdia cited by the company.

Alibaba restructured its reporting segments this quarter, moving chip arm T-Head into the cloud business and creating a new unit for model labs and consumer AI applications. Year-over-year comparisons have been adjusted to reflect the new structure.

AI Labs unit burns cash

The newly formed AI Labs and Applications segment, which includes the Qwen model labs, consumer app, and QwenWork agent, generated $492 million in revenue but posted an adjusted EBITA loss of $2.04 billion, compared to a $475 million loss a year earlier. The company attributed the widening loss to AI investment and the cost of running inference for the Qwen app.

Chief executive Eddie Wu defended the spending on the analyst call, stating that "our AI business's capacity to self-fund and sustain itself is strengthening." He projected annualized revenue from AI products would approach $10 billion in the current quarter, up from approximately $7.3 billion in the June quarter. Executives said the company expects to recoup its overall AI investment within the three-year window it set in early 2025, when it pledged more than RMB 380 billion.

Bloomberg Intelligence analysts took a more skeptical view, writing that AI will "continue to depress, not enhance, returns" at China's leading AI companies, with Alibaba's AI business expected to generate cash losses for the next three years.

Capital spending drivers

Alibaba cited three factors behind the capital expenditure surge: fluctuating procurement cycles, adding CPU-compute capacity in anticipation of customers adopting AI agents, and rising prices across a broad range of chip components—the same supply constraint that affected Chinese hardware companies this quarter.

The company's traditional e-commerce business continues to shrink, with China e-commerce revenue falling 8% to $16.35 billion. Alibaba has been divesting assets to fund its AI pivot, including selling its games division for $1.5 billion this month.

Why it matters

Alibaba's financial results illustrate the tension facing major tech companies pursuing AI leadership: massive infrastructure spending depresses near-term profitability even as cloud services begin generating meaningful returns. The company is executing both sides of the AI business model simultaneously—building the infrastructure and monetizing it through cloud services—while absorbing losses on consumer-facing AI products. The question for investors is whether Alibaba can reach profitability on its AI investments within its three-year timeline, or whether the capital requirements will continue to outpace revenue growth.

The company also recorded a provision for a €550 million European Commission fine against AliExpress under the Digital Services Act, which contributed to the jump in general and administrative expenses to 4.7% of revenue from 3.0%.

In August, Alibaba released the weights of Qwen3.8-Max, its largest model at 2.4 trillion parameters, marking the first time the company has open-sourced its top-tier model class. The company's target is to quintuple cloud and AI revenue to $100 billion within five years.

Details were first reported by AI Watch, with additional coverage from CNBC, Bloomberg, the South China Morning Post, and Nikkei.