Alibaba’s Cloud Intelligence Group used its Q4 fiscal 2026 earnings call to disclose ¥36 billion in annualized AI product revenue — apparently the first recurring-revenue AI line item any Chinese hyperscaler has published in a form investors can compare with Microsoft and Google. The number’s size is unremarkable. What matters is that it exists. Until last week, the Chinese cloud story was told in capex pledges and YoY growth rates, neither of which gives an analyst a comparable figure to set against Redmond’s. Now there is a denominator, and what the market does with it will shape the next year of AI-cloud disclosure on both sides of the Pacific.

The figure is constructed the way Microsoft’s is: AI product revenue inside Cloud Intelligence Group for the March quarter, multiplied by four, gives ¥36 billion. This is annualized in-quarter revenue, not contracted backlog. The aggregate grew at triple-digit rates for the eleventh consecutive quarter, and Alibaba has chosen to be measured against the same frame Microsoft built.

The ¥36 billion covers the total AI product line. CEO Wu guided separately that the narrower model-and-application-services subset would surpass ¥30 billion in full-year ARR by December — a smaller denominator, a different slice of the same business. The year-end target sits below the current total-line figure not because the business is contracting, but because the two numbers are measuring different things.

What Microsoft, Google, and Amazon report differs in ways the new Alibaba number throws into relief. Microsoft’s $37 billion AI business run rate bundles OpenAI’s Azure consumption with first-party Copilot revenue. Alphabet reports total Google Cloud as a single line with a contracted backlog of $460 billion, but has not disaggregated the AI portion. Amazon has not published an equivalent figure. Alibaba has split the difference: an AI-specific line, built by the Microsoft method, sitting one rung below the bundled cloud number that always appears in the press release. The effect is to put one figure on the wall every other hyperscaler can be benchmarked against, and to apply pressure to the three that have not.

The strongest objection is that an ARR built from a single quarter is not really an ARR. A figure that annualizes one quarter benefiting from enterprise migrations, captive consumption by Alibaba’s own commerce business, or both, would inflate the headline and quietly hollow out the comparison the disclosure was meant to invite. The honest reply is that the same critique applies to Microsoft, whose run rate captures OpenAI’s compute spend on Azure inside the very business line being measured, and to anyone using ARR outside contracted SaaS. Both numbers are constructions. What they offer the market is not an audited figure but a published one — a metric a CFO has put a name to and will be asked to defend at next quarter’s call. That is enough to make them comparable, and comparable is what global investors did not have before.

A figure that did not exist as a Chinese-cloud comparable last quarter is now the figure Tencent, Huawei, and Baidu will be asked either to match or to explain not matching.

Retailers who buy Chinese cloud capacity now have a number to benchmark their spend against. The ¥36 billion covers AI product revenue from Alibaba’s commerce platforms and Cloud Intelligence Group’s external customers, a base that includes Qwen inference and the Taobao AI assistant. If a retailer’s share of that line is rising, the cloud bill is no longer a black box but a tracked component of the disclosure their vendor reads on its earnings call. If their share is falling while their bill is rising, the questions a retail CFO will face about why they are paying for AI compute that is not showing up in commerce growth will get sharper. The pressure runs both ways. What Alibaba has given retail buyers, and given itself, is the same metric to argue with.

The choice now sits with the Chinese cloud businesses that have not disclosed. Tencent’s business services revenue grew in the teens in the third quarter of 2025, with AI repeatedly cited as a driver but never quantified separately. The reluctance is a strategic choice, not an accounting one. Huawei’s external cloud revenue fell 3.5% in 2025 and the annual report breaks out nothing on AI; Baidu has not published an AI revenue line. Each can publish a comparable number and risk the comparison, or decline and let the comparison proceed on Alibaba’s terms. Alibaba has decided the second option is the worse one.