Microsoft finally puts a number on Azure: $29.4 billion a quarter

Two numbers landed this morning that say more about where AI money is actually moving than any benchmark table: Microsoft broke out what Azure earns, and Alibaba's Tmall started selling model tokens next to the phone credit.

Microsoft disclosed quarterly Azure revenue for the first time ever on Wednesday — $29.4 billion in the June quarter, $101.9 billion for the fiscal year ended June 30. The figure came in a presentation that also collapses Microsoft's three reporting segments into two: "Agents and Infra," which holds Azure, Microsoft 365 cloud, server products and the enterprise services Microsoft now calls frontier and support services, and "Devices and Consumer," which takes Windows, Xbox, and search and ads from Bing and LinkedIn. The old structure had stood since 2015, and the new one starts with fiscal first-quarter 2027 earnings in late October. Investors liked it: shares rose about 1.4% in after-hours trading.

The context is the part worth dwelling on. Amazon began disclosing AWS revenue in 2015 and Alphabet started reporting Google Cloud in 2020, which left Microsoft — the number two cloud by revenue — as the only one of the three publishing a growth rate rather than a dollar figure. That was convenient when the growth rate was the story; it stops being convenient when analysts need to size the business against rivals. The new number puts Azure behind AWS at $42.2 billion in its most recent quarter and ahead of Google Cloud at $24.8 billion. It grew 42% year over year, and now represents almost a third of Microsoft's total revenue. Management guided to 44% to 45% constant-currency growth for the September quarter, slightly below July's 45% call, but said the difference is bookkeeping: some GitHub sales move out of Azure and into M365 Cloud, where most of the advanced AI products now sit. Overall company guidance is unchanged.

The redefinition is the real signal. Azure sheds GitHub cloud services, developer cloud services, Security Copilot and the healthcare and life sciences cloud products, leaving what CEO Satya Nadella called "more purely our consumption-based platform and infrastructure business." Read that as a bet that the money in AI is earned two ways — selling the compute, and selling the agent on top of it — and that Microsoft wants investors to see both lines clearly rather than blended into one cloud number. It also sharpens an uncomfortable concentration: Stifel estimated in July that roughly half of Azure's fiscal 2026 revenue growth came from OpenAI, which now spreads its training across Amazon and others. Publishing the number makes that dependency easier to model and harder to wave away.

Tmall opened a token recharge center on Wednesday, putting subscriptions from Alibaba Cloud, Zhipu, Kimi and MiniMax on the same shelf as mobile phone top-ups. Users in the Taobao app can search "token" and buy Token Plan, Coding Plan or pay-as-you-go bundles delivered as card codes or direct credit. Zhipu opened its official flagship store the day before, and AIbase reported search volume for its products rose about 40 times on opening day. The platform is pairing it with hardware — computers preloaded with agent software, Qwen-powered learning machines, smart speakers and AI glasses.

This is the consumerization of the API business, and it is more consequential than another storefront. Chinese model labs have spent two years selling to developers and enterprises; Tmall converts a metered, technical purchase into a standardized, auto-renewing consumer product with a checkout flow hundreds of millions of people already know how to use. We covered the supply side of this in Zhipu's API revenue jumps 27x as China's AI labs prove the token economy — the labs proved tokens sell to businesses. Now they're testing whether tokens sell to everyone, with Alibaba collecting the toll either way.

What to watch: whether Microsoft's October print makes the OpenAI share of Azure growth look like a strength or a concentration risk — and whether Tmall's token sales hold once the novelty of a flagship store wears off.

Which is the better read on where AI revenue actually lands this year — infrastructure billed by the minute, or subscriptions sold like phone credit? Tell us in the comments.

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Read more

AI 101 — How do AI video generators work?

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone