Beijing tells Alibaba and ByteDance to file Nvidia purchase plans

Share
Beijing tells Alibaba and ByteDance to file Nvidia purchase plans

For a year, China's approach to Nvidia has been subtraction — stop ordering, buy domestic. Sunday's report is Beijing trying addition instead, on a chip that was never on the banned list.

China's Ministry of Industry and Information Technology has asked Alibaba, ByteDance and other domestic companies to report their plans to buy Nvidia's RTX Pro 5500, and told them the government intends to approve the purchases. The Information's Qianer Liu reported the approach on Sunday, citing two people familiar with the matter. Reuters and Bloomberg both carried the story, with Reuters spelling out the ministry's wording — that the government "intends to approve" — and noting it could not independently verify the report. Nvidia, Alibaba and ByteDance have not commented.

The chip matters as much as the permission. The RTX Pro 5500 is a Blackwell workstation card with 84 GB of GDDR7 memory and 21,760 CUDA cores — the core count of a consumer RTX 5090, with well over twice the memory — sold as a professional graphics and simulation part rather than a data-center AI accelerator. That classification is the point. U.S. export rules keep Nvidia's top data-center parts out of China; they say much less about a workstation board that ships by the pallet. At 84 GB, one card holds models that would otherwise need a restricted server, and a rack of them is inference capacity that arrives through the front door.

The timing reverses a posture Beijing spent the summer building. From August, regulators were reported to have told domestic firms to stop placing new Nvidia orders and to steer toward domestic alternatives; the exception was a small number of import licences for H200s to ByteDance and Tencent — Beijing lets ByteDance and Tencent import Nvidia H200 chips. Approving RTX Pro 5500s points the same way: Chinese labs need compute now, domestic supply cannot yet close the gap, and the state would rather see the orders than lose sight of them.

Read the verb carefully, though. The instruction is to report plans, not to proceed. Beijing is not lifting a restriction; it is putting itself in the middle of the purchase so it can decide case by case. That is the instrument it has used on rare earths and on the H200 licences — control exercised through the queue rather than the statute book — and it gives the ministry something a ban never did: visibility into which lab is scaling what, and a lever over who is served first. The reported word "intends" is doing deliberate work, and it is not a promise.

What the report does not say: how many chips, at what price, on what timeline, or whether approval reaches companies beyond the ones named. This is a single-outlet scoop resting on two unnamed sources. The scale is unknown, and the arrangement can still narrow.

What to watch: whether a volume figure or official confirmation follows in the next few days, and whether Washington reads an 84 GB workstation card as a loophole in the export regime rather than a product category of its own.

If a workstation GPU can substitute for a data-center accelerator, is export control still controlling anything? Tell us in the comments.

Read more

Akhetonics says its all-optical CPU reaches a customer in 2026

Akhetonics says its all-optical CPU reaches a customer in 2026

Light-based computing keeps promising more than it delivers — but one Munich startup has just put a date on its bet, and the interview laying it out is doing the rounds on Hacker News this week. Akhetonics says it will deploy its first commercial machine with a major customer by the end of 2026, with several more planned for 2027. The company, founded by Michael Kissner and Leonardo Del Bino, is building a computer where data enters as light, is switched as light, and circulates through memory

The Week in AI — October 5–11, 2026

The Week in AI — October 5–11, 2026

Every big claim this week turned out to rest on fine print more interesting than the headline: revenue only the company reporting it can define, safety tests sandboxed while the product keeps the web, and a Pentagon phase-out nobody would confirm until reporters kept asking. The week's top 5 1. OpenAI's revenue was $20 billion below the numbers everyone quoted — and the gap was definitional. The Financial Times reported Thursday that OpenAI's annualized revenue runs roughly $20 billion unde

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Russia's largest tech company is learning what the AI era's infrastructure war looks like from the receiving end — three data centers in four days, and with them much of the cloud layer Russian businesses run on. A Ukrainian drone strike knocked out Yandex's data center in Vladimir early Sunday morning, the third of the company's facilities hit since October 8. The site — reported at roughly 50 MW and designed for about 2,880 server racks — stopped operating completely after the attack, Yandex

Agent teams cost up to 5x more, barely score higher

Agent teams cost up to 5x more, barely score higher

The multi-agent hype train hit a benchmark this weekend — and the grid and the trucking regulators had quiet weeks of their own. Vals AI put agent teams head-to-head with single agents on its Vibe Code Bench, and the teams cost between 1.8 and 5.1 times more for almost no extra quality. The evals company ran GPT-6 Sol and Claude Opus 5.5 solo and in teams across 50 apps at two reasoning efforts; out of four comparisons, only one was statistically significant — Sol at medium effort, where the t