Design flaw massively delays launch of new AI chip Blackwell B200

At Nvidia, the production of its announced AI chip “Blackwell” B200 is said to be significantly delayed due to a design flaw. According to reports, Nvidia has now informed Microsoft and at least one other cloud provider that there are problems. BoliviaInteligente / Unsplash

Long delay

This is evident from a Insider report from The Information Production of the B200 chip is said to take at least three months longer than initially planned. Nvidia is now reportedly working with chipmaker Taiwan Semiconductor Manufacturing Company (TSMC) on a new series of test runs and will not ship large numbers of Blackwell chips before the first quarter of 2025. The Information writes that Microsoft, Google and Meta have ordered the chips worth “several billion dollars”.

The delay is said to have been caused by a design flaw that was discovered “unusually late in the production process,” according to two unnamed sources, including a Microsoft employee, as The Information writes. Other online magazines are also reporting on the delays. Nvidia expects production of the B200 chip to “start in the second half of the year,” Nvidia spokesman John Rizzo told The Verge. “Beyond that, we don’t comment on rumors.” There is currently no confirmation of the details of the report – but there is also no denial. So at the moment these are rumors that cannot be verified.

The B200 chips are the successor to the extremely popular H100 chips that power large parts of the artificial intelligence cloud landscape and have made Nvidia one of the most valuable companies in the world. The report of the problems comes just months after Nvidia announced that Blackwell-based products would be available from partners starting in 2024.

The new chips are expected to kick off a new annual cadence of AI chips from the company, while several other technology companies like AMD are working to develop their own AI chip competitors. Nvidia had only recently confirmed that there had been quality issues with a GDDR6X shipment. This is now causing production bottlenecks for Nvidia RTX 40 cards from the 4070 onwards. Since Micron is the only supplier, the situation is getting worse.