Nvidia has allegedly informed Microsoft and at least one other cloud provider that the production of its “Blackwell” B200 AI chips will be delayed by at least three months compared to the original plan, as reported by The Information. This delay is attributed to a design flaw that was detected “unusually late in the production process,” as stated by two unnamed sources, including a Microsoft employee, cited by the outlet.

The B200 chips are the successors to the highly popular and hard-to-acquire H100 chips, which power extensive areas of the artificial intelligence cloud landscape and have significantly contributed to making Nvidia one of the most valuable companies globally. Nvidia anticipates that the production of the chip “will ramp up in the second half of the year,” as per a statement shared by Nvidia spokesperson John Rizzo with The Verge. “Beyond this, we do not comment on rumors.”

Currently, Nvidia is reportedly conducting a new set of test runs with chip producer Taiwan Semiconductor Manufacturing Company and won’t ship a large quantity of Blackwell chips until the first quarter. The Information indicates that Microsoft, Google, and Meta have placed orders worth “tens of billions of dollars” for these chips.

This report emerges just a few months after Nvidia announced that “Blackwell-based products will be accessible from partners” starting in 2024. The new chips are expected to initiate a new annual rhythm of AI chips from the company while several other tech firms, such as AMD, strive to develop their own AI chip competitors.