(Reuters) – Nvidia’s new Blackwell AI chip is already facing delays and problems with its accompanying servers overheating, giving some customers enough time to get new data centers up and running. The information newspaper reported on Sunday that he is worried that he will not have time for the job. .
The report, citing sources familiar with the matter, said Blackwell’s graphics processing units overheated when connected to server racks designed to hold up to 72 chips.
The semiconductor maker has asked its suppliers to change rack designs several times to resolve the overheating issue, according to Nvidia employees working on the issue, as well as customers and suppliers familiar with the issue, who reported anonymously. It is stated in supplier.
“NVIDIA works with leading cloud service providers as an integral part of our engineering teams and processes. Engineering iterations are normal and expected,” a company spokesperson told Reuters. said in a statement.
Nvidia announced Blackwell chips in March and previously said they would ship in the second quarter before delays that could impact customers such as Meta Platforms, Alphabet’s Google and Microsoft. .
Nvidia’s Blackwell chip combines two squares of silicon, the same size as the company’s previous products, into one component, making tasks like providing responses from chatbots 30 times faster.
(Reporting by Gursimran Kaur in Bengaluru; Editing by Bill Berkrot and Lisa Shoemaker)
