A new problem with Nvidia AI chips is dealing with overheated servers, The Information reports

(Reuters) – Nvidia’s new Blackwell AI chips, already facing delays, have run into problems with accompanying servers overheating, prompting some customers to worry they won’t have enough time to update data centers, a report said on Sunday. .

Blackwell’s image processing units overheat when connected together in custom server racks the company built, the report said, citing sources familiar with the matter.

The AI ​​chip maker has asked its suppliers to change the design of the racks several times to solve overheating problems, according to Nvidia employees who have been working on the issue, as well as customers and suppliers with knowledge of it, the report said without naming the company. suppliers.

Nvidia did not immediately respond to a request for comment outside regular business hours.

Nvidia introduced the Blackwell chips in March and had previously said they would ship in the second quarter before encountering delays, which could affect customers such as Meta Platform, Alphabet’s Google and Microsoft.

Nvidia’s Blackwell chip takes two squares of silicon the size of the company’s previous offering and commits to a single chip that’s 30 times faster for tasks like providing responses from chatbots.

(Reporting by Gursimran Kaur in Bengaluru; Editing by Bill Berkrot)