(Reuters) – Nvidia's new Blackwell AI chips, which have already faced delays, have encountered problems with associated servers overheating, causing some customers to worry they won't have enough time to get new data centers up and running, the information reported on Sunday .
Blackwell's graphics processing units overheat when connected together in server racks designed for up to 72 chips, the report said, citing sources familiar with the problem.
The chipmaker has asked its suppliers several times to change the design of the racks to fix overheating issues, according to Nvidia employees who worked on the problem, as well as customers and suppliers with knowledge of the problem, the report said without naming to mention. suppliers.
“Nvidia works with leading cloud service providers as an integral part of our engineering team and process. The technical iterations are normal and expected,” a company spokesperson said in a statement to Reuters.
In March, Nvidia unveiled Blackwell chips and had previously said they would ship in the second quarter before delays hit, potentially impacting customers like Meta Platforms, Alphabet's Google and Microsoft.
Nvidia's Blackwell chip takes two squares of silicon the size of the company's previous offering and ties them into a single component that's 30 times faster at tasks like providing chatbot responses.
(Reporting by Gursimran Kaur in Bengaluru; Editing by Bill Berkrot and Lisa Shumaker)
