To obtain tech updates in your inbox, signal as much as the e-newsletter ARPU.
Nvidia’s extremely anticipated Blackwell AI chips are reportedly encountering overheating issues inside server racks, doubtlessly delaying the deployment of latest knowledge facilities for some clients. Reuters, citing The Info, reported that this subject provides to earlier delays skilled by the chipmaker.
The stories, citing sources acquainted with the matter, state that the Blackwell GPUs overheat when a number of chips are linked inside server racks designed to carry as much as 72 chips. Nvidia has reportedly requested design adjustments from its suppliers a number of occasions to handle the overheating downside.
“Nvidia is working with main cloud service suppliers as an integral a part of our engineering group and course of. The engineering iterations are regular and anticipated,” an Nvidia spokesperson instructed Reuters in an announcement.
The Blackwell chips, unveiled in March, have been initially slated to ship within the second quarter of 2024 however skilled delays. These delays may affect main clients akin to Meta Platforms, Google, and Microsoft, all of whom are important traders in AI infrastructure.
Nvidia’s Blackwell chip boasts a major efficiency enchancment over its predecessor, using two silicon squares to realize a 30-fold pace enhance in duties like chatbot responses. Nonetheless, the overheating subject presents a serious problem that would affect the rollout and adoption of this important new know-how.
The views and opinions expressed herein are the views and opinions of the writer and don’t essentially mirror these of Nasdaq, Inc.
