Nvidia’s Blackwell AI Chips Hit With Overheating Issues: Report

Nvidia’s highly anticipated Blackwell AI chips, designed to hugely enhance performance for AI tasks, have encountered a series of setbacks, including overheating issues with the servers intended to house them. 

The graphics processing units (GPUs), which were expected to ship in the second quarter of this year, are causing concern among key customers, including Meta Platforms, Google, and Microsoft. The chips are overheating when installed together in server racks designed to hold up to 72 units, according to a report by The Information.

According to the report mentioned that Nvidia has already requested its suppliers to revise the server rack designs several times in an effort to address the heating problems. The overheating issue is exacerbating the delays that have already affected the Blackwell chip’s rollout, leaving cloud service providers with concerns about the impact on their ability to deploy new data centers in time. 

Nvidia, however, remains optimistic, with a company spokesperson telling Reuters that such engineering iterations are standard and expected in the development process. The company continues to work closely with cloud providers as part of their engineering efforts to resolve the issue.

The Blackwell chips, first unveiled in March, are expected to deliver a massive leap in performance. Nvidia claims the chips are 30 times faster than their predecessors, making them ideal for AI-powered applications like chatbots. 

(Inputs from The Information and Reuters)