Tesla CEO Elon Musk Reiterates Bringing Up Large AI Training Cluster Currently 'Extremely Difficult'

Tesla Inc TSLA CEO Elon Musk on Wednesday shed light on the limitations of AI training compute at the company.

What Happened: “Bringing up a large training cluster of H100's is currently extremely difficult,” Musk wrote on X, formerly Twitter. “Same was true of A100's when they first came out, but now they run smoothly.”

Tesla activated Nvidia Corp‘s NVDA 10,000-unit H100 GPU cluster in late August, which Musk said is three times faster than the A100 in the company's tests. However, the CEO has previously said that the uptime and performance with A100 cluster was low at first but then improved with ‘lots of work’ from both Tesla and Nvidia.

Why It Matters: Tesla will use both NVIDIA and Dojo going forth to process the large amount of data from Tesla’s growing fleet of vehicles and train its FSD software.

Dojo is Tesla's custom-built supercomputer platform, specifically designed from scratch for AI machine learning purposes, with a particular focus on video training using data obtained from its fleet of vehicles.

Check out more of Benzinga’s Future Of Mobility coverage by following this link.

Related Link: Elon Musk Expresses Frustration At Fish And Wildlife Agency Induced Delay In Starship’s Second Flight