Diverting Tesla’s AI chips to X made sense given the logistics that come with assembling a supercomputer, the automaker’s former head of AI infrastructure said on Tuesday.
“The practicalities of building a supercomputer are insane once you get down to it,” Tim Zaman wrote in an X post.
“Lets assume you have a datacenter location and ordered GPUs,” he added. “Just taking delivery of thousands of GPUs is a crazy military operation, even before any racking, stacking and cabling or bringup.”
Prior to joining Tesla in 2019, Zaman had spent three years working for Nvidia as an AI infrastructure systems software engineer, per his LinkedIn profile. He left Tesla to join Google’s DeepMind as a software engineer last year.
“True,” Musk replied.
Zaman weighed in on the issue after a CNBC report on the same day said that Tesla CEO Elon Musk had rerouted $500 million worth of Nvidia chips to his social media platform.
Musk confirmed CNBC’s report on Tuesday but said the move was made because “Tesla had no place to send the Nvidia chips to turn them on.”
“They would have just sat in a warehouse,” Musk wrote in an X post.
Musk’s decision to redirect the highly coveted AI chips has highlighted his companies’ increasingly competing interests.
The mercurial billionaire’s sprawling business empire spans multiple industries, whether it be making rockets with SpaceX or drilling tunnels with The Boring Company.
In January, Musk said that he was “uncomfortable growing Tesla to be a leader in AI & robotics without having ~25% voting control.” Under Musk, the EV giant has been working hard to realize its ambitions of building self-driving cars and sentient humanoid robots.
Tesla, however, isn’t the only Musk-owned company that has set its sights on conquering the AI industry.
On May 26, Musk revealed that his AI startup, xAI, raised $6 billion for their Series B funding round. The latest fundraising round has bolstered xAI’s total valuation to $24 billion, making it the second-most valuable AI company behind Sam Altman’s OpenAI.
But money isn’t the only issue AI players like Musk and Altman have to contend with. Companies hoping to make a splash in the highly competitive field need to acquire Nvidia’s coveted AI chips, which, unfortunately, are in short supply.
This has become a problem — even for Nvidia. In February, Nvidia’s CEO Jensen Huang said in an earnings call that the company was trying its best to distribute chips fairly.
“Why allocate something when the data center’s not ready? Nothing is more difficult than to have anything sit around,” Huang said.
“At the core of it, we want to allocate fairly, avoiding waste and looking for opportunities to connect partners and end users,” Huang added.
