Eric Schmidt, who led Google for a decade, told Stanford students that, while he’s not one to give investment advice, he sees an obvious play in the stock market that doesn’t involve buying shares of his former employer.
Schmidt said, in a video that was posted by Stanford this week and subsequently removed, that big technology companies are planning for increasingly large investments into Nvidia-based artificial intelligence data centers, which could cost as much as $300 billion to build.
“I’m talking to the big companies, and the big companies are telling me they need $20 billion, $50 billion, $100 billion — very very hard,” said Schmidt, adding that he’s a “close friend” to OpenAI CEO Sam Altman.
Schmidt suggested a huge amount of this spending was going to Nvidia, which makes the dominant data center AI chips, and has already seen revenue increase by more than 200% for three straight quarters. Google has developed chips called Tensor Processing Units (TPUs), which could compete with Nvidia’s processors but are still at a much earlier stage.
“If $300 billion is all going to Nvidia, you know what to do in the stock market,” Schmidt said. “That’s not a stock recommendation.” Schmidt didn’t say if he owns Nvidia shares.
Schmidt, who was CEO of Google from 2001 to 2011 and remained on the board until 2019, was addressing a class of undergraduates on the topic of AI. He told the Wall Street Journal that he requested the video to be taken down because he had misspoken about Google’s lax work culture in the session.
But his candid remarks illustrate the driving force behind Nvidia’s rise and the company’s central place in the generative AI boom that began in late 2022.
While Nvidia is seeing soaring demand from cloud companies and the leading developers of AI models, Wall Street is asking whether the chipmaker’s top clients are overspending on AI infrastructure. Nvidia will provide its latest update on the market when it reports quarterly results on Aug. 28
Schmidt suggested Nvidia won’t be the only winner in AI, but there aren’t a lot of other obvious choices. He said that he currently believes big companies who can invest more money into Nvidia chips and data centers will gain a technological lead over smaller competitors that can’t spend as freely.
“At the moment, the gap between the frontier models — there are only three — and everyone else appears to be getting larger,” Schmidt said. “Six months ago, I was convinced that the gap was getting smaller, so I invested lots of money in the little companies. Now I’m not so sure.”
Meta CEO Mark Zuckerberg, whose company has purchased about 600,000 of Nvidia’s pricey GPUs, said earlier this month that Meta’s next-generation model — in the Llama family — will require access to about 10 times the amount of computing power.
“And future models will continue to grow beyond that,” Zuckerberg said on an earnings call.
Altman, meanwhile, is reportedly working with OpenAI’s lead backer Microsoft to build a $100 billion data center for AI called “Stargate.”
“When Microsoft did the OpenAI deal, I thought that was the stupidest idea I’d ever heard, outsourcing essentially your AI leadership to OpenAI and Sam and his team,” Schmidt said. “And yet, they’re on their way to being the most valuable company.”
Schmidt said it will be difficult for competitors to catch up with Nvidia because many of the most important open source tools that AI developers use are based on the company’s CUDA programming language. He said AMD’s software that translates Nvidia’s CUDA code for its own chips “doesn’t work yet.”
Schmidt, who started venture firm Innovation Endeavors in 2010, still holds about 147 million shares of Alphabet, according to Bloomberg, valued at about $24 billion. In addition to startup investing, he has been a philanthropist and advised several government committees on technology.