Nvidia Is ‘Slowly Becoming the IBM of the AI Era,’ According to the Leader of a $2 Billion AI Startup

Jim Keller, an engineer who has worked at AMD, Apple, and Tesla and is now CEO of an AI chip startup taking on Nvidia, says that Nvidia is “becoming the IBM of the AI era.”

On a Sunday podcast episode of DemystifySci, Keller brought up Nvidia’s AI chips and called attention to companies like Microsoft and Google that are using Nvidia’s technology to power their own innovations.

“All the big tech companies are in an arms race and they’re all calling Nvidia,” Keller said.

Keller, who now leads the $2 billion AI chip startup Tenstorrent, which has funding from Samsung and Hyundai, stated that Nvidia currently has “the best processors by functionality.”

He then said that Nvidia is “slowly becoming the IBM of the AI era,” adding, “We’ll see how that goes. I run an AI tech company so I have opinions about that too.”

^{Jim Keller, chief executive officer of Tenstorrent. Photographer: SeongJoon Cho/Bloomberg via Getty Images}

Nvidia is now the industry leader for AI chips, with over 80% of the market share. It benefits from a first-mover advantage in AI computing; Nvidia claims to have started investing in AI and machine learning development starting in 2006.

IBM, too, could be considered a first mover in the PC market. Though IBM did not invent the PC, the company’s 1981 personal desktop opened computers up to a broader audience and generated $1 billion in revenue in its first year.

IBM “set a technology standard” with its first PC, according to IEEE Spectrum.

Keller has previously weighed in on the cost of AI chips, both from Nvidia and from ChatGPT-maker OpenAI, which currently uses Nvidia’s chips. He claimed in April that Nvidia could have cut research and development costs and in February that he could build AI processors for all workloads and AI companies at one-eighth of the cost suggested by OpenAI CEO Sam Altman.

Nvidia CEO Jensen Huang said in March that its next-generation AI chip would cost more than $30,000.

Source