Nvidia’s Jensen Huang sets AI benchmark: $250K token use expected from top engineers

Jensen Huang warns top engineers: insufficient use of AI tokens is a major concern.

The CEO of Nvidia mentioned on the "All-In Podcast," released Thursday, that he would be "seriously concerned" if one of the leading engineers used too few AI tokens.

"If that $500,000 engineer didn't utilise at least $250,000 worth of tokens, I would be very concerned," Huang stated.

"At the year's end, I’ll ask the $500,000 engineer how much they spent on tokens. If they say $5,000, I'd focus on something else," he added.

Huang, when questioned about Nvidia possibly investing $2 billion on tokens for its engineering team, said: "That's what we're aiming for."

"It's similar to a chip designer saying, 'Guess what? I'll just use a pen and paper,'" Huang commented regarding top engineers not effectively using AI tokens.

Earlier this week, at the GPU Technology Conference, Huang remarked that tokens might be part of his engineer recruitment approach.

"Engineers are set to make a few hundred thousand dollars annually in base pay," Huang stated. "I'll likely provide half as much in tokens to enhance their productivity by 10 times."

"Now, it's among Silicon Valley's recruiting resources: How many tokens are included in my job offer?" Huang noted. "The logic here is straightforward, as every engineer with token access becomes more efficient."

Tokens are the base unit AI systems use for processing text. The more text an AI reads or generates, the higher the token consumption, hence companies often charge based on usage per thousand or million tokens.

Huang isn't the only one advocating that engineers need ample AI computing power — and that companies should be prepared to cover the costs.

Business Insider's Alistair Barr pointed out earlier this month that tech firms might be exploring a novel method to attract talent: providing AI processing power alongside salary, bonuses, and stock options.

Tomasz Tunguz from Theory Ventures informed Barr that tokens might be an emerging "fourth element" of compensation.

Peter Gostev, from Arena, a startup evaluating model performance, proposed that OpenAI and Anthropic "develop recruitment platforms where clients can post job roles, indicating the token budget alongside salary expectations."

Thibault Sottiaux, heading engineering on OpenAI's Codex squad, mentioned on X that applicants are frequently asking about their compute access levels.

OpenAI CEO Sam Altman has previously suggested that AI tokens could be a form of universal basic income.