Jensen Huang: AI Tokens Should Be Included in Engineers' Compensation Structure, Have Become New Hiring Condition in Silicon Valley

Silicon Valley is seeing a brand-new element in salary negotiations. Fortune reports that NVIDIA CEO Jensen Huang, during his keynote at GTC 2026, proposed that AI Tokens (the basic billing unit for processing text with large language models) should be included in engineers’ compensation packages. “How many Tokens are in your offer?” has become a new factor in Silicon Valley recruiting negotiations.
(Background: Full speech of Jensen Huang at GTC 2026: AI demand reaches trillions of dollars, computing power jumps 350 times, OpenClaw turns every company into AaaS)
(Additional context: The hottest jobs in the AI era: What is prompt engineering? The highest annual salary starts at 7.5 million)

What is your annual salary?

Now, Silicon Valley engineers are starting to ask one more question: what is your Token quota?

Jensen Huang said at GTC 2026 that a tech engineer with a $300,000 annual offer, if also given a Token quota worth half of the salary, could boost productivity tenfold. This is not a metaphor; it’s the actual evolving logic of compensation design.

“How many Tokens are attached to the position” has become a concrete negotiation point in Silicon Valley headhunting and offer discussions, just like stock options, and is now included in contracts.

A Token is roughly equivalent to 0.75 English words, and now it’s being written into paychecks. (But how do you tax it?)

Sam Altman takes it further: everyone gets a share of computing power

Back in 2024, Sam Altman introduced the concept of “Universal Basic Compute” on the All-In podcast. He envisioned that in the future, every “citizen of Earth” could receive a GPT-7 compute quota, which they could use themselves, resell on the market, or donate to charity.

It sounds like science fiction, but the logic isn’t complicated: if intelligence ultimately flows like electricity, then “ownership of compute power” becomes a distributable resource.

This month, at the BlackRock Infrastructure Summit, Altman re-explained this logic, stating that the core business of AI model providers is “selling Tokens,” just like power companies sell kilowatt-hours or water companies sell liters—metered and billed by usage, with no exceptions.

From this perspective, Huang’s idea of including Tokens in salary statements isn’t a strange metaphor; it’s essentially a subsidy for employees’ electricity costs.