AI tokens as compensation: Jensen Huang wants to give engineers $250,000 in compute

A new trend is turning Silicon Valley compensation upside down. The idea is simple: in addition to salary, equity, and bonuses, employees would also receive a budget of AI tokens, the computational units that power tools like Claude, ChatGPT, and Gemini.

Nvidia CEO Jensen Huang ignited the debate at the company's GTC conference this week, suggesting engineers should receive tokens worth roughly half their base salary. Top engineers at Nvidia could, by his math, burn through $250,000 in AI compute over the course of a year. He calls it a recruiting tool and predicts it will become standard across Silicon Valley.

The logic is that access to more compute makes engineers more productive, and more productive engineers are worth more. Venture investor Tomasz Tunguz of Theory Ventures observed as early as February that tech startups were already adding inference costs as a fourth component to engineering compensation. With a top-quartile engineer at $375,000 plus $100,000 in tokens, roughly one dollar in five is now compute.

This is no coincidence. Agentic AI has been taking off, and token consumption has exploded. Where someone writing an essay might use 10,000 tokens in an afternoon, an engineer running a swarm of agents can blow through millions of tokens in a day, automatically, in the background, without typing a word.

The New York Times reported Saturday that engineers at companies including Meta and OpenAI are competing on internal leaderboards tracking token consumption. Generous token budgets are quietly becoming a standard job perk, akin to dental insurance or free lunch.

For CIOs and technology leaders, this signals that AI infrastructure is no longer just an operating cost, but a strategic competitive advantage in the talent war. The question is no longer whether companies should provide AI access to their developers, but how much and in what form.