Tech leaders pitch AI token budgets as a new part of engineer pay

In short: Some tech leaders want to add AI token budgets to engineering pay, but the real value and fairness are still unclear.

What's going on

Some companies are starting to treat AI tokens as a fourth part of compensation, alongside salary, bonus, and equity. Tokens are the units people spend to use an AI model, like credits on a prepaid phone plan (more credits means more calls).

Nvidia CEO Jensen Huang talked about this idea at the company’s GTC 2026 event. He suggested giving engineers an annual token budget worth about half of their base salary, on top of normal pay. Huang argued that engineers with more tokens can get more work done because they can use AI more often.

OpenAI leaders have also emphasized access to tokens and “inference compute,” meaning the paid computer power used to run AI answers (like paying for electricity and time on a very fast calculator). OpenAI President Greg Brockman said that having more of this available will increasingly drive software productivity. The report says job candidates are already asking about token budgets in interviews, and some OpenAI employees reportedly use more than 10 billion tokens per week.

Investors are pushing the idea forward, too. Tomasz Tunguz of Theory Ventures estimated that if a company adds $100,000 a year in AI usage costs to a $375,000 engineer salary, then more than 20% of the total package could be AI usage. Some companies are already tracking token use per employee.

What to watch

The big open question is whether the productivity gains are real at scale. If an engineer uses $100,000 a year in AI, the company needs to see a big jump in output to justify it. Engineers may also want to watch for token budgets replacing cash pay, changing in value over time, or benefiting some roles much more than others.

Source: TechCrunch AI