AI JOBS
Do you take tokens?
AI tokens are starting to look like a new part of pay in Silicon Valley.
Instead of offering engineers just salary, equity, and bonuses, some companies are now discussing giving them AI tokens too.
These tokens power tools like ChatGPT, Claude, and Gemini, and can be used to run agents, automate work, and speed up coding.
The idea is simple: more compute could mean more output.
Nvidia CEO Jensen Huang helped push the idea into the spotlight this week.
He suggested engineers could get token budgets worth up to half their base salary on top of normal pay.
In some cases, that could mean $250,000 a year in AI compute.
The idea has been building for a while.
In February, investor Tomasz Tunguz said some startups were already treating AI costs as a fourth part of engineering compensation.
As AI agents become more common, token use is climbing fast.
While a casual user may spend thousands of tokens, engineers using multiple agents can burn through millions in a day.
The New York Times also reported that engineers at companies like Meta and OpenAI are comparing token usage on internal leaderboards.
In some firms, token budgets are starting to feel like a job perk.
In brief:
Some tech companies are starting to treat AI tokens as part of compensation.
As AI agents grow, token spending is rising fast.
Tokens may boost output, but they do not offer the long-term value of salary or equity.
But there’s a catch
More tokens may help engineers work faster, but they can also bring more pressure to deliver.
And if a company is spending as much on AI compute as it is on salary, it may start questioning how many people it really needs.
There’s also the issue of value.
Tokens are useful, but they are not cash or equity.
They do not grow, they do not vest, and they are unlikely to help in future salary talks.
So while tokens may become a bigger part of pay in tech, it is still unclear how much they really benefit employees.
Sorry, whenever I hear them earning tokens all I can think of is them earning XP - MG


