Are AI tokens a new signing bonus or a cost of doing business?

This week, a topic that has been buzzing around Silicon Valley came to the fore: AI tokens as compensation. The idea is straightforward enough – instead of just giving engineers salaries, equity, and bonuses, companies can also give them a budget of AI tokens, the computational units of digital tools like Claude, ChatGPT, and Gemini. Spend them running agents, changing jobs, improving code. The point is that getting more computing power makes engineers more productive, and that better engineers are more valuable. It’s money in the person holding them, that’s the idea.

Jensen Huang, Nvidia’s leather-jacketed CEO, seemed to be taking everyone’s cue when he floated at the company’s annual GTC event earlier this week that engineers should get about half of their pay back — in tokens. His superiors, and his mathematics, may be burned $250,000 a year in AI computing. He called it a recruiting tool and predicted it would become a fixture in Silicon Valley.

It’s not clear where the idea started, well, imagined. Tomasz Tunguz, a well-known VC in the Bay Area who runs Theory Ventures and focuses on AI, data, and SaaS startups – and whose writing on all things has gained credibility over the years – was talking about this in mid-February, writing that tech startups were already raising funds for shows like “the fourth district to engineer compensation.” Using data from the compensation tracking website Levels.fyi, they put the highest salary for a software engineer at $375,000. Add $100,000 in tokens and you’ve topped $475,000 — meaning about one in five dollars is now accounted for.

It was not an accident. Agent AI has been on the rise, and release of OpenClaw towards the end of January he accelerated the negotiations. OpenClaw is an open-source AI assistant designed to run continuously – moving tasks, spawning sub-agents, and working on to-do lists while the user sleeps. It’s part of a broader shift toward “agent” AI, meaning systems that don’t just respond to information but act in real-time.

The result is that the use of symbols has exploded. While someone writing an article can spend 10,000 tokens in a day, an engineer running a support team can blow millions a day – automatically, in the background, without typing a word.

By the end of this week, the New York Times had produced a intelligent form on the so-called tokenmaxxing trend, finding that engineers at companies including Meta and OpenAI are competing on internal bulletin boards that target the use of tokens. Generous budgets are quietly becoming the norm, the report said, in the way that dental insurance or free lunches were. An Ericsson analyst in Stockholm told the Times that he probably spends more on Claude than he earns, though his boss picks up the tab.

Perhaps tokens will indeed be the fourth pillar of compensation. But engineers may want to hold the line before accepting this as an outright victory. More tokens may mean more power in the short term, but when we consider the situation, it doesn’t mean more security. For one thing, a large portion of the brand comes with high expectations. If a company pays for a second engineer’s estimate on your behalf, the real pressure is to produce twice as much (or more).

And there is a muddled problem under this: when the company’s mark is used for each employee reaching or exceeding the employee’s salary, the financial perspective of the accountant begins to look different from his financial group. If the computer is doing the work, the question of how many people need to coordinate becomes difficult to answer.

Jamaal Glenn, a Stanford MBA from the East Coast and former VC turned financial services CFO, is the same it shows that what may appear to be benefits can be a smart way for companies to increase the perceived value of compensation without increasing costs or expenses – things that add up to the employee over time. Your brand budget is insufficient. It doesn’t appreciate it. It doesn’t show up in your next interview the way a base salary or financial aid does. If companies successfully adapt tokens as payment, they may find it easier to maintain cash comp while pointing to growing interest rates as proof of investment in their people.

This is good for the company. Whether it’s good for engineers depends on questions that most engineers don’t have much to answer.