In the highly mature compensation market of Silicon Valley, a new type of "benefit" is quietly changing the rules of talent competition. This week, the topic of "AI Tokens (computing units) as compensation" has sparked heated discussions in the tech circle. In addition to traditional base salaries, stock options, and bonuses, top tech companies are now offering engineers substantial AI computing budgets, viewing it as an essential resource for career development.
Recently, NVIDIA CEO Jensen Huang gave this trend a boost at the GTC conference. He proposed that companies should provide engineers with AI computing quotas equivalent to 50% of their base salary. According to his calculation, top talent may consume up to $250,000 worth of computing power annually. He predicts that this "computing power benefit" will soon become a standard in Silicon Valley hiring.
The Productivity "Arms Race"
This trend did not emerge by accident. With the release of the open-source AI assistant OpenClaw in early 2026, the way engineers work has undergone a fundamental change. These AI agents, which can run 24/7, break down subtasks autonomously, and consume massive computing power. While an ordinary person might use only 10,000 tokens to write an article, an engineer running an agent cluster can burn millions of tokens in a single day.
Currently, internal "computing power consumption rankings" have even appeared within companies such as Meta and OpenAI. For developers, having sufficient tokens means being able to hire more "AI employees" to work for them, achieving a significant advantage in output.
A Benefit or an "Invisible Trap"?
However, financial experts and some experienced professionals have expressed caution. Jamal Glenn, a Stanford MBA and senior CFO, pointed out that while token budgets are convenient, they differ fundamentally from cash or stocks: they do not appreciate over time, cannot be redeemed for the next job, and will not enter your retirement account.
A deeper crisis lies in the fact that when the cost of computing power paid by a company approaches or even exceeds its salary, the finance department will inevitably reassess the value of "people." If computing power takes on most of the work, will the demand for high-salary human employees decrease accordingly?
The emergence of this new compensation structure reflects the reshaping of labor relations in the AI era. Whether tokens are a "digital bonus" that empowers engineers or an "expensive consumable" used by companies to avoid cash expenditures will become a key point in workplace battles in the coming years.
