Nvidia CEO Jensen Huang says engineers should be evaluated on basis of how many AI tokens they use

Nvidia CEO Jensen Huang says the rise of AI is changing software engineering so much that companies may soon judge engineers by how many AI tokens they use instead of just how much code they write. He believes access to AI compute will become a key productivity tool, and engineers who use more AI effectively will deliver better results.

In Short

• Nividia CEO says companies should use AI tokens as a key part of engineer compensation
• Jensen Huang suggests that AI tokens could become a standard for evaluating high-value tech roles
• He says token budgets will also attract top talent in competitive tech hiring

If any profession is being reshaped the most by the rise of artificial intelligence, it’s software engineering. Over the past few years, AI tools have completely changed how engineers write, test, and execute code, and the transformation is far from over. As AI becomes a core part of everyday development, Nvidia CEO Jensen Huang says the way engineers are evaluated may need to change too. Speaking on the All-In Podcast, Huang suggested that in the AI era, engineers should be judged based on how many AI tokens they use, and the compute usage should become a new measure of productivity.

According to Huang, with AI becoming an integrated part of the work software engineers do, token budgets, which are the units of AI compute, could also become a central part of how companies measure productivity, hire talent, and even decide compensation. According to him, as AI becomes deeply embedded in software development, access to computers will matter as much as salary or experience.

What are AI tokens?

AI tokens are the basic units used by AI systems whenever they process text, write code, analyse data, or perform automated tasks. Every prompt you give an AI and every response it generates uses tokens. It basically represents the amount of computing power needed by AI to complete that task.

Since companies have to pay for this computing, the number of tokens used shows how much an engineer is using AI. According to Huang, companies are taking this usage as a new way to measure the productivity of the engineers in AI-heavy roles.

For reference, for the flagship Claude 4.6 Sonnet model, Anthropic’s pricing is around $0.000003 per input token and $0.000015 per output token, which works out to a few dollars per million tokens. Similar pricing models are used by AI systems such as Google Gemini and OpenAI’s ChatGPT.

Tokens should be seen as productivity

According to Huang, in the future engineers may not just be given salaries, but also large token budgets so they can use AI freely to speed up their work. He argues that expensive talent should be using powerful tools extensively, otherwise companies are not getting full value from them.

“Let’s say you have a software engineer or AI researcher and you pay them $500,000 a year if that person did not consume at least $250,000 worth of tokens, I am going to be deeply alarmed.”

Nvidia chief Jensen Huang believes that the token usage could become a standard benchmark for high-value technical roles, especially as companies shift towards AI-assisted development. He argues that the goal is not to spend more money on compute, but to make sure engineers fully use the technology available to them. Huang also suggests that moving forward, companies that provide more compute will attract better talent, because access to AI tools directly affects how productive an engineer can be.

In short, Huang believes the rise of AI and token-based compute will redefine what it means to be a strong engineer. Instead of measuring productivity only by how much code someone writes, companies will increasingly judge engineers by how effectively they use AI to multiply their output.

- Ends