Jensen Huang, Nvidia's chief executive, has a unique way of evaluating engineers - one that comes with a token budget attached. As he discussed on the All-In Podcast at the close of GTC 2026, if an engineer's annual AI token consumption is less than half their salary, the decision to keep them is made in favor of keeping more talented employees.
This approach aims to strike a balance between investing in top talent and controlling expenses. Huang stated that he has a test for whether an engineer is worth keeping, which involves checking their AI token consumption against their annual salary. If it comes in under half, he will likely choose to retain them over less skilled candidates who could be more cost-effective.
The token budget method serves as a tool for Nvidia to evaluate the quality of its engineers and make informed decisions about hiring and retaining talent. By implementing this strategy, the company can identify and reward the most valuable members of its team while also maintaining control over costs.