Minimizing team size while maximizing impact is a challenge that many tech companies face, and Nvidia's Jensen Huang has identified a key metric to gauge an engineer's worth. Speaking on the All-In Podcast at the close of GTC 2026, Huang revealed that if an engineer's annual AI token consumption came in under half their salary, they would be exempted from being replaced by a cheaper engineer. This test is designed to determine whether keeping an $500,000 engineer is worthwhile.
The test attaches a token budget to the engineer's salary, which is used to calculate their AI token consumption rate. If this rate falls below 50%, Huang claims that he will not replace the engineer with someone cheaper. This approach allows companies to weigh the cost of hiring and retaining an engineer against their productivity gains through increased innovation.
While the test provides a clear guideline for managers, implementing it effectively requires careful consideration. Companies need to ensure that they are transparent about the costs associated with different approaches and provide regular updates on the effectiveness of their token budget management strategy. By adopting this approach, companies can optimize their hiring decisions while minimizing the negative impact on their bottom line.