Jensen Huang has been vocal about his approach to hiring engineers. Speaking on the All-In Podcast at the close of GTC 2026, he outlined a test that involves calculating an engineer's worth based on their annual AI token consumption and whether it comes in under half of their salary. The idea is to determine if an engineer is worth keeping on the team.
The test itself is quite straightforward: if an engineer spends less than $500,000 on AI tokens per year compared to their annual salary, they are deemed unsuitable for keeping on the team. However, the fact that this cost includes a token budget attached makes it all the more important to keep track of expenses carefully.
Huang's approach may seem punitive, but its intention is to weed out underperforming engineers and focus on those who have the potential to drive innovation at Nvidia. By providing clear guidelines for what constitutes a "worthwhile" salary, Huang hopes to attract top talent from around the world. As the tech industry continues to grapple with issues of burnout and overpay, it will be interesting to see how this approach plays out in practice.