Jensen Huang, Nvidia's CEO, recently shared a test for whether an engineer is worth keeping, and it comes with a significant financial incentive attached. The test involves evaluating an engineer's annual salary in relation to their AI token consumption. If the engineer's token consumption ($500,000) falls below half of their annual salary, they are deemed disposable. This means that Nvidia would be willing to let them go if it means saving costs and maintaining a leaner workforce.
Huang made this statement during an appearance on the All-In Podcast at the close of GTC 2026. He emphasized that the decision is not based solely on individual performance but also takes into account the overall impact of each engineer's salary on Nvidia's bottom line. The company uses AI token consumption as a proxy to gauge employee productivity, as it has been shown to be an effective way to measure and manage workforces.
The test is designed to help Nvidia identify potential areas where costs can be reduced without sacrificing talent or performance. By attaching a financial reward to the test, Huang hopes to encourage engineers to optimize their AI token consumption while maintaining their job security. This approach could potentially lead to more efficient use of resources and a leaner workforce, which would have significant implications for companies like Nvidia that are struggling with budget constraints.