Unlocking cost savings through AI token budgets has become a widely discussed topic within the tech industry. Jensen Huang, CEO of Nvidia, recently addressed this issue on his podcast at the close of GTC 2026, revealing that he has a test for whether an engineer is worth keeping - and it comes with a token budget attached.
According to Huang, if an $500,000 AI engineer's annual token consumption comes in under half their salary, they are deemed "worth keeping". This means that the company will not terminate them without considering the potential cost savings. The figure of 50% is a key benchmark, as Huang has stated he is willing to keep engineers who meet this threshold.
The concept of AI token budgets has gained traction within the tech industry, with many companies exploring ways to optimize their spending on AI development and deployment. By implementing token budgets, organizations can ensure that their investments are aligned with cost savings, while also maintaining a healthy work environment for their team members.