NVIDIA CEO Jensen Huang announced an innovative recruitment strategy at this year's GTC conference: offering engineers token budgets. As the first executive in Silicon Valley to publicly propose this concept, Huang stated that every engineer may soon require an annual token allocation, with NVIDIA willing to provide additional resources accordingly. He envisions providing tokens equivalent to half of base salary as a subsidy for engineers earning hundreds of thousands of dollars annually, thereby boosting their productivity tenfold. Huang noted that tokens have evolved into one of Silicon Valley's critical hiring tools. Engineers often gauge job value by the number of tokens included in compensation packages. This is because access to tokens enables more frequent invocation of AI models for inference and generation, significantly enhancing work efficiency. Tokens serve as fundamental units for processing text or code within artificial intelligence systems, typically measuring computational resource consumption based on usage costs. This initiative comes amid surging demand for AI computing power. Huang predicts that leveraging NVIDIA's enhanced token-generation capabilities through its Blackwell and Vera Rubin architectures could drive procurement orders exceeding $1 trillion before 2027. Previously, media reports indicated Silicon Valley companies were exploring AI inference compute capacity as a fourth major recruiting lever beyond traditional salaries—a trend investors have closely monitored, urging firms to explicitly list token allocations in job postings. NVIDIA's move marks a pivotal shift where compute resources have transformed from mere infrastructure into direct talent acquisition instruments.