Silicon Valley is redefining engineer compensation by integrating AI compute capacity as a critical fourth component alongside salary, bonuses, and equity. As generative AI tools become central to software development, the cost of running underlying models, known as inference, has emerged as a key driver of productivity and a significant budget line for finance executives. Tech candidates are increasingly inquiring about their dedicated inference budgets during job interviews. Thibault Sottiaux, an engineering lead at OpenAI, noted that requests for access to AI coding tools like Codex are rising sharply. Usage rates per user are outpacing overall user growth, signaling that AI compute is becoming a scarce and valuable resource. Greg Brockman, OpenAI President, emphasized that the inference compute available to an individual will increasingly determine their overall software productivity. Consequently, engineers without access to substantial compute risk producing significantly less than their peers, potentially threatening their career prospects. Early indicators suggest this shift is already underway. Hakeem Shibly, a data specialist at Levels.fyi, observed a compensation submission from a software engineer that listed a Copilot subscription as a specific benefit. While minor, this represents a symbolic move toward making AI access a standard perk. Some industry voices, such as Peter Gostev of Arena, have proposed that recruitment platforms should explicitly list token budgets alongside salary ranges, treating compute allocation as a transparent job requirement. Investors are also taking note. Tomasz Tunguz of Theory Ventures predicts that companies will soon treat AI inference as a formal component of engineering compensation. He argues that as AI usage contributes to total cash burn, CFOs must track these costs with the same rigor as traditional headcount expenses. Tunguz estimates that with a 75th percentile software engineer salary of $375,000, adding $100,000 in annual inference costs could mean AI expenses account for over 20% of the fully loaded compensation cost by 2026. The fundamental question for finance leaders is the return on investment for this spend. Tunguz suggests that just as cloud infrastructure is judged by profit per hour of GPU use, employee value should be measured by productive work generated per dollar of inference cost. He noted that if an engineer burns $100,000 in AI costs annually, they must be eight times more productive to justify the expense. Tunguz has already begun integrating AI tools into his daily workflow, automating 31 tasks daily at an annual cost of approximately $12,000. He predicts that by 2026, engineers will negotiate their roles not only in dollars and equity but also in tokens, the standard unit for pricing AI model usage. This trend signals a structural shift in the tech labor market where computational power is as essential to an engineer's toolkit as a computer or a salary.