Uber's AI budget burn exposes ROI measurement gap across enterprises
Tiffany Luck of NEA says companies are struggling to measure returns on AI spending after early adopters like Uber exhausted annual budgets in months.
Enterprises are still struggling to measure return on investment from AI deployments, according to Tiffany Luck, a partner at New Enterprise Associates. The comment arrives as companies reckon with the costs of the "tokenmaxxing" trend that swept Silicon Valley earlier this year, when executives pushed employees to maximize AI usage without clear guardrails on spending.
Uber reportedly burned through its entire annual AI budget in just a few months. Other companies have responded by cutting Claude licenses for parts of their organizations, while Meta shut down an internal leaderboard that tracked AI usage. The pattern suggests that the initial enthusiasm for unrestricted AI access has collided with finance teams demanding accountability.
The ROI question is particularly thorny because many AI deployments don't map cleanly to traditional software metrics. A chatbot that answers customer service queries might reduce ticket volume, but quantifying the value of a coding assistant or a research tool requires new frameworks that most companies haven't built yet. Finance teams are asking for hard numbers on productivity gains, and engineering leaders are finding those numbers difficult to produce. Companies that can demonstrate concrete ROI—whether through headcount savings, faster time-to-market, or measurable quality improvements—will have an easier time justifying expanded budgets. Those that can't may find their AI initiatives scaled back or restructured around narrower, more measurable use cases.
