Enterprise AI spending hits reality check as Uber burns annual budget in months

The AI spending spree that defined early 2026 is colliding with budget reality. Tiffany Luck, a partner at New Enterprise Associates, told TechCrunch this week that the "tokenmaxxing" trend — where executives pushed employees to use AI tools as aggressively as possible — has given way to a sharp ROI reckoning. Uber reportedly exhausted its entire annual AI budget in just a few months, while other companies have started cutting Claude licenses for parts of their organizations. Meta went further, shutting down an internal leaderboard that tracked AI usage across teams.

The shift marks a turning point for enterprise AI adoption. What started as an all-in bet on productivity gains has run into the hard math of per-token costs at scale. Companies that once treated AI budgets as infinite are now asking which use cases actually justify the spend. Luck's comments suggest the industry is moving from experimentation to cost discipline — a phase that will likely accelerate through Q3 earnings calls, where finance teams will finally quantify whether the tokenmaxxing era delivered promised returns or became a costly detour.

Luck also discussed the emerging IPO pipeline for AI companies and the rise of personal agents — software that acts on a user's behalf across multiple services. Both categories face the same reckoning: sustainable business models require demonstrable ROI, not just adoption velocity. The next phase of enterprise AI will separate companies that found genuine productivity wins from those that simply burned through budgets chasing novelty. Watch for the first major AI vendor to publicly disclose per-customer unit economics in an earnings call — that transparency will signal whether the market has truly shifted from growth-at-all-costs to profitability.