Venture investor Tiffany Luck, partner at NEA, says many enterprises are still working out how to measure and justify returns from AI deployments. She points to a recent shift from rapid experimentation toward tighter budgeting and more controlled usage. Multiple reports describe “tokenmaxxing” practices—where companies encourage employees to use AI tools extensively—as a phase that has run into financial limits. For example, Uber is reported to have spent through its annual AI budget within a few months. Other companies have reportedly reduced or rescinded AI product access in parts of their organizations, including cuts to Claude licenses. Meta has also been cited for ending internal AI usage incentives, such as a company leaderboard that encouraged engagement.
Taken together, the accounts suggest enterprises are moving from volume-driven adoption toward cost management, governance, and clearer ROI expectations. Luck’s comments emphasize that the ROI “reckoning” is ongoing: organizations are learning how to align AI usage with measurable outcomes, rather than focusing solely on usage growth.
