NEA’s Tiffany Luck says enterprises are still figuring out their AI ROI
TL;DR
TechCrunch frames the current AI mood swing: after the „tokenmaxxing“ push, companies are now staring at the actual AI bill. The piece cites Uber reportedly burning through its annual AI budget early, some firms trimming Claude licenses, and Meta ending an internal leaderboard. NEA partner Tiffany Luck says this is where the enterprise gap sits: companies want broader AI use, but ROI measurement is still immature.
Nauti's Take
Tokenmaxxing was a very Silicon Valley reflex: turn usage all the way up, then ask what the invoice means. That can be useful as a learning phase, but it is weak as an enterprise strategy.
Internal AI rollouts need more than enthusiasm; they need hard measurement around which task becomes faster, cheaper, or better. Without that discipline, AI budgets risk becoming prestige spend with impressive usage charts and thin outcomes.
Briefingshow
The key point is not simply that AI costs more than expected. Many companies appear to have confused usage with value: more prompts, more tokens, more licenses. The conversation is now moving from adoption to impact, meaning cost per workflow, saved time, quality gains, and measurable business outcomes.
Sources