Uber COO Andrew Macdonald stated in a Rapid Response interview that higher AI token usage has not produced proportional increases in consumer features, coinciding with CEO Dara Khosrowshahi's announcement of slowed hiring to offset AI investments. Macdonald cited Uber CTO Praveen Neppalli Naga's April comments to The Information on exhausting the 2026 Claude budget as triggering internal reviews of token consumption versus output. Primary reporting from Business Insider May 2026 details these trade-offs without direct metrics tying spend to shipped features. Macdonald's assessment aligns with Uber's Q1 2026 earnings transcript, where Khosrowshahi explicitly linked AI spend to headcount restraint. Duolingo's April reversal on AI usage in performance reviews, reported in its CEO podcast, shows parallel pushback at other firms evaluating tokenmaxxing. No primary data from either company quantifies token-to-feature conversion rates. Uber's internal engineering discussions, per the interview, found no verifiable correlation between elevated token stats and 25% more useful features. The Information's Naga interview and Uber earnings call transcripts provide the core citations; Business Insider adds context on headcount impacts without independent verification of ROI claims.