According to Bloomberg, the Silicon Valley Data LLM Token Expenditure Index has fallen nearly 20% from its May peak as AI enterprises face mounting cost sensitivity from customers, signaling potential weakness in pricing power amid the industry's massive capital spending boom.
Research firm Allianz highlighted growing concerns about AI investment efficiency, noting the gap between investment and sales growth has reached 46%, even more severe than the 32% divergence observed during the 2001 telecom bubble. Separately, emerging regulatory pressures in the EU and U.S. are pushing enterprises toward lower-cost AI models, further pressuring margins for high-end platforms.