The situation

AI budgets are tightening as enterprises shift from experimental "tokenmaxxing" to ROI-focused deployment. Uber recently implemented spending tiers to cap monthly AI expenditures, while AI startup Lindy switched 100% of its traffic from Anthropic's Claude to DeepSeek's cheaper open-weight alternatives—saving millions within months. As Lindy CEO Flo Crivello stated: "It's a matter of survival for the business." The implication for AI labs is clear: the era of uncapped enterprise spending is ending. OpenAI and Anthropic face a new reality where customers demand efficiency, not just capability. This favors leaner inference providers and open-source models that can deliver comparable performance at fraction of the cost.

Why it matters

The AI market is entering a more demanding phase. Capability still matters, but the bigger question is whether companies can turn model demand, compute access, and enterprise adoption into durable economics.

What to watch

Watch whether the facts create second-order effects: policy responses, customer behavior, capital flows, competitive pressure, and whether the story becomes a one-day headline or a lasting shift.