Companies raced to adopt AI, expecting instant productivity boosts and cost cuts. Instead, they face soaring bills and underwhelming returns.

Microsoft pulled back on Anthropic’s Claude Code licenses after just six months. The tool’s rapid adoption inflated costs beyond what finance teams could justify. Employees embraced AI assistants, but spending spiraled faster than any measured productivity gains.

Uber burned through its entire 2026 AI coding budget in four months. The company even ranked teams by AI usage to push adoption. Yet, the promised link between AI use and better consumer features remains missing. Uber’s COO called it a “head-exploding moment.”

Many firms report employees running AI tasks for trivial purposes—like checking the weather through expensive AI models. This behavior inflates costs without adding value. One CTO described such uses as “incredibly expensive and roundabout.”

Token-based pricing, the backbone of AI billing, charges companies per unit of data processed. Heavy AI users rack up massive bills that outpace human labor costs. NVIDIA’s VP admitted compute costs surpass employee expenses on some teams.

Executives once feared AI would cut jobs. Now, some leaders walk back claims of massive labor displacement. Instead, companies face a cost-management crisis. Most admit they underestimate AI expenses by 25% or more, complicating budgeting and financial forecasting.

To control costs, many enterprises plan to shift AI workloads from public clouds back to private or hybrid infrastructure. Sixty-one percent already use hybrid AI systems, but few accurately track on-premise AI expenses. This blind spot hampers cost discipline.

Workers question why AI-driven productivity gains haven’t translated into better wages or profit sharing. Labor tensions rise as employees resist unchecked AI adoption. In South Korea, Samsung narrowly avoided strikes over AI-related profit sharing demands.

Microsoft and Uber’s pullbacks signal a broader reckoning. The AI race is no longer about experimentation alone. Boards demand tangible results and tighter cost controls. Investors grow wary as AI’s promised profits lag behind runaway infrastructure bills.

Some industry voices call for smarter AI use, not just token “maxxing.” The tech world is pivoting from hype to hard questions: Does AI justify its colossal price tag? So far, the answer is murky.