The Daily Froth: A Token of Appreciation, and an Invoice - The Cost of Enterprise AI

Research evidence excerpt

The Daily Froth: A Token of Appreciation, and an Invoice - The Cost of Enterprise AI

Economics

U.S. Economics

The Daily Froth: A Token of Appreciation, and an Invoice - June 19, 2026

The Cost of Enterprise AI

The first phase of enterprise AI was about proving it worked. The second phase is about proving it is worth

the cost. Uber recently revealed that it exhausted its entire 2026 AI budget in just four months and has since

imposed limits on employee AI usage to keep costs under control. Similarly, Walmart announced it has placed

limits on employee usage of its in-house AI tool to manage their rising operational costs. Even one of AI's biggest

hyperscalers, Microsoft, is asking employees to be more deliberate about how they use AI. As Satya Nadella said,

"Don't use frontier models for non-frontier problems." The conversation is increasingly shifting from capability to

economics.

These headlines have raised questions about whether cost is becoming a meaningful headwind to AI adoption.

We think the bigger story is that enterprises are moving from experimentation to optimization. Adoption has

progressed more slowly than many expected, but that likely reflects the practical challenges of implementation

(including governance, compliance, security, and workflow integration) rather than concerns about the technology

itself.

While costs may rise alongside model capabilities, so too should the value AI generates. For most enterprises,

the question is no longer whether to use AI, but how to generate the greatest return from it.

Recent data from Ramp suggest that AI spending remains modest for the average employee, with median spending

of just $11.38 per month. Costs rise dramatically among heavy users, however, with spending among the top 1%

approaching $7,500 per month.