Tokens or humans? The new corporate trade-off

• Roughly 95% of enterprise AI runs on expensive frontier models even for simple tasks; routing work to cheaper model tiers could deliver 10x cost savings
• AI currently delivers less value than it costs, creating an 'unsustainable path' as new models are approximately 2x more expensive per token than predecessors
• Fortune 500 buyers are becoming more price-sensitive than the AI investment thesis assumes, challenging the market's assumption of indifferent demand at record-high valuations

Summary: Corporate AI Costs Force Trade-offs Between Technology and Headcount

Enterprise AI costs are significantly exceeding expectations, forcing CFOs at Fortune 500 companies to choose between AI spending and future headcount growth—a trade-off unprecedented in corporate technology budgets.

Key Findings

Cost Crisis: Companies report AI budgets being exhausted in one to two months despite being allocated annually. According to Arvind Jain, CEO of enterprise AI company Glean, each new frontier AI model costs roughly twice as much per token as its predecessor, creating an "unsustainable path."

Inefficient Spending: Approximately 95% of enterprise AI usage runs on the most expensive frontier models, even for simple tasks that could use cheaper alternatives. This represents a significant efficiency problem, with potential 10x savings available through proper model routing.

Evolution of AI Adoption: Matan Grinberg, CEO of Factory AI, describes three phases companies have experienced: initial board pressure to adopt AI, "zombie mode" (deploying AI regardless of cost), and current reassessment of premium model necessity.

The Core Problem: AI technology delivers value but doesn't yet pay for itself. The cost of implementation currently exceeds the return on investment for most enterprises.

Market Implications

This cost sensitivity challenges the prevailing AI investment thesis, which assumes demand will remain strong regardless of price. The revelation that Fortune 500 buyers are becoming price-sensitive poses risks to companies like OpenAI and Anthropic, whose business models rely on premium pricing.

The situation suggests potential headwinds for frontier AI model providers and opportunities for companies offering model routing optimization solutions that can deliver substantial cost savings.