Overview

Investing.com reports that rising consumption costs are creating a budget pressure known as “token shock,” where AI usage expands faster than per‑token price declines. Bernstein’s analyst note highlights that software providers are moving from fixed unlimited subscriptions to usage‑based pricing to protect margins against growing inference expenses.

Cost Pressures (Token Shock)

Agentic AI intensifies the strain: an EY‑reviewed project showed that converting simple large‑language‑model prompts into multi‑step agentic workflows increased the cost per interaction by roughly thirty‑fold. Inference expenses can represent about 80 % of an AI model’s lifetime cost, shifting focus from upfront training spend to ongoing operational outlays.

Corporate Responses

  • Uber Technologies exhausted its 2026 AI coding budget within four months and introduced a $1,500 cap per employee for each agentic coding tool.
  • Walmart has capped employee access to an internal AI agent after previously offering unlimited tokens.
  • A major French insurer curtailed its use of Anthropic’s Claude two months after launch when costs exceeded expectations.
  • HubSpot launched a pricing model that ties charges to outcomes rather than token consumption, illustrating a move toward result‑based billing.

Project Outcomes

Discussions with technology and consulting providers reveal that many clients have postponed, scaled back, or cancelled at least one AI project due to cost concerns. One provider estimated that fewer than one‑quarter of its pilot projects generated positive returns, underscoring the uncertainty around AI economics.

Cost Management Strategies

Enterprises are managing expenses by reserving large, advanced models for complex tasks while routing simpler work to smaller, cheaper alternatives. Open‑source models and private infrastructure are gaining traction for high‑volume, predictable workloads.

Market Outlook

The outlook remains guardedly optimistic; while enterprise adoption may be slower than anticipated, software firms could respond with specialised models, smarter model routing, and outcome‑based pricing to mitigate token‑related cost pressures.