Accenture caps Claude Enterprise funding at $29 per employee after heavy users spent ~$1,000 a month

Accenture, one of the largest enterprise buyers of AI services, is tightening its Claude budget. According to an internal memo first reported by 404 Media, the firm will centrally fund just $29 a month per employee for Claude Enterprise. Some employees had been running up roughly $1,000 a month. The memo warns that costs beyond the centrally funded amount will be handled differently, effectively pushing heavy usage onto teams or projects.

The gap between $29 and $1,000 shows how agentic coding changes cost. A chat assistant used a few times a day has a predictable cost. An agent that reads entire repositories, runs tool loops and retries can burn through millions of tokens per developer per day, especially on top-tier models. Seat-based thinking breaks down when usage is metered by tokens and a small minority of power users dominates the bill.

The memo lands at a sensitive moment for Anthropic. The company is heading into a November IPO with investors reportedly debating a $1.5T versus $2T valuation, and enterprise consumption is central to its revenue story. It has also just shipped tools that address this exact pressure: Claude Haiku 5.5 at $0.10/$0.50 per Mtok and client-side auto model selection in Claude Code, which route cheaper models for simple steps. OpenAI and Google face the same dynamic with Codex and Gemini agents.

The debate in developer communities is whether caps like this will slow adoption or simply professionalize it, with budgets, model routing and usage dashboards becoming standard. Watch whether other consultancies and large employers follow Accenture, and whether vendors respond with enterprise pricing built around predictable per-seat bundles rather than raw token metering.