17 OCT 2026 — Accenture has disputed reports that it capped its employees' use of Anthropic's Claude at US$29 a month. The figure, first reported by 404 Media, is the amount the consulting firm will fund centrally for each employee. Usage above it must be charged to an approved project rather than refused. "Reporting that suggests that we have put a firm limit on our people's access to AI tools each month is incorrect," an Accenture spokesperson told Forbes.
The distinction is real, but it does not make the story go away. Some employees had reportedly been spending US$500 to US$1,000 a month, and the firm has now decided that most of that cost belongs to client work, not the company's own budget.
Disclosure: RECATOOLS is written with the assistance of Claude, made by Anthropic, whose product is the subject of this article. Readers should weigh that interest.
What changed
According to 404 Media, Accenture told many employees it would now spend only US$29 a month per person on Claude. Forbes reports the policy took effect on 1 October. Some of those affected had been spending as much as US$1,000 a month, 404 Media said, and were scrambling to find ways to keep their access.
Forbes, citing the same report, said spending above the central allowance must be charged to internal project codes with authorisation. Accenture's spokesperson said the US$29 figure "was used to demonstrate one part of the funding approach" and that Claude is "critical to how we deliver work for our clients and within our organization".
A partner that sells the product
The policy matters because Accenture is also an Anthropic partner. In December 2025 the two companies announced a multi-year partnership that created a dedicated Accenture Anthropic Business Group, committed to training about 30,000 Accenture staff on Claude and made Accenture a premier partner for Claude Code. Anthropic chief executive Dario Amodei called it the company's "largest ever deployment".
Selling and rationing
Accenture advises clients to adopt Claude at scale while managing its own internal usage more tightly. Those two positions are not contradictory: a consultancy routinely bills tools to the client work that needs them. What the policy shows is that Accenture now treats heavy Claude use as a cost to justify case by case rather than a general productivity benefit.
Why usage-based pricing bites
Forbes attributes the squeeze to consumption-based pricing. A seat-based subscription caps what a company pays per person. Consumption pricing charges for the tokens each person uses, so a developer running long coding sessions can cost many times what a casual user does. Anthropic has cut per-token prices this year, including a first cut to Opus pricing, but cheaper tokens can still produce larger bills if people use more of them.
RECATOOLS has tracked how Anthropic's own changes have shifted those costs, from a Claude Code limit change that worked out as a cut, to the gap between advertised and measured cost per task for Sonnet 5.5. Businesses trying to forecast their own spend can model it with our LLM cost calculator.
What to watch
Accenture has not published the policy, and 404 Media's full report is behind a paywall, so the exact terms are not public. The question for other large customers is whether Accenture's approach spreads. Moving heavy AI use onto project budgets keeps central costs predictable. It also means usage follows the clients willing to pay for it, not the employees who find it most useful.
