Accenture has reportedly limited central funding for many employees’ Claude Enterprise usage to $29 per person a month, with additional spending directed to project or team budgets from October 1. For workers already using hundreds of dollars of the AI assistant each month, keeping the same workflow may now depend on another budget absorbing the bill.

404 Media’s Joseph Cox reports that some employees were spending as much as $1,000 monthly. The change concerns who pays for that use: the reported arrangement leaves a route to spending above $29, rather than imposing a company-wide ban on Claude.

$29Reported monthly central allowance per affected employee
$500 to $1,000Reported monthly usage for one employee’s Claude Code workflow

The October 1 date and funding instructions appear in The Next Web’s account of the internal announcement reviewed by Cox. Additional consumption must be charged to a project or team WBS, an internal budget code. That account traces back to 404’s reporting; it is not separate confirmation, and the internal notice has not been independently authenticated by this publication.

The employee messages make the budget consequence concrete. According to that account of Cox’s reporting, one worker sought a faster Premium upgrade because monthly usage above $500 would affect project costs. Another described spending $500 to $1,000 a month using Claude Code, Anthropic’s coding assistant, to build and test an accelerator for several clients. The messages do not establish whether an upgrade would resolve those costs.

At that second worker’s reported usage level, a $29 central allowance would leave $471 to $971 each month for another budget to cover, assuming spending stayed the same. A third employee reportedly said their account had been revoked, but the reporting does not establish that reaching $29 caused the loss of access. The employees are anonymous, and their underlying messages were not available for independent inspection.

Who pays above $29?

Under the reported arrangement, central funding covers $29 per affected employee each month, with additional consumption directed to a project or team WBS budget. The figure is an internal allowance, not an established Claude subscription price or a blanket ban. The reporting does not say whether additional billing requires approval or what happens technically when an account reaches $29.

A budget decision closer to the work

Claude’s enterprise billing helps explain why access and usage can be separate expenses. Anthropic’s current documentation says usage-based Enterprise plans charge a seat fee for access, then bill consumption separately at API rates. Older Standard and Premium plans include usage limits and can add metered usage credits with spending controls. Accenture’s contract and configuration are not public, so those documents explain possible billing mechanisms without establishing which applies there.

The affected workforce is also unresolved. The reported notice says Claude remains available according to role and business need, while warning that a license and WBS do not guarantee continued access. It does not establish an affected headcount. Accenture and Anthropic’s December 2025 partnership announcement promised Claude training for approximately 30,000 professionals and Claude Code availability for tens of thousands of developers. Those commitments are not counts of active users or recipients of this notice.

In its response to 404, as relayed by The Next Web, Accenture emphasized matching tools to work and continuing to invest in AI capabilities. It did not directly confirm or deny the reported allowance, effective date or budget process.

Our read is that the funding shift brings the cost decision closer to the teams receiving the work. That does not establish whether the spending was wasteful or worthwhile. BCG’s advisory framework argues for measuring cost per successful outcome, including human effort and review; it is not an assessment of Accenture. The employee spending figures come without evidence of time saved, output quality or rework. The consequential test for a budget owner is whether a $500 workflow produces enough useful work to justify continuing it.