Accenture shifts from AI push to token rationing as bills soar

Accenture has started trying to stop non-technical workers from using company AI tokens on trivial tasks such as converting PDFs into presentation slides. The move marks a sharp turn for a firm that had pushed regular AI use so hard that senior staff were told promotions could hinge on it.

The cost pressure is showing up in internal budgets, not just in strategy decks. Accenture is seeing “soaring token spend” from workers outside technical teams, a sign that everyday tasks can quickly turn into material expense when employees treat generative AI as an all-purpose shortcut. What once looked like a broad productivity push is now forcing leaders to ask which uses actually justify the bill.

That shift is colliding with the way Accenture has promoted AI in its own business. CNBC reported on February 19, 2026, that the company was linking promotions for senior staff to regular use of its AI tools, especially for senior managers and associate directors. Accenture’s public leadership page identifies Julie Sweet as chair and chief executive officer, and the company has also described AI as a growth engine while launching Accenture Edge to help mid-market companies adopt AI and modernize technology. The message to clients and employees has been expansion; the message inside the company is becoming restraint.

The new caution fits a broader finance debate now running through corporate AI programs. Deloitte said on February 12, 2026, that AI computing demand was surging and AI costs were volatile and complex. By March 18, Deloitte was arguing that many organizations now treat AI as a core budget allocation rather than a discretionary spend item. Its 2026 guidance said token consumption can quietly push costs across applications, cloud and infrastructure, and that CFOs need clearer unit economics, governance, and forecasts for when spending will jump.

That matters beyond one consulting firm. Investors have also been pressuring AI-dependent businesses and memory-chip makers as questions grow about returns on heavy AI spending. Inside companies, the argument is no longer about whether employees should use generative AI at all. It is about who gets access, which tasks deserve it, and how fast the token bill can outrun the savings.