Walmart limits Code Puppy AI tokens for employees

Walmart has begun assigning fixed AI token limits to employees who use its internal assistant Code Puppy to manage rising per‑token costs as large language models move to pay‑per‑use billing. The limits apply across tasks such as spreadsheet analysis and presentation creation and affect the company’s roughly 2.1 million workers.

Earlier this year, Walmart encouraged employees to experiment with Code Puppy and other AI tools and provided training on how to use them. After demand exceeded expectations and providers shifted higher‑tier plans to usage billing, the company introduced per‑employee caps to control expenses.

Walmart has advised staff to “use AI where it can create value” and issued guidance on selecting the most suitable AI tool for a task. Employees retain access to other company‑funded AI platforms, but interaction counts and complexity are now treated as measurable costs. The company has not disclosed exact token allotments or how limits will vary by role and says it will monitor usage and adjust allowances as needed.

Enterprise AI providers have moved portions of their offerings from flat subscriptions to metered pricing. Some higher‑tier plans from major model providers now charge based on tokens processed. Microsoft implemented paid access for GitHub Copilot beginning June 1. These pricing changes have increased the visibility of per‑request costs for companies running large-scale internal AI programs.

Technical factors that drive token consumption include larger models that perform internal reasoning steps, workflows that run multiple agents, and automated pipelines that loop through iterative prompts. Each additional model call or agent cycle increases token use and the associated bill.

Several firms have reported faster than expected budget depletion under usage billing. One large employer revealed it used its annual AI budget within the first four months of the year, an example of how per‑use charges can accelerate spending when demand is high.

Walmart frames the token limits as a cost‑control measure and a way to make AI spending more predictable. The company says the policy is intended to steer workers toward appropriate models for different tasks and to preserve productive use while keeping a closer account of AI expenses.