Microsoft weighs shifting Copilot to usage-based pricing as AI costs climb

Microsoft is reportedly rethinking how it prices Copilot, with an eye toward shifting the enterprise AI assistant toward usage-based pricing as usage grows and underlying AI costs become harder to forecast.

In a report carried by Yahoo Finance, Microsoft is described as evaluating changes to Copilot Cowork, its enterprise offering that helps teams work with business data through AI. The article says the company is considering moving from pricing that is less directly tied to consumption to a model that charges based on usage, which would better reflect how compute and model access costs scale with how intensively customers use Copilot features.

The report frames the shift in the context of higher and more variable AI costs across the industry. If customers intensify their use of Copilot, the cost to deliver that capability typically rises as well, particularly when services involve repeated queries, longer context, or more complex responses. Usage-based pricing can potentially reduce mismatches between what Microsoft pays for AI inference and what it collects under a fixed-fee structure.

Beyond pricing, the Yahoo Finance article also says Microsoft is evaluating alternative AI models for Copilot. It specifically references options that include DeepSeek, suggesting Microsoft may be looking beyond one provider or model family to manage cost, performance, or availability. However, the report does not provide further details on whether those model evaluations are limited to certain copilots or workflows, or whether any changes are imminent.

Microsoft has not, in the information available here, published a full explanation of the proposed pricing structure. Key details such as what counts as “usage” (for example, number of prompts, tokens, seats, or active workflows), which customer tiers would be affected first, and whether any contracts already in place would be renegotiated are not described in the Yahoo Finance post.

This comes as enterprises across the technology sector continue to expand their pilots and deployments of generative AI tools. Copilot is one of the most prominent enterprise-focused rollouts, integrated into Microsoft’s productivity and business software ecosystem. For Microsoft, pricing strategy is especially significant because Copilot value is often tied to how frequently employees use it, and usage-based economics can influence customer behavior, procurement decisions, and expectations around total cost of ownership.

Even with the reported direction, the company’s exact timeline and end-state remain unclear. The Yahoo Finance report, based on its description, points to evaluation rather than a finalized product announcement, and it does not lay out specific rates, contract terms, or which geographic markets or enterprise segments would be covered first.

Investors and customers may watch for clearer guidance from Microsoft through pricing pages, enterprise agreements, or product updates around Copilot Cowork. Additional transparency on model choices and how Microsoft plans to manage AI inference costs could also become a focus, particularly if usage-based pricing is tied to those technical changes.