Microsoft moves Copilot Cowork to consumption-based pricing, eyes cheaper AI models

Microsoft has announced a shift to usage-based pricing for its enterprise AI assistant, Copilot Cowork, ending the previous unlimited-use model. Under the new structure, businesses will pay according to the computing resources and AI processing power they consume, a change the company says reflects the variable and intensive nature of agentic automation.

Copilot Cowork is an agentic AI tool designed to autonomously handle complex, multi-step tasks by continuously invoking large language models, retrieving data, and interacting with external applications. Unlike simpler chatbots, such tools can substantially improve productivity, but they also incur significantly higher computational costs if left unmonitored.

Microsoft explained that internal testing revealed some customers were making extensive use of the platform, making a flat-fee unlimited model unsustainable. The pay-as-you-go approach is intended to align costs more closely with actual usage while allowing the company to responsibly scale the service.

The announcement coincides with a broader surge in demand for agentic AI capabilities across industries. At the same time, organizations are keenly aware of the financial overhead tied to running advanced AI models at scale. To address this, Microsoft is also exploring more cost-efficient AI models, including a version of DeepSeek, as it seeks to lower both its operating expenses and the price passed on to customers.

While Microsoft has not disclosed exact per-use rates, it indicated that customers will have granular visibility into their consumption and associated bills. The move mirrors a wider trend in the tech industry toward metered cloud and AI services, giving enterprises more control over expenditure while challenging them to optimize how agentic systems are deployed.

The change is effective immediately for new Copilot Cowork customers, with existing subscribers transitioning over the coming months. Microsoft expects the pricing model to encourage more targeted and efficient use of agentic automation, benefiting organizations that can balance task complexity with resource consumption.