
3 min readJun 17, 2026 03:07 PM IST
Microsoft has said companies using Copilot Cowork will now pay based on how much compute and AI processing they consume. (Image: Microsoft)
Microsoft is changing how businesses pay for its Artificial Intelligence-powered Copilot Cowork platform, introducing usage-based pricing as the company expands access to the enterprise productivity tool and explores cheaper AI models, including a version of DeepSeek, to reduce costs.
The move comes as demand for agentic AI tools continues to surge. Unlike traditional chatbots, tools such as Copilot Cowork, Anthropic’s Claude Code and OpenAI’s Codex can autonomously carry out complex tasks by repeatedly calling AI models, retrieving data, and using external tools. While this can significantly boost productivity, it can also lead to rapidly increasing computing costs.
Microsoft has said companies using Copilot Cowork will now pay based on how much compute and AI processing they consume. The company said its testing showed the service could not be offered on an unlimited-use basis because some customers were already using it extensively.
“We have users who do hundreds of tasks a week, which is great — they’re way productive — but the consequence is the costs can go very high,” Charles Lamanna, Microsoft’s Executive Vice President for Copilot, Agents and Platform, told Axios.
Tapping DeepSeek
As part of efforts to reduce costs, Microsoft is evaluating a fine-tuned version of DeepSeek V4 or another open-source AI model as a lower-cost alternative to the Anthropic and OpenAI models currently powering Copilot Cowork. The company expects to announce its final choice in the coming weeks.
The possibility of Microsoft incorporating technology from the Chinese AI company DeepSeek could attract scrutiny, especially amid ongoing concerns about data security and geopolitical tensions surrounding AI development.
However, Microsoft stressed that any DeepSeek-based offering would be optional for customers and fully hosted on Azure. Customer data would remain within Microsoft’s cloud infrastructure and continue to be protected by Azure’s enterprise security, compliance and data residency controls.
Story continues below this ad
The company also said it has fine-tuned the Copilot Cowork model and added additional safeguards designed to reduce bias and improve reliability.
The development reflects Microsoft’s broader shift toward a multi-model AI strategy. While the company remains a major partner of OpenAI, it has increasingly added support for other AI models to give customers more flexibility in balancing performance and cost.
Microsoft recently announced the general availability of Copilot Cowork after a three-month preview period. According to the company, more than half of the Fortune 500 have already tested the platform. The service is designed to execute complex, long-running tasks across multiple business applications and data sources, delivering completed work rather than simply generating recommendations.
Copilot Cowork currently runs on Anthropic’s Opus and Sonnet models, while GPT models are available through Microsoft’s Frontier program. Microsoft’s own “Cowork 1” model is also expected to launch in the coming weeks, with the company positioning it as a lower-cost option for everyday enterprise workloads.
View original source — Indian Express ↗

