U.S. — Microsoft is considering making China's DeepSeek AI model available for its Copilot Cowork AI agent and is transitioning the service to usage-based pricing. The company is evaluating open-source models as lower-cost alternatives to products from Anthropic and OpenAI.

Charles Lamanna, Microsoft's executive vice president for Copilot, addressed the reason for the pricing change, stating, "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." The Trump administration previously limited access to the most advanced U.S. AI models.

Gary Marcus, a professor emeritus of psychology and neural science at New York University, stated that more U.S. companies are using cheaper, open-source Chinese AI models. He noted that enormous capital expenditures have not eliminated errors produced by large language models (LLMs) and have removed technical moats that might give hyperscalers a competitive edge.

Marcus said that this situation "has led to price wars coexisting with high operating expenses (needed to run bigger data centers to train and operate new models) and low or even negative margins, since all are building more or less the same product." Meta, Microsoft, Alphabet, Amazon, and other hyperscalers are collectively spending hundreds of billions of dollars annually on data centers, chips, and other infrastructure, with collective spending expected to exceed $700 billion in 2026.

Marcus stated that most chips depreciate as more advanced versions are released and that high-end AI models could be displaced by more efficient rivals that do not rely on as many expensive chips. Security researchers have indicated that China's Zhipu AI can now match U.S. models in finding security bugs, though Anthropic's and OpenAI's products remain superior at other tasks.

OpenAI floated the idea of a government backstop for data center financing but later retracted it after widespread outcry. Airlines received federal bailouts after the September 11, 2001, terrorist attacks and again during the COVID-19 pandemic. Marcus stated that a worst-case scenario could involve governments bailing out AI giants if the fallout impacted pension funds, banks, or the global economy. He concluded by saying, "To make the bet on the version we have now is premature."