Microsoft is transitioning Copilot Cowork from flat-rate to usage-based billing and exploring a Microsoft-hosted version of DeepSeek as a cheaper alternative model option.
Microsoft is restructuring its Copilot Cowork enterprise AI tool with two major changes aimed at cost management and broader accessibility.
The company will shift from flat-rate pricing to usage-based billing, according to Copilot head Charles Lamanna, who cited sustainability concerns with the current model. This adjustment reflects a broader industry trend toward consumption-based pricing for AI services.
Simultaneously, Microsoft is evaluating a fine-tuned version of DeepSeek V4 as a lower-cost model option for Copilot Cowork users. DeepSeek, an open-source AI model from China, has gained attention for delivering competitive performance at a fraction of typical enterprise AI costs. A Microsoft-hosted version would allow enterprise customers to access these cost benefits within Microsoft's infrastructure.
The dual approach addresses different customer needs: usage-based pricing reduces barriers for organizations with variable or lighter AI usage patterns, while the DeepSeek option provides a budget-conscious alternative to Microsoft's proprietary models.
These changes come as enterprises increasingly scrutinize AI spending. By offering flexible pricing and lower-cost model alternatives, Microsoft aims to expand Copilot Cowork's addressable market while maintaining revenue through high-volume users and premium model tiers.
The moves signal Microsoft's willingness to integrate third-party models alongside its own technology—a pragmatic response to competitive pricing pressures in the enterprise AI market.
Companies deploying large language models face unexpectedly high token usage charges, forcing a reassessment of AI implementation costs. A Silicon Valley software maker and ecommerce firm are grappling with the economics of generative AI adoption.
South Korea's central bank cautioned that surging bonuses from the AI-driven semiconductor sector could trigger broader wage growth and consumer spending, complicating inflation control.
Zhipu has released GLM-5.2, a flagship model offering significant improvements in coding and agentic tasks. The model features a 1 million token context window and is available under an MIT license.
Databricks reported annualized revenue of $6.9B, up 80% year-over-year from $5.4B in Q4, driven by strong demand for its data analytics platform. However, CEO Ali Ghodsi warned that increased AI agent usage is raising costs and pressuring profit margins.