Microsoft has entered the decision model market, launching Microsoft-Decision-1 on Alibaba's Qwen3.5-9B as a foundation while signaling plans to move to proprietary technology in future versions.
Decision models represent a distinct category of AI designed to optimize business outcomes across operational and strategic choices—distinct from text or image generation. The category was pioneered by Jev.
Microsoft's use of Qwen3.5-9B is a tactical choice: rapid market entry without building custom infrastructure from scratch. But the company has signaled this is a temporary arrangement. Internal development of decision models would give Microsoft several advantages: proprietary architectures that competitors cannot replicate, lower inference costs at scale, and tighter integration with its enterprise software stack—making the technology stickier for cloud and Office 365 customers.
The economics of decision models resemble SaaS: high-margin subscriptions or usage-based pricing for corporate clients seeking optimized outputs, advanced diagnostics, and custom training. Controlling the underlying model reduces reliance on external providers and protects pricing power as the category matures.
Building proprietary decision models also addresses a strategic vulnerability. Dependence on third-party models limits Microsoft's ability to differentiate, improve margins, or lock in enterprise customers. For a company that has spent decades building moats through software integration and ecosystem lock-in, the path to competitive advantage in decision AI runs through vertical ownership of the model layer.
