Microsoft and OpenAI have ended their exclusive partnership. On Monday, both parties announced that Microsoft is canceling its exclusive sales rights to OpenAI's technology—meaning OpenAI can now sell its models to any cloud platform, including Google Cloud and AWS. In exchange, Microsoft no longer needs to share revenue with OpenAI, and OpenAI's debt to Microsoft has been capped. This deal sends a strong signal: the AI "iron triangle" that has been tightly bound for nearly seven years is starting to loosen. Meanwhile, at the same time, another piece of news broke: OpenAI is working with Qualcomm and MediaTek to develop a custom AI smartphone chip. According to renowned analyst Ming-Chi Kuo, the core logic of this phone isn't about installing apps but using an AI Agent to replace applications—the phone continuously understands your context and coordinates between local small models and cloud-based large models to get things done. Looking at these two pieces of news together makes it even more interesting. OpenAI is "divorcing" Microsoft on one hand, while starting to build its own hardware entry point on the other. Previously, all of OpenAI's capabilities ran on Azure. While it seemed like an advantage, it was actually a shackle. Now, the cloud is free, and the company is starting to lay out plans for the terminal side as well—if they can pull it off, OpenAI won't just be "the ChatGPT company" anymore; it'll be a full-stack AI company controlling everything from chips to models to endpoints. Of course, the phone likely won't go into mass production until 2028, but Qualcomm's stock is already up 7% today. The capital market clearly thinks this direction has more imagination than simply making APIs.