On July 27, Microsoft is prioritizing the allocation of its limited AI computing power to its own products, such as Microsoft 365 Copilot and GitHub Copilot, while the needs of Azure cloud customers are being placed in a secondary position. According to a report by Business Insider, this allocation strategy has put Microsoft CEO Satya Nadella under rare pressure since taking office: despite Microsoft's significant investments in building AI infrastructure, it is still struggling to meet the computing power demands of both its internal products and external customers. Microsoft's Chief Financial Officer Amy Hood stated in January this year that the company would prioritize meeting the computing power needs of its own AI products, followed by long-term research and development and innovation, and finally Azure customers. This means that Azure customers may not have priority access to the most scarce AI resources when purchasing Microsoft's cloud services. Microsoft's current core contradiction is that the rapid growth in AI demand has outpaced the expansion of data center and chip capacities. Hood cited an example, stating that if all GPUs launched in the first half of a fiscal year were allocated to Azure, its growth rate could have exceeded 40%, rather than the actual 39%. This gap exposes Microsoft's difficult choice in computing power allocation: whether to prioritize supporting Azure customers or its own products, such as Microsoft 365 Copilot and GitHub Copilot. Although the sales quota for the Azure sales team has been increased by 30% this year, computing power supply remains a challenge for internal balance. After GitHub experienced failures, Microsoft once used Amazon's cloud capacity to alleviate pressure and explored renting Oracle's infrastructure, but these efforts were not pursued due to security and compliance issues. The impact on Azure customers is that cloud resources have become even scarcer in the AI era, and enterprises may need to compete with Microsoft's own products for computing power resources when training models and deploying AI applications. Three years ago, Microsoft was seen as a leader in the AI wave due to its bet on OpenAI. However, as AI moves from product launches to daily use, the pressure has shifted from "having products" to "ensuring products can operate stably, affordably, and continuously." In the office software sector, although Microsoft 365 is a stable cash source for Microsoft, more and more users are beginning to use AI tools for document writing, summarization, etc. Microsoft needs to allocate sufficient computing power to Copilot to maintain its competitiveness. In the developer tools sector, although GitHub Copilot has a first-mover advantage, it also faces competition from tools such as Cursor and Anthropic's Claude Code. Against the backdrop of rapid growth in AI usage, Microsoft needs to address the challenges of expanding infrastructure and maintaining stability simultaneously. The shortage of computing power has also forced Microsoft to readjust its organizational divisions. Nadella has delegated more commercial business to Judson Althoff, while focusing on AI himself alongside the core engineering team. However, for Microsoft, the real challenge lies in resource allocation: ensuring that its own Copilot products do not fall behind while also preventing Azure customers from being left without the required capacity for extended periods. Nadella was once praised for his early bet on AI, but now he must face the resource dilemma that comes with AI's success: when its own Copilot, R&D teams, and Azure customers compete for the same batch of GPUs, Microsoft must decide which party will be served first with its limited computing power.