Feng Huang Wang·2025-10-09 23:17Core Insights - Microsoft is facing a prolonged shortage of data center resources, impacting its ability to meet cloud service demand, with limitations on new Azure subscriptions expected to last until mid-2026 [1] - The shortage is more severe than previously anticipated, with earlier estimates suggesting the situation would last until the end of 2025 [1] - The capacity issues affect both AI-related GPU servers and traditional CPU-based data centers, which handle conventional cloud service tasks [1] Group 1 - Microsoft is experiencing physical space and server shortages in key data center regions, including Northern Virginia and Texas [1] - The company has stated that most of its Azure services in the U.S. still have available capacity for existing customers to expand their workloads [1] - Microsoft plans to implement "capacity protection measures" to balance customer demand across its data center network in case of unexpected demand surges [1] Group 2 - The demand for computing power driven by AI applications, particularly since the launch of ChatGPT and GPT-4, has outpaced Microsoft's ability to expand capacity [2] - Microsoft is under pressure not only in the AI sector but also in its traditional cloud infrastructure, which supports various internet applications and websites [2] - OpenAI is identified as Microsoft's largest customer in the CPU-based workload segment [2]