Oracle Commits $300B to 'Stargate' AI Datacenter Deal
Oracle has signed a $300 billion, decade-long AI infrastructure deal known as "Stargate," which will build out 4.5GW of datacenter capacity across multiple continents. The agreement is considered the world's largest public cloud commitment and a direct challenge to AWS, Google, and Microsoft in the AI infrastructure arms race. To help equip these new facilities, Oracle is also expanding its partnership with AMD to deploy 50,000 of the company's GPUs to meet customer demand.
- The "Stargate" initiative is a broader $500 billion AI infrastructure project involving OpenAI, Oracle, and SoftBank, aiming to build large-scale data centers across the U.S. Oracle's $300 billion agreement with OpenAI is a five-year contract to provide the land, power, and cooling for these facilities. - This deal is part of a massive acceleration in AI-related capital expenditures, with the five largest U.S. cloud and AI infrastructure providers—Microsoft, Alphabet, Amazon, Meta, and Oracle—collectively committing to spend between $660 billion and $690 billion in 2026. This represents a near-doubling of spending from 2025. - The 4.5 GW of power capacity secured by Oracle for this project represents roughly a quarter of the current operational data center capacity in the United States. Globally, data center electricity consumption is projected to more than double between 2024 and 2030, reaching 945 terawatt-hours. - The AMD partnership will equip the new Oracle Cloud Infrastructure (OCI) facilities, making OCI the first hyperscaler to deploy the new AMD Instinct™ MI355X GPUs. This is part of a larger trend where hyperscalers are actively investing in their own custom silicon to optimize performance and reduce long-term costs for specific AI workloads. - Oracle's strategy of not developing its own large-scale competing AI models has positioned it as a neutral and appealing platform for AI developers, a key factor in securing the deal with OpenAI. This contrasts with competitors like Google, Amazon, and Microsoft, who are developing their own custom AI chips and models. - To finance the hardware and energy demands of the Stargate project, Oracle's corporate debt has surpassed $100 billion, though its Remaining Performance Obligations (RPOs) have increased to over $455 billion. - The initial deployment of 50,000 AMD GPUs is scheduled to begin in the third quarter of 2026, with plans for further expansion in 2027 and beyond. The overall Stargate initiative aims to deploy over 2 million chips in total. - Oracle's late entry into the cloud market meant many of its 162 data centers were built recently, making them better equipped to handle the significant hardware and power requirements of modern AI model training.
