OpenAI partners with Oracle on Stargate: A data center leap for the AI future
Artificial intelligence continues reshaping the tech landscape, and OpenAI’s latest move signals a massive infrastructure-level bet on that future. In collaboration with Oracle, OpenAI has unveiled “Stargate” — a multibillion-dollar data center project designed to deliver unprecedented compute capabilities for large language models like GPT-5 and beyond. With staggering projected power demands and a timeline stretching into the latter part of the decade, Stargate sits at the center of a wider U.S.-based supercomputing buildout expected to support the next leap in generative AI. Here’s what we know about the project, what it could mean for OpenAI’s future, and why Oracle is now a critical pillar in the AI arms race.
Stargate: A new era of AI infrastructure
The Stargate project isn’t just another server farm — it’s a moonshot endeavor aimed at meeting the computational demands of next-generation artificial intelligence. Positioned to become operational by 2028, the facility will reportedly offer exascale-level computing power. That’s on the scale of 1 quintillion calculations per second. For OpenAI, whose ambitions rely on scaling deep learning models to unprecedented sizes, this is a foundational step.
The center will be located in the United States, emphasizing domestic infrastructure as a key leverage point in the global AI competition. Sources suggest the facility could cost upwards of $100 billion through 2030, making it one of the most expensive tech projects ever undertaken.
Oracle’s cloud infrastructure becomes AI-critical
The partnership with Oracle is more than symbolic. While Microsoft remains OpenAI’s most prominent investor and cloud provider, Oracle is carving out space as a decentralized but powerful alternative in AI infrastructure support.
Oracle brings with it expertise in high-performance computing (HPC) and a proven data center strategy that emphasizes modular buildouts and energy efficiency. The company’s Gen2 Cloud Infrastructure has already been used by OpenAI to train some of its models, and Stargate may see a dramatically scaled-up version of this tech stack. Larry Ellison, Oracle’s CTO, has publicly promoted the firm as the most cost-efficient and secure cloud option for AI training workloads, especially in GPU-intensive applications.
Power demand and scalability factors
Building exascale data centers isn’t just about racks of GPUs — it’s about power, water, heat, and sustainability. Stargate is set to require several gigawatts of power, likely sourced from a mix of renewable and carbon-neutral energy to align with long-term environmental targets.
The demand for energy highlights a broader challenge that hyperscalers face: the balance between AI acceleration and ecological responsibility. This will necessitate cooperation with utility providers and local governments, along with new cooling technologies and AI-optimized data center design to maintain sustainability standards.
The broadband AI race: Why this matters
Stargate is emblematic of a wider rush by major tech firms to lock in data processing capacity now — before demand outpaces availability. As generative AI spreads across industries, compute hardware and optimized data centers will define competitive advantage more than models or algorithms alone.
OpenAI’s move is also strategic. By diversifying its hosting partners beyond Microsoft, it ensures redundancy, negotiates pricing leverage, and risks mitigation in its long-term scaling plans. Oracle’s role underscores the broader shift where infrastructure companies become co-architects in AI development, not just landlords for storage and compute.
Final thoughts
Stargate isn’t just a data center — it’s OpenAI’s declaration of intent for the next decade. By teaming up with Oracle, it’s betting on a future where compute is the currency of progress, and access defines capability. The $100B price tag reflects the enormity of that ambition, and it may just be the start of a new phase in the AI infrastructure arms race. For infrastructure providers like Oracle, this is a golden moment to become indispensable in AI’s foundational layer. And for everyone watching, from investors to developers, Stargate marks a clear signal: model innovation is only as powerful as the hardware built to run it.
Image by: Lisa Li
https://unsplash.com/@lisa_li
