Oracle is reportedly set to invest approximately $40 billion in Nvidia’s high-performance chips as part of a large-scale effort to build a U.S.-based data center designed to support OpenAI’s growing compute needs.
Highlights
The initiative, part of the broader Stargate Project, was detailed in a report by the Financial Times, citing unnamed sources familiar with the deal.
A Major Move in AI Infrastructure Development
The planned data center will be located in Abilene, Texas, and is expected to become operational by mid-2026. Oracle will own and operate the facility, with OpenAI leasing compute power to scale its models.
Rather than building its own infrastructure, OpenAI is opting for a partnership-based approach to access accelerated computing resources—particularly critical as demand increases for services like ChatGPT and other AI systems.
The facility will utilize Nvidia’s top-tier GB200 Grace Blackwell chips, with plans to deploy approximately 400,000 units over a multi-phase rollout. An initial batch of 16,000 chips is expected to be running by summer 2025, scaling up to 64,000 by the end of 2026.
Financing and Project Scope
The infrastructure buildout is backed by significant financial support. According to the Financial Times:
- JPMorgan is providing two loans totaling $9.6 billion.
- Investment firms including Crusoe and Blue Owl Capital have contributed around $5 billion in equity.
Oracle has committed to a 15-year lease on the Abilene facility, indicating long-term strategic interest in AI infrastructure.
While neither OpenAI, Oracle, nor Nvidia have issued public statements regarding the report, the scale of the initiative highlights a broader shift in how major tech firms approach compute resource management.
The Stargate Project
The Abilene data center is part of the Stargate Project, a collaboration involving OpenAI, Oracle, SoftBank, and MGX (an Abu Dhabi-backed firm). The consortium reportedly aims to invest up to $500 billion over four years to develop AI infrastructure across the United States.
The Abilene site is expected to deliver 1.2 gigawatts of power, marking it as one of the most energy-intensive AI data centers globally. The initiative is also being framed as a contribution to U.S. economic revitalization and technological self-sufficiency.
Global Expansion
The Stargate initiative extends beyond the U.S. Plans are also underway for a similar data center in Abu Dhabi, United Arab Emirates, projected to host over 100,000 Nvidia chips. The first operational phase of the UAE site is scheduled for 2026.
Strategy of Oracle
Oracle’s role in the Stargate Project signals its growing ambition to compete in the AI and cloud infrastructure space.
By aligning with OpenAI and Nvidia, Oracle is positioning itself alongside other cloud leaders such as Amazon, Microsoft, and Google, particularly in the high-performance computing segment critical to future AI development.
The partnership also illustrates OpenAI’s strategy to diversify its compute sources beyond its long-standing collaboration with Microsoft Azure. This move may help alleviate resource bottlenecks and improve scalability for increasingly complex AI models.
