OpenAI & SoftBank Fuel AI Race with $1B Bet on Energy Infrastructure

REDWOOD CITY, CA – January 09, 2026 – In a landmark move signaling a new phase in the artificial intelligence arms race, OpenAI and SoftBank Group have announced a strategic partnership with SB Energy, injecting $1 billion to construct the massive physical infrastructure required to power next-generation AI. The deal, centered on building and operating enormous data center campuses, is a cornerstone of the ambitious 'Stargate' initiative, a $500 billion U.S.-based AI infrastructure project.

As part of the agreement, OpenAI and SoftBank are each investing $500 million into SB Energy, a SoftBank subsidiary. Underscoring the project's immense scale, OpenAI has also signed a 1.2-gigawatt (GW) data center lease with SB Energy for an initial site in Milam County, Texas. This single facility's power consumption is equivalent to the needs of roughly 800,000 homes, highlighting the staggering energy requirements of advanced AI.

This partnership represents a critical shift in the tech industry, moving beyond software and algorithms to focus on the capital-intensive world of power plants and physical construction. It aims to solve one of the biggest bottlenecks for AI supremacy: the availability of reliable, massive-scale computational power.

"Partnering with SB Energy brings together their strength in data center infrastructure and energy development and OpenAI's deep domain expertise in data center engineering," said OpenAI co-founder and President Greg Brockman. "The result is a fast, reliable way to scale compute through large, highly optimized AI data centers."

The 'Stargate' Supercomputer Takes Shape

The collaboration is a major step in realizing the 'Stargate' project, a multinational joint venture unveiled in January 2025 with the goal of investing up to $500 billion in U.S.-based AI infrastructure by 2029. With SoftBank as the lead financial partner and OpenAI helming operations, the initiative has been compared in scale and national importance to the Manhattan Project. Its stated goal is to build a nationwide network of exclusive, advanced data centers to secure American leadership in AI.

Today's announcement accelerates a project already moving at a breakneck pace. By late 2025, the Stargate initiative had already announced sites totaling nearly 7 GW of planned capacity, putting it on track to meet its initial 10 GW goal far ahead of schedule. The Milam County site joins a growing list of locations in Texas, New Mexico, and Ohio, forming the physical backbone of OpenAI’s future computational needs.

This aggressive build-out is a direct response to what insiders describe as a constant state of being "under compute." The demand for the powerful GPUs and specialized servers needed to train and run models like ChatGPT is skyrocketing, forcing OpenAI and its competitors to secure not just the chips, but the buildings and power grids to support them.

The Energy-AI Nexus

The partnership places SB Energy, an integrated energy and infrastructure developer, at the heart of solving AI’s existential energy problem. The immense power draw of facilities like the 1.2 GW Milam County campus is forcing a convergence of the technology and energy sectors. Reports have circulated of internal OpenAI memos outlining ambitions for as much as 250 GW of compute capacity by 2033—a demand comparable to the entire electricity consumption of India.

To meet this challenge, SB Energy plans to do more than just build data centers; it will build the power sources for them. The company announced its intention to construct new power generation to support the Milam County facility, a move designed to add capacity to the Texas grid and insulate local ratepayers from the facility's enormous demand. The company has a growing track record in this area, having previously developed large-scale solar and energy storage projects, including a 900-megawatt solar park near Austin that supports one of Google's data centers.

"SB Energy's strategic partnership with OpenAI accelerates our delivery of advanced AI data center campuses and associated energy infrastructure at the scale required to advance Stargate and secure America's AI future," said SB Energy co-CEO Rich Hossfeld.

The project also addresses other environmental concerns, with the data center designed to minimize water usage—a critical factor in water-stressed regions and a growing concern for the sustainability of hyperscale computing.

Building an AI Behemoth, Brick by Brick

Executing such a monumental construction plan requires specialized expertise in speed, cost discipline, and logistics. To bolster its capabilities, SB Energy recently acquired Studio 151, a firm specializing in data center construction management, procurement, and engineering. The acquisition integrates a team with a track record spanning 20 data center campuses and one that has already been working with OpenAI and SB Energy on multiple projects.

"Bringing these capabilities in-house strengthens our ability to deliver repeatable campuses at scale and with certainty," noted SB Energy co-CEO Abhijeet Sathe, highlighting the importance of execution discipline in such a complex endeavor.

This vertical integration—combining OpenAI’s first-party data center designs with SB Energy’s expertise in construction and integrated energy delivery—is intended to create a new, faster, and more efficient model for building AI infrastructure. The partnership is further solidified by a previous $800 million investment in SB Energy from Ares Infrastructure Opportunities funds in 2025, demonstrating strong financial backing for the company’s expansion.

For communities like Milam County, the project promises a significant economic infusion, with thousands of construction jobs and further investment in workforce development. However, it also places immense pressure on local infrastructure, particularly the Electric Reliability Council of Texas (ERCOT) grid, which must now accommodate the unprecedented, round-the-clock power demands of an AI supercomputer.