IA 360
Current Affairs

Stargate: OpenAI, SoftBank and Oracle pledge $500bn for data centers

Unveiled at the White House the day after Trump's inauguration, the Stargate project commits $500 billion over four years to AI infrastructure, with $100 billion deployed immediately.

Admin IA360 6 min read Leer en español
Stargate: OpenAI, SoftBank and Oracle pledge $500bn for data centers

OpenAI has announced the creation of a new company, The Stargate Project, which will invest $500 billion in AI infrastructure over the next four years. The figure makes it the largest compute commitment announced to date, and it wasn't unveiled in a corporate auditorium but at the White House, one day after Donald Trump's inauguration.

Of that total, $100 billion will be deployed "immediately," according to OpenAI. The data centers built under the project will be for OpenAI's exclusive use, expanding its compute capacity for generative AI.

Who's putting up the money, and who's in charge

The equity investors in Stargate are SoftBank, OpenAI, Oracle and MGX, Abu Dhabi's sovereign fund. The division of labor is clear: SoftBank has financial responsibility, OpenAI has operational responsibility. SoftBank founder Masayoshi Son will chair the new company.

On the technology side, the initial partners will be Arm (owned by SoftBank), Microsoft, Nvidia, Oracle and OpenAI itself. The company said Oracle, Nvidia and OpenAI will work closely together to build and operate the computing system, drawing on a relationship with Nvidia dating back to 2016 and a more recent partnership with Oracle.

OpenAI was also careful to stress that Stargate doesn't mean a break with Microsoft: "This also builds on the existing OpenAI partnership with Microsoft. OpenAI will continue to increase its consumption of Azure as OpenAI continues its work with Microsoft with this additional compute to train leading models and deliver great products and services," the company said. That's a meaningful detail: for months, OpenAI's reliance on Microsoft's infrastructure has been a recurring talking point, and this announcement diversifies that base without shutting the door on Azure.

Construction is already underway in Texas

Building has already begun, starting in Texas. At the White House press conference, Oracle founder Larry Ellison said 10 data centers were currently under construction at the Texas site. OpenAI added that it is "evaluating potential sites across the country for more campuses as we finalize definitive agreements."

The specifics of what OpenAI plans to build haven't been shared yet. What is known is that last year the company handed the White House a document outlining plans for 5GW data centers — which would make them the largest facilities in the world. To put that in perspective: 5GW is roughly the amount of electricity consumed by several major cities, and powering a single complex of that size requires generation capacity that simply doesn't exist yet.

That's why OpenAI has put out an open call to "connect with firms across the built data center infrastructure landscape, from power and land to construction to equipment, and everything in between." It's an implicit acknowledgment that no single tech company can pull off a project like this alone: it requires builders, landowners, equipment manufacturers and, above all, electricity.

Energy as the bottleneck — and now a matter of state

The most telling part of the announcement wasn't a number, but Trump's presence and his explicit offer of government intervention. "What we want to do is we want to keep it in this country. China is a competitor, others are competitors. We want to be in this country, and we're making it available," the president said.

He went further on the power problem: "I'm gonna help a lot through emergency declarations, because we have an emergency, we have to get this stuff built. So they have to produce a lot of electricity. And we'll make it possible for them to get this production done easily, at their own plants if they want."

That line sums up the industry's new equation. The AI bottleneck is no longer just about getting hold of Nvidia chips; it's about securing enough energy to power them, and enough land and permits to build at this scale. A sitting president offering emergency declarations and letting companies generate their own electricity shows just how far AI data centers have traveled — from a private infrastructure matter to a geopolitical objective.

What half a trillion dollars actually means

Sam Altman, OpenAI's CEO, summed up the ambition: "I think this will be the most important project of this era." Hyperbole is part of the script at these announcements, but the number forces you to take it seriously. $500 billion is an order of magnitude until recently reserved for government programs, not a consortium of private companies chasing a single goal.

Still, the announcement is worth reading with some caution. A four-year investment commitment isn't money already spent — it's a stated intention that depends on finalizing "definitive agreements," as OpenAI itself acknowledges, and on solving the energy, land and construction problems the company admits it can't tackle alone. Of the $500 billion, only $100 billion carries the "immediate" label.

There's also a story here about the balance of power in the AI ecosystem. With SoftBank holding financial responsibility and chairing the company, and with Oracle and Nvidia as top-tier operating partners, OpenAI is reducing its dependence on any single infrastructure provider. The presence of MGX, Abu Dhabi's sovereign fund, among the equity investors also confirms that Middle Eastern sovereign capital is now a structural part of financing AI at scale.

What's still unclear

The announcement leaves more questions than concrete answers. There's no detail on the specifications of the data centers beyond the 5GW document previously shared with the White House, no detailed timeline, and no confirmed locations beyond Texas. It's also unclear how Trump's electricity promises will actually materialize, or what form those emergency declarations will take.

What is clear is the framing: the AI race is now being fought in concrete, copper and megawatts, with the biggest tech companies, a sovereign fund and the White House sitting at the same table on the second day of a new administration. The coming months — as those definitive agreements get signed and construction in Texas progresses — will show how much of the announced half-trillion dollars turns into actual data centers.

Share this article

Related articles

Apple v. OpenAI, week 2: what's in the case file and what may come this week
Current Affairs

Apple v. OpenAI, week 2: what's in the case file and what may come this week

After Friday's headlines, the docket in case 5:26-cv-07078 spent its first weekend unchanged: the twelve entries are still the ones from day one. We go through them entry by entry — fees, summonses, an eight-lawyer team — and lay out the four procedural milestones that may land this week: the scheduling order, the first proofs of service, the injunction motion Apple has announced, and the July 24 decision on who judges the case.

3 min
Apple sues OpenAI: the partnership that brought ChatGPT to the iPhone ends up in court
Current Affairs

Apple sues OpenAI: the partnership that brought ChatGPT to the iPhone ends up in court

Apple accuses OpenAI, its hardware chief Tang Tan, engineer Chang Liu and io Products of misappropriating iPhone trade secrets to accelerate its first consumer device. The federal lawsuit, filed in San Jose, brings six claims and seeks injunctions and damages; OpenAI replies that it has "no interest in other companies' trade secrets." The case will test where the line sits between hiring talent and taking confidential know-how in the middle of the AI hardware race.

5 min

This website uses cookies to improve the browsing experience. Cookie policy.