One year after the White House press conference, there has still been no substantive progress; OpenAI's $500 billion "Stargate" plan is stuck in a stalemate.

```

The $500 billion "Stargate" AI data center plan, which Trump announced with great fanfare at the White House, has come to a standstill more than a year after its launch. OpenAI has been forced to abandon its ambition to build its own data centers and instead is relying on partnerships with cloud service providers to meet its urgent computing power needs.

According to insiders, the joint venture formed by OpenAI, Oracle, and SoftBank has yet to assign any personnel, nor has it developed any OpenAI data centers so far. Within weeks after the project announcement, the three parties became divided over key issues such as leadership, role allocation, and cooperation structure, stalling the "Stargate" plan.

Facing a shortage of computing power, OpenAI once planned to build large-scale data center campuses to reduce reliance on cloud service providers. However, when lenders refused to provide multi-billion dollar funding to a business model that remains unproven and burns billions annually, the plan was shelved. Months later, after confronting financial realities, OpenAI returned to its "Stargate" partners seeking to move the project forward.

This setback has put pressure on OpenAI’s finances. Last year the company's gross margin was lower than expected because it had to buy expensive computing power at the last minute. OpenAI has also raised its compute spending forecast for 2030 from $450 billion to $665 billion. Meanwhile, Google and Anthropic have grown into more threatening competitors over the past year.

Project Stalled by Three-Party Disputes

In January 2025, when President Trump announced the "Stargate" project, the three companies responsible for developing the project promised to advance at lightning speed, almost immediately investing $100 billion to build 10 gigawatts of computing capacity.

But, according to three insiders involved in the plan, within weeks of the announcement, "Stargate" ran into difficulties, lacking leadership and coordination. OpenAI, Oracle, and SoftBank were unable to agree on matters such as responsibility assignments and how the partnership should be structured.

Some executives proposed splitting "Stargate" into an independent entity that would build the facilities and lease them back to OpenAI, with some OpenAI employees possibly transferring to that entity. Executives also discussed whether "Stargate" could serve as a funding vehicle to raise capital for chips and infrastructure. However, these ideas ultimately came to nothing.

Last year, OpenAI failed to achieve its established goal of securing a commitment to 10 gigawatts of capacity from SoftBank and Oracle over the next three years. During the process of finalizing the "Stargate" deal, OpenAI instead signed compute agreements with other suppliers such as Amazon Web Services and Google Cloud to meet its computing needs.

Self-Build Plans Hit Financial Reality

After the White House announced "Stargate", OpenAI's senior leadership began to explore how to build their own data centers. Company employees visited sites around the country, searching for potential locations capable of supporting 800 megawatts to 1.2 gigawatts campuses, prioritizing places that could provide sufficient power in 2026 and 2027.

But as OpenAI assessed sites and calculated costs, executives realized the company would have to pay a steep price to secure needed financing. If a better-credit tenant like Oracle signed the lease and then subleased capacity to OpenAI, costs would drop significantly—this is the model OpenAI had used in other data center deals.

It was not until spring that OpenAI’s "Stargate" team met with Oracle’s executive team, including then-head of Oracle's cloud business and current co-CEO Clay Magouyrk. The two parties agreed on a new vision: Oracle would directly reach a major data center deal with OpenAI covering 4.5 gigawatts of power.

According to meeting participants, this broader architecture allowed both companies to spread risk across multiple campuses, rather than signing one-off projects. Within about two months, OpenAI paused its own data center building plans. The first campus the company surveyed, such as the Vantage Data Centers project in Wisconsin, was assigned to Oracle for development.

Compromise: Control Without Ownership

At the end of July 2024, Oracle and OpenAI announced a deal to develop 4.5 gigawatts of data center capacity at multiple locations in the US. According to two insiders, both sides agreed to share some economic risk for this large-scale project in this rare arrangement. This means if delays or cost overruns occur, both will share the cost; if costs come in under budget, savings are also shared.

After the announcement, OpenAI employees began preparing to revive ambitions of building their own data centers. One project was particularly important in its plans: a 1-gigawatt campus in Milam County, Texas, between Austin and Waco.

At the start of 2025, OpenAI had made plans for the project, but Oracle negotiations caused the project to be shelved and construction paused, with OpenAI even losing the general contractor due to delays. Eventually, the company did not go ahead with the project on its own but instead established a partnership with another "Stargate" partner, SoftBank, and publicly stated it would jointly develop the site with SoftBank Energy.

However, according to two participants in discussions, the two parties clashed over control. OpenAI wanted the Texas campus to be its first self-built data center, while SoftBank wanted to develop and own the project.

From September to October, the "Stargate" team repeatedly traveled to Japan to resolve issues with SoftBank's Masayoshi Son. In one meeting in a large conference room one floor below Son’s main office, negotiations lasted hours. OpenAI employees repeatedly went to a nearby 7-Eleven to buy soda and milk tea, and at the end of the meeting the table was filled with empty bottles.

In the end, the two parties reached a compromise: OpenAI would sign a long-term lease and control the facility's design, with SoftBank Energy responsible for developing and owning the project. Within OpenAI, this moment signaled that its ambition to immediately build its own data center was no longer a near-term goal.

The two companies broke ground in October. Although this is not considered OpenAI’s first self-built data center, employees breathed a sigh of relief, as the deal nearly a year in preparation was done. OpenAI hopes this project can serve as a template for expanding more capacity. The deal grants OpenAI design control without the startup directly investing in or owning the project, with control extending to decisions such as cluster architecture, cooling systems, rack configuration, and power infrastructure.

New Leadership Reshuffles Infrastructure Team

OpenAI CFO Sarah Friar’s comments at last month’s World Economic Forum in Davos illustrated this shift. In an interview with The Information editor-in-chief Jessica Lessin, she said OpenAI is "leveraging our (cloud) partners because that's how you keep the balance sheet light."

"We haven’t fully built ourselves today, because frankly, we have really good partners," she said. "But you know, where will we be three years from now?"

According to two OpenAI employees, the company still plans to build its own data centers, but this is no longer a near-term priority. Instead, the company is using the "Stargate" label to build a series of cloud and infrastructure agreements, some of which grant special rights to design data centers.

The evolution of "Stargate" and OpenAI’s realization that it cannot build and own its own infrastructure for now coincides with leadership changes in its infrastructure organization. Last November, OpenAI hired former Intel CTO and AI head Sachin Katti to lead the infrastructure team, reporting to OpenAI president and co-founder Greg Brockman.

Risk Warning and DisclaimerThe market has risks; investment needs caution. This article does not constitute personalized investment advice, nor does it take into account individual users’ specific investment goals, financial situation or needs. Users should consider whether any opinions, views, or conclusions in this article suit their particular circumstances. Investments based on this are at your own risk. ```