OpenAI's $500 billion Stargate project hits a snag: progress stalls, team shortages

Once regarded as a milestone in U.S. artificial intelligence infrastructure, the $500 billion “Stargate” project has now stalled. The ambitious blueprint to build a 10GW ultra-large data center cluster has encountered obstacles due to internal coordination and funding issues. Facing rising computing demands and rapid expansion by competitors, OpenAI is adjusting its strategy, shifting toward multi-party collaborations to fill the gap.

The “Stargate” blueprint remains unimplemented: team formation incomplete, facilities not yet under construction

The Stargate project was announced in early 2025 by U.S. President Trump in partnership with industry giants such as OpenAI, Oracle, and SoftBank, aiming to establish a 10GW AI data center in the United States as a core infrastructure to support next-generation AI models. However, over a year has passed, and actual progress has been limited.

According to The Information, the joint venture has yet to establish a dedicated full team or commence large-scale facility development. Disagreements among partners over leadership structure, financial responsibilities, and decision-making authority have stalled overall progress. What was once seen as a symbol of industry integration and capital mobilization remains at the planning stage.

OpenAI’s funding and financial pressures emerge: in-house plans put on hold

As progress in the joint venture stalls, OpenAI considered building and owning the data centers independently, evaluating large-scale debt financing. However, due to the company’s ongoing high investments and the lack of a fully mature long-term profit model, some financial institutions remain cautious about providing substantial loans.

Ultimately, the in-house plan was suspended. The goal to reach 10GW capacity by the end of 2025 was not achieved; current capacity is approximately 7.5GW. As estimates for computing demand are revised upward, future spending projections through 2030 have increased from about $450 billion to $665 billion, adding financial pressure.

(OpenAI’s annual revenue surpasses $20 billion, but The New York Times remains skeptical: bankruptcy possible by 2027)

From “ownership” to “control”: OpenAI deepens cloud and chip collaborations

Facing difficulties in advancing joint ventures and self-build plans, OpenAI is adopting more flexible cooperation models. In mid-2025, the company signed an agreement with Oracle to jointly develop approximately 4.5GW of data center capacity, sharing construction and cost overrun risks to reduce individual capital burdens.

Additionally, OpenAI has signed extra compute contracts with multiple cloud service providers to address short-term resource shortages. The company is also diversifying chip supply sources to reduce reliance on a single GPU vendor, shifting toward a multi-hardware partnership strategy.

Chief Financial Officer Sarah Friar stated that the current priority is to ensure long-term compute access and design autonomy while avoiding excessive balance sheet expansion. Building data centers remains a long-term direction but is not the immediate priority.

OpenAI shifts toward splitting collaborations: compute infrastructure becomes a key battleground

The delay of the Stargate project coincides with a rapid escalation in global AI competition. Many rivals are expanding their compute capacity, strengthening integration of data centers and chip resources, shifting industry focus from model capabilities to energy and hardware acquisition speed.

Meanwhile, the surge in power demand from large data centers has raised concerns among grid operators about supply stability. AI development is no longer just a software and algorithm race; it also involves multiple challenges in electricity, land, and capital markets.

The $500 billion Stargate plan originally symbolized a comprehensive upgrade of AI infrastructure, but due to practical constraints and complex collaborations, progress has been slower than expected. OpenAI is restructuring its compute strategy, moving from centralized joint ventures to dispersed partnerships to ensure that technological development is not hindered by infrastructure bottlenecks.

This article, “OpenAI’s $500 Billion Stargate Plan Stalls: Progress Delays, Team Shortages,” first appeared on Chain News ABMedia.