Oracle and OpenAI Cancel Major AI Data Center

The Thesis Journal

Oracle and OpenAI Cancel Major AI Data Center Expansion in Texas

Adrian Schimpf • March 6th, 2026

Artificial Intelligence "Space" race

Oracle and OpenAI Cancel Major AI Data Center Expansion in Texas

Oracle and OpenAI have abandoned plans to expand a major artificial intelligence data center in Texas after prolonged negotiations over financing and changing infrastructure needs stalled the project.

The cancelled expansion involves a large development at a data center campus in Abilene, Texas, part of the high profile Stargate AI infrastructure initiative announced last year. While construction at the existing site continues, the two companies have decided not to move forward with a proposed expansion that would have significantly increased the facility’s computing capacity.

The development highlights the growing complexity and financial challenges surrounding the massive infrastructure required to support modern AI systems.

Meta Emerges as Potential Replacement Tenant

The collapse of the Oracle OpenAI expansion plans has created an opportunity for another major technology company to step in.

According to people familiar with the discussions, Meta is now considering leasing the expansion space from the site’s developer, Crusoe. The social media giant has been aggressively expanding its AI infrastructure to support new machine learning products and services.

Nvidia, the leading manufacturer of AI chips, has reportedly played a role in facilitating discussions between Meta and the developer. Sources say Nvidia has already provided a $150 million deposit to help secure the expansion project while a new tenant is finalized.

The move reflects Nvidia’s interest in ensuring that its AI hardware remains the primary technology powering the new facility.

The Stargate Project Remains One of the Largest AI Infrastructure Builds

The Abilene data center campus is part of the Stargate project, a large scale AI infrastructure initiative announced in Washington last year with support from government and industry leaders.

The site spans roughly 1,000 acres and is designed to host one of the largest concentrations of AI computing capacity in the world.

Even without the cancelled expansion, the project remains massive. Oracle previously agreed to develop approximately 4.5 gigawatts of data center capacity for OpenAI across multiple locations.

For perspective, a single gigawatt of computing infrastructure requires enormous electrical capacity, comparable to the power output of a nuclear reactor and capable of supplying electricity to hundreds of thousands of homes.

The Abilene campus currently operates at around 1.2 gigawatts of capacity, and the proposed expansion would have increased that figure to roughly 2 gigawatts.

Financing and Demand Uncertainty Complicate Expansion

Several factors contributed to the breakdown of the expansion negotiations.

Building large AI data centers requires billions of dollars in investment and coordination between developers, energy providers, hardware suppliers, and cloud service companies.

Sources say financing arrangements became increasingly complex as the project progressed. In addition, OpenAI’s rapidly evolving demand forecasts made it difficult to finalize long term infrastructure commitments.

Artificial intelligence development is advancing quickly, and companies often adjust computing needs as new models and training techniques emerge.

These uncertainties can complicate long term planning for data center operators, particularly when projects require massive upfront capital investments.

Technical Issues Also Strained Relationships

Operational challenges at the site may have also contributed to the decision to halt the expansion.

Earlier this year, several buildings at the Abilene campus reportedly experienced outages lasting multiple days after winter weather disrupted parts of the liquid cooling systems used to regulate high performance AI servers.

Modern AI data centers generate enormous amounts of heat due to the intense computing power required to train and run large language models and other advanced systems. Maintaining reliable cooling infrastructure is therefore critical for maintaining operations.

Although both Oracle and Crusoe have stated that their partnership remains strong, the incident added another layer of complexity to an already challenging expansion effort.

Massive AI Infrastructure Spending Continues

Despite the setback in Texas, the race to build AI infrastructure remains intense across the technology sector.

Major technology companies are committing enormous resources to build data centers capable of supporting increasingly powerful artificial intelligence models.

Meta, for example, has projected capital expenditures of up to $135 billion in 2026, much of which will be directed toward AI related infrastructure. The company is currently developing large data centers in several U.S. states, including Louisiana and Indiana.

In addition, Meta recently signed an agreement to deploy roughly 6 gigawatts of AI hardware supplied by AMD, highlighting the scale of the ongoing infrastructure race.

Meanwhile, companies such as Nvidia, AMD, and specialized cloud providers continue to compete to supply the processors and systems powering the next generation of AI computing.

Markets React to Shifting AI Infrastructure Plans

The news also affected several companies tied to the artificial intelligence infrastructure ecosystem.

Shares of Oracle slipped modestly following the report, while other companies involved in AI computing infrastructure, including Nvidia and AMD, also saw declines during trading.

Investors remain highly focused on developments in the AI sector, which has become one of the primary drivers of technology investment and market valuation in recent years.

Any changes in major infrastructure projects can therefore influence sentiment across the broader technology industry.

Overall

The decision by Oracle and OpenAI to cancel their planned expansion in Texas illustrates both the enormous scale and the growing complexity of building artificial intelligence infrastructure.

While demand for AI computing power continues to rise rapidly, the financial, technical, and logistical challenges involved in constructing massive data centers remain significant.

As companies race to build the next generation of AI systems, partnerships between technology firms, infrastructure developers, and semiconductor manufacturers will play a crucial role in determining how quickly the global AI ecosystem can expand.

For now, the Abilene site remains one of the most ambitious AI infrastructure projects underway, even as the search for a new tenant for its planned expansion continues.

Data & Methodology:

BEA GOV

Bloomberg

Finance Yahoo

Brody Ford

Ed Ludlow

Dina Bass

Aquire for direct sources

Monthly research brief. No commentary. No noise.

Email *

We respect your inbox. Unsubscribe anytime.

Back to Research

The Thesis Journal

Oracle and OpenAI Cancel Major AI Data Center Expansion in Texas

Oracle and OpenAI have abandoned plans to expand a major artificial intelligence data center in Texas after prolonged negotiations over financing and changing infrastructure needs stalled the project.

The development highlights the growing complexity and financial challenges surrounding the massive infrastructure required to support modern AI systems.

Meta Emerges as Potential Replacement Tenant

The collapse of the Oracle OpenAI expansion plans has created an opportunity for another major technology company to step in.

According to people familiar with the discussions, Meta is now considering leasing the expansion space from the site’s developer, Crusoe. The social media giant has been aggressively expanding its AI infrastructure to support new machine learning products and services.

Nvidia, the leading manufacturer of AI chips, has reportedly played a role in facilitating discussions between Meta and the developer. Sources say Nvidia has already provided a $150 million deposit to help secure the expansion project while a new tenant is finalized.

The move reflects Nvidia’s interest in ensuring that its AI hardware remains the primary technology powering the new facility.

The Stargate Project Remains One of the Largest AI Infrastructure Builds

The Abilene data center campus is part of the Stargate project, a large scale AI infrastructure initiative announced in Washington last year with support from government and industry leaders.

The site spans roughly 1,000 acres and is designed to host one of the largest concentrations of AI computing capacity in the world.

Even without the cancelled expansion, the project remains massive. Oracle previously agreed to develop approximately 4.5 gigawatts of data center capacity for OpenAI across multiple locations.

For perspective, a single gigawatt of computing infrastructure requires enormous electrical capacity, comparable to the power output of a nuclear reactor and capable of supplying electricity to hundreds of thousands of homes.

The Abilene campus currently operates at around 1.2 gigawatts of capacity, and the proposed expansion would have increased that figure to roughly 2 gigawatts.

Financing and Demand Uncertainty Complicate Expansion

Several factors contributed to the breakdown of the expansion negotiations.

Building large AI data centers requires billions of dollars in investment and coordination between developers, energy providers, hardware suppliers, and cloud service companies.

Sources say financing arrangements became increasingly complex as the project progressed. In addition, OpenAI’s rapidly evolving demand forecasts made it difficult to finalize long term infrastructure commitments.

Artificial intelligence development is advancing quickly, and companies often adjust computing needs as new models and training techniques emerge.

These uncertainties can complicate long term planning for data center operators, particularly when projects require massive upfront capital investments.

Technical Issues Also Strained Relationships

Operational challenges at the site may have also contributed to the decision to halt the expansion.

Earlier this year, several buildings at the Abilene campus reportedly experienced outages lasting multiple days after winter weather disrupted parts of the liquid cooling systems used to regulate high performance AI servers.

Modern AI data centers generate enormous amounts of heat due to the intense computing power required to train and run large language models and other advanced systems. Maintaining reliable cooling infrastructure is therefore critical for maintaining operations.

Although both Oracle and Crusoe have stated that their partnership remains strong, the incident added another layer of complexity to an already challenging expansion effort.

Massive AI Infrastructure Spending Continues

Despite the setback in Texas, the race to build AI infrastructure remains intense across the technology sector.

Major technology companies are committing enormous resources to build data centers capable of supporting increasingly powerful artificial intelligence models.

Meta, for example, has projected capital expenditures of up to $135 billion in 2026, much of which will be directed toward AI related infrastructure. The company is currently developing large data centers in several U.S. states, including Louisiana and Indiana.

In addition, Meta recently signed an agreement to deploy roughly 6 gigawatts of AI hardware supplied by AMD, highlighting the scale of the ongoing infrastructure race.

Meanwhile, companies such as Nvidia, AMD, and specialized cloud providers continue to compete to supply the processors and systems powering the next generation of AI computing.

Markets React to Shifting AI Infrastructure Plans

The news also affected several companies tied to the artificial intelligence infrastructure ecosystem.

Shares of Oracle slipped modestly following the report, while other companies involved in AI computing infrastructure, including Nvidia and AMD, also saw declines during trading.

Investors remain highly focused on developments in the AI sector, which has become one of the primary drivers of technology investment and market valuation in recent years.

Any changes in major infrastructure projects can therefore influence sentiment across the broader technology industry.

Overall

The decision by Oracle and OpenAI to cancel their planned expansion in Texas illustrates both the enormous scale and the growing complexity of building artificial intelligence infrastructure.

While demand for AI computing power continues to rise rapidly, the financial, technical, and logistical challenges involved in constructing massive data centers remain significant.

As companies race to build the next generation of AI systems, partnerships between technology firms, infrastructure developers, and semiconductor manufacturers will play a crucial role in determining how quickly the global AI ecosystem can expand.

For now, the Abilene site remains one of the most ambitious AI infrastructure projects underway, even as the search for a new tenant for its planned expansion continues.

Subscribe