News
EveryDatacenter
Stargate Abilene

Stargate Abilene

AI data centre campus in Texas operated by Oracle for OpenAI

Stargate Abilene is an artificial intelligence data centre campus in Texas, developed by Crusoe on Lancium’s Clean Campus. Operated by Oracle for OpenAI, it launched with two buildings in September 2025 and aims for 1.2 gigawatts of capacity.

Share on WhatsApp

  • Operated by Oracle for OpenAI use
  • Total capacity planned at 1.2 gigawatts
  • Two buildings launched in September 2025
  • Expansion to 2GW cancelled due to financing
  • Uses liquid cooling for GPU hardware

What it is

Stargate Abilene is a large-scale artificial intelligence data centre campus located in Abilene, Texas. It forms part of the broader Stargate joint venture initiative. The facility is developed by Crusoe Energy Systems on land provided by Lancium’s Clean Campus. Oracle operates the site specifically to support the computational needs of OpenAI. The project represents a significant investment in dedicated AI infrastructure within the United States, focusing on high-density computing capabilities required for advanced machine learning models.

How it came about

The project was announced as part of a major joint venture involving substantial projected spending. Two initial data centre buildings became operational in September 2025. Plans originally included six additional buildings to reach a total capacity of approximately 1.2 gigawatts, with further expansion to 2 gigawatts intended. However, reports indicate that these expansion plans were cancelled due to financing difficulties and shifting demand forecasts from OpenAI. Changes in leadership at OpenAI and strategic pivots towards cloud contracts also influenced the decision to halt further growth at this specific location.

The building and the halls

The campus is situated on Lancium’s Clean Campus in Texas. The initial phase consisted of two data centre buildings that went live in late 2025. The design supports high-performance computing hardware, including Nvidia GPUs. The site was prepared to host up to eight buildings in total during its initial planning stages. The infrastructure is built to accommodate dense server racks necessary for AI training workloads, with specific attention paid to the physical layout required for efficient cooling and power distribution.

Power and cooling

The facility targets a total IT capacity of around 1.2 gigawatts upon completion of the initial eight-building phase. Power is supplied through the Lancium Clean Campus infrastructure. The site utilizes liquid cooling systems to manage the heat generated by high-performance graphics processing units. Reports suggest that power readiness issues delayed some deployments, influencing decisions about which hardware generations would be installed. The energy demands are significant, reflecting the intensive nature of large-scale AI model training and inference tasks.

What runs inside

OpenAI is the primary tenant, using the compute resources for its artificial intelligence operations. Oracle manages the day-to-day operations and infrastructure maintenance. The site is designed to host Nvidia Blackwell GPUs, although delays led to considerations for newer chip architectures. The capacity is intended for training and running large language models and other AI applications. There were discussions about Meta potentially leasing unused capacity, facilitated by Nvidia, but no final agreement was reported in the available information.

The place

Abilene, Texas, was selected for its available land and power infrastructure through the Lancium Clean Campus. The rural setting provides space for large-scale industrial facilities. The location is part of a broader trend of developing AI infrastructure in regions with accessible energy resources. The site serves as a flagship location for the Stargate initiative in the United States, though future expansions have been redirected to other states such as Wisconsin.

Environment and impact

The project faced challenges related to financing and operational reliability. A multi-day outage caused by winter weather affecting liquid cooling equipment reportedly strained relationships between partners. The cancellation of expansion plans reflects broader financial pressures on the involved companies, including debt accumulation and cost-cutting measures. Local economic impacts include construction activity, but the reduced scope may limit long-term job creation compared to initial projections. Environmental concerns regarding energy consumption are managed through the Clean Campus framework.

Network and links

The focus of the facility is on internal compute capacity for AI workloads rather than public interconnection. Connectivity is likely established to support data transfer between the campus and OpenAI’s other infrastructure, but technical network specifications remain undisclosed in the available reports.

Questions people ask about Stargate Abilene

Who operates the Stargate Abilene data centre?

Oracle operates the facility. It was developed by Crusoe on Lancium’s Clean Campus land. The primary tenant is OpenAI, which uses the compute resources for its artificial intelligence projects.

What is the power capacity of the site?

The initial phase aims for a total capacity of around 1.2 gigawatts across eight buildings. Plans to expand further to 2 gigawatts were cancelled due to financing challenges and changing demand forecasts.

Why was the expansion of the Abilene site cancelled?

Financing difficulties and OpenAI’s shifting view of the Stargate project led to the cancellation. Additionally, power readiness delays meant newer chips might be deployed elsewhere, reducing the need for immediate expansion at this location.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Stargate Abilene yet, so the facts on this page come from the sources listed above: openai.com/index/stargate-community/, datacenterdynamics.com/en/news/oracleopenai-drop-plans-to-expand-flags. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryDatacenter wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

    Every Monday: the data centre of the day, new pages on EveryDatacenter, and the conferences and anniversaries of the week ahead.