TLDRs;

  • Oracle’s stock experienced a slight decline following the cancellation of plans to expand a major Texas AI data center in collaboration with OpenAI.
  • The Abilene campus continues to be operational, yet negotiations regarding financing and shifting infrastructure requirements put a stop to the proposed expansion phase.
  • It is reported that Meta Platforms is contemplating leasing the unused capacity as competition for AI infrastructure escalates on a global scale.
  • Nvidia has assumed a strategic role in the project, advocating for its chips to drive the large-scale facility.

Shares of Oracle (ORCL) inched downward after reports that the company and OpenAI had given up on plans to expand a massive artificial intelligence data center situated in Abilene, Texas. The proposed expansion was part of a larger endeavor to scale AI infrastructure able to meet the rapidly growing computational needs of advanced models.

According to individuals acquainted with the situation, negotiations between the companies hit a snag due to disagreements regarding financing structures and changing projections regarding OpenAI’s computing capacity needs. Consequently, the expansion plan for the facility was called off, although the overall partnership between the two firms remains in place.

In spite of the halted expansion, Oracle is still working on previously agreed infrastructure commitments. The company inked a deal last year to develop around 4.5 gigawatts of capacity for OpenAI, and that project is still ongoing as both companies pursue other AI computing initiatives.

The Abilene site is owned by infrastructure developer Crusoe Energy Systems, which has been shifting from cryptocurrency mining operations to constructing large-scale AI data centers.

Massive Stargate Infrastructure Vision

The site is widely regarded as a flagship location associated with the ambitious Stargate AI infrastructure concept. The initiative initially aimed to build one of the largest computing hubs ever constructed, potentially involving up to five gigawatts of computing power and requiring tens of billions of dollars in funding.

ORCL Stock Card

Financing for the Abilene campus alone is said to have reached approximately $15 billion, making it one of the most costly data center developments ever funded. The project started ramping up operations in 2025, when the first phase of the facility became operational after months of equipment deployment and infrastructure setup.

During that rollout, Oracle started delivering advanced server racks powered by the latest AI chips manufactured by Nvidia. These racks incorporate the company’s next-generation GB200 architecture, which is specifically designed for large-scale AI training clusters.

However, while the facility remains in operation and expands gradually, the canceled expansion underscores the complexity of constructing massive AI infrastructure at the gigawatt scale. Industry observers point out that infrastructure companies entering the AI race frequently encounter engineering, financial, and logistical challenges that can delay development timelines.

Nvidia’s Growing Strategic Influence

One of the most prominent aspects of the situation is the changing role of Nvidia within the AI infrastructure ecosystem. Traditionally recognized as the leading supplier of GPUs used in AI training, Nvidia has increasingly taken a more active part in shaping large infrastructure projects.

Reports suggest that Nvidia deposited $150 million with Crusoe related to the Abilene site. Besides supplying hardware, the company has also assisted in connecting potential partners who could make use of the facility’s computing capacity.

Specifically, Nvidia is said to have encouraged interest from Meta Platforms, which is assessing the possibility of leasing part of the data center’s infrastructure. Such a move could ensure that Nvidia’s chips continue to be the primary computing engines driving the site rather than hardware from competitors such as Advanced Micro Devices.

This kind of strategic involvement points to a shift in the AI supply chain, where hardware companies are no longer just vendors but also enablers of partnerships between infrastructure providers and AI developers.