Stargate Project Pivots: OpenAI Drives AI Infrastructure, SoftBank Steps Back

A monumental undertaking to construct the world's most powerful AI infrastructure, known as Project Stargate, is moving forward, but with a significant shift in its foundational partnerships. While OpenAI, the project's operational lead, is aggressively pushing ahead with development, its initial key financial partner, SoftBank, appears to be increasingly absent from the unfolding reality of the ambitious venture. This has led to a recalibration of both the project's immediate goals and its long-term funding and strategic structure, raising questions about the future of this half-trillion-dollar effort to secure American leadership in artificial intelligence.

Initially announced in January 2025 as a $500 billion joint venture, Stargate was presented as a collaboration between OpenAI, SoftBank, Oracle, and the Emirati investment firm MGX.[1][2] The project's grand vision is to build a sprawling network of advanced data centers across the United States, providing the immense computational power necessary to train and operate the next generations of artificial intelligence.[3][4] The initiative, which was projected to create over 100,000 jobs, was positioned as a critical step for both the re-industrialization of the U.S. and its national security.[5][3] SoftBank was slated to take on the primary financial responsibility for the massive undertaking, with OpenAI managing the operational aspects.[5] However, recent developments indicate a significant divergence from this original plan, with OpenAI proceeding on major data center deals without SoftBank's direct involvement.[6][7]

The most prominent example of this new direction is the massive 1.2-gigawatt data center currently under construction in Abilene, Texas.[7][8] While labeled as the first leg of the Stargate project, reports indicate it is proceeding without SoftBank, which notably holds the trademark for the "Stargate" name.[6][7] Instead, OpenAI is working closely with Oracle on this facility.[8] Further cementing this shift, OpenAI has reportedly struck a massive deal to lease 4.5 gigawatts of data center capacity from Oracle, a contract valued at approximately $30 billion annually.[9] OpenAI has also entered into a smaller but significant deal with CoreWeave for additional GPU capacity.[6][7] These independent moves by OpenAI align with the initial $100 billion immediate investment target for Stargate but sideline SoftBank from the core infrastructure build-out.[6] Disagreements over key partnership terms, including the locations of data center sites and choices of energy suppliers, are reported to be the primary cause of the friction between OpenAI and SoftBank.[7][10]

Despite the apparent friction with SoftBank, the broader Stargate initiative continues to evolve. The initial ambitious plan to immediately deploy $100 billion has been scaled back, with the "official" joint project now targeting a more modest goal of building a small data center in Ohio by the end of the year.[6][7][8][10] In a joint statement, both OpenAI and SoftBank have affirmed their commitment to the project, stating they are moving with "urgency on site assessments" and advancing projects in multiple states.[10] Meanwhile, OpenAI is not limiting its ambitions to the United States. The company has launched "OpenAI for Countries," an initiative to partner with governments worldwide to build localized AI data centers, tailoring services to specific languages and cultures.[4] This global expansion includes plans for a "UAE Stargate" in the United Arab Emirates, expected to open in 2026.[1]

The immense scale of Stargate brings with it unprecedented challenges, most notably the colossal power requirements. The proposed data centers are incredibly energy-intensive, with some estimates suggesting a single campus could require up to 5 gigawatts of power, equivalent to the output of a large nuclear power station.[11][12] This has led to discussions around using alternative energy sources, including nuclear power, to meet the demand.[12][13] The project also faces the hurdle of sourcing millions of specialized AI chips, a market already facing significant supply constraints.[13] Amidst these technical and logistical challenges, the role of Microsoft, OpenAI's long-standing and largest investor, remains pivotal. While the Stargate announcement altered Microsoft's exclusive cloud provider status to a "right of first refusal" on new capacity, the two companies' partnership remains deeply intertwined.[14][15][16] Microsoft continues to be a major investor and provider of Azure cloud infrastructure for OpenAI's current products and model training, and the Stargate project is seen as complementary to their existing collaboration.[5][14]

In conclusion, the Stargate project, envisioned as a transformative leap in AI infrastructure, is navigating a complex and evolving landscape. While the initial partnership structure with SoftBank as the lead financial backer appears to have fractured, OpenAI is aggressively forging ahead, securing massive data center capacity through new and expanded alliances, particularly with Oracle. The project's initial timeline and scope have been adjusted, but the underlying ambition to build unparalleled computational power for the future of AI remains. The success of this monumental undertaking will ultimately depend on overcoming immense logistical hurdles, securing sustainable power sources, and navigating the intricate web of partnerships that now define the race to build the foundation for artificial general intelligence.