Stargate

What Is the Stargate Project?

Stargate is an American multinational joint venture announced in January 2025 that aims to invest up to $500 billion in AI data center infrastructure in the United States by 2029. The venture is led by OpenAI (operational responsibility) and SoftBank (financial responsibility), with Oracle and Abu Dhabi-based investment firm MGX as equity partners. Key technology partners include NVIDIA, Microsoft, and Arm. Stargate represents the single largest committed investment in AI compute infrastructure in history, reflecting a belief among its backers that the scaling of AI model training and inference will continue to demand exponentially more compute for years to come.

Scale, Architecture, and Hardware

Stargate launched with an initial $100 billion commitment and plans to scale to $500 billion across multiple phases. The flagship facility in Abilene, Texas began operations in September 2025 with two buildings, expanding to a planned eight buildings housing over 450,000 NVIDIA GB200 GPUs. As of early 2026, OpenAI, Oracle, and SoftBank announced five additional U.S. sites—in Shackelford County, Texas; Doña Ana County, New Mexico; Lordstown, Ohio; Milam County, Texas; and an undisclosed Midwestern location—bringing planned capacity to nearly 7 gigawatts and over $400 billion in committed investment. The project's total target is roughly 10 GW of AI data center capacity, a scale that rivals the total electricity consumption of small nations and underscores the enormous energy demands of AI. OpenAI is also developing a custom AI chip codenamed "Titan" in collaboration with Broadcom, to be fabricated on TSMC's 3nm process node, with mass production targeted for late 2026—a move that could reduce dependence on NVIDIA GPUs over time.

Financing and Corporate Structure

The financial architecture of Stargate is itself unprecedented. SoftBank secured a $40 billion bridge loan in March 2026 to deepen its commitment, while SB Energy (a SoftBank subsidiary) received $500 million each from OpenAI and SoftBank to build and operate next-generation power infrastructure for the data centers. Microsoft's role is notable: rather than contributing capital expenditures or compute capacity, it serves as a "technology partner" that may rent computing power from Stargate facilities. The venture has not been without friction—reports surfaced in 2025 and 2026 of disputes between OpenAI, Oracle, and SoftBank over governance, cost allocation, and operational control, temporarily stalling progress before being resolved. The project's corporate structure as Stargate LLC reflects the complexity of aligning four major institutional backers with different strategic priorities across cloud computing, semiconductor fabrication, and AI research.

Global Expansion and Sovereign AI

While the U.S. remains the primary focus, Stargate has expanded internationally. In 2025, OpenAI announced Stargate-branded data center partnerships in the UAE (with G42, Oracle, SoftBank, NVIDIA, and Cisco), Argentina (with Sur Energy, up to $25 billion and 500 MW), Norway, and the United Kingdom. These international deployments align with the growing movement toward sovereign AI infrastructure, where nations seek domestic compute capacity to ensure AI capabilities are not entirely dependent on foreign cloud providers. The UAE campus is expected to open in 2026, making it one of the first non-U.S. Stargate facilities to come online.

Strategic Implications for the Agentic Economy

Stargate is not merely a data center buildout—it is a bet on the future shape of the agentic economy. The project's massive scale is designed to support the next generation of foundation models, agentic AI systems, and inference-heavy workloads that power autonomous agents operating across industries. As AI shifts from training-dominated to inference-dominated compute demand, infrastructure at Stargate's scale becomes the substrate for an economy where AI agents transact, reason, and act on behalf of humans and organizations. The project also intensifies competition in the AI accelerator and high-bandwidth memory supply chains, placing enormous pressure on TSMC, SK Hynix, and other upstream semiconductor manufacturers. Whether Stargate achieves its full $500 billion ambition will depend on sustained demand for frontier AI compute, the resolution of energy infrastructure bottlenecks, and the ability of its partners to maintain alignment over a multi-year, multi-hundred-billion-dollar deployment.

  • AI Data Centers — the physical infrastructure Stargate is building at unprecedented scale
  • AI Energy Consumption — the enormous power demands driving Stargate's energy partnerships
  • GPU Computing — the NVIDIA GB200 GPUs at the core of Stargate's compute architecture
  • Sovereign AI Infrastructure — the geopolitical dimension of Stargate's international expansion
  • AI Inference — the inference-heavy workloads Stargate is designed to support at scale
  • AI Accelerators — custom chips like OpenAI's Titan that may supplement GPU infrastructure
  • Semiconductor Fabrication — TSMC's role in manufacturing Stargate's custom silicon
  • Agentic Economy — the economic paradigm that Stargate's compute capacity enables
  • Cloud Computing — how Stargate relates to and disrupts traditional cloud infrastructure
  • Liquid Cooling — thermal management technologies required at Stargate's density and scale

Further Reading