Project Stargate Unveiled: OpenAI’s $500 Billion Bet on an AI-Driven Future

Bottom Line Up Front: Project Stargate is OpenAI’s $500 billion bet on the future of AI infrastructure—a data center of unprecedented scale designed to power the next generation of models like ChatGPT-7. Backed by global investors and tech giants, Stargate reflects a strategic shift toward private-sector-driven AI innovation, with massive energy demands, nuclear power exploration, and a timeline that redefines what’s possible in computational capability. More than a facility, Stargate signals the arrival of AI as national infrastructure.

Project Stargate finally came out of stealth mode last week. News of this project has been spoken of in leaks and behind closed doors now for over a couple of years. What emerged confirms the whispers: Stargate is poised to redefine the limits of computational infrastructure.

Estimated to be one of the largest and most advanced data centers in the world, Stargate spans several hundred acres and is projected to consume up to 5 gigawatts of power—a figure that rivals the energy demands of some small nations. This initiative represents the next leap in OpenAI’s mission to push the boundaries of artificial intelligence. Its sheer scale underscores its importance in training and operating increasingly sophisticated AI models.

How is it being funded? The project is set to cost $500 billion. With an initial build out of $100 billion to get the project going. No funding will come from the federal government; it will all be sourced from private-sector investments. The project is backed by four founding companies:

• OpenAI: U.S.-based AI research organization which will have operational responsibility will provide approximately $20b
• SoftBank: Japan-based investment company which will have financial responsibility will provide approximately $20b
• Oracle: U.S.-based computer technology company which has approximately $11b to invest on its balance sheet
• MGX: Investment firm based in the United Arab Emirates has approximately $100b on its balance sheet to invest in AI.

The project will have the following technology partners:

• Arm: UK-based semiconductor and software design company which will provide hardware and software designs.
• Microsoft: US-based technology and software company which seems like it will provide compute for the runtime of the model in its Azure Data Centers to the tune of $80b
• NVIDIA: US-based graphics processing and AI hardware company which will probide hardware, software, Build and Operational Expertise.
• Oracle: US- US based computer technology company which will provide build and operational expertise
• OpenAI: US- US based AI research organization which will provide build and operational expertise.

What is the federal government’s involvement? Former President Donald Trump announced his intent to intervene when necessary to expedite the project’s progress, citing its significance for maintaining U.S. leadership in artificial intelligence. “What we want to do is, we want to keep it in this country,” Trump stated. “China is a competitor, and others are competitors. We want it to be in this country, and we’re making it available. I’m going to help a lot through emergency declarations because we have an emergency. We have to get this stuff built.”

Additionally, Trump revoked an executive order issued by President Joe Biden that called for regulation and oversight of artificial intelligence. The new directive aimed to eliminate barriers for U.S. companies to develop AI. “This order revokes certain existing AI policies and directives that act as barriers to American AI innovation, clearing a path for the United States to act decisively to retain global leadership in artificial intelligence,” the order stated.

Central to Stargate’s purpose is the development of “ChatGPT 7”, a generational leap forward in AI. To achieve this, Stargate will likely rely on cutting-edge hardware, including Rubin Ultra 200 NVIDIA chips and other advanced 2-nanometer chipsets provided by ARM. NVIDIA’s roadmap has consistently pushed the envelope for AI acceleration, and Stargate’s adoption of this technology reflects a commitment to unparalleled processing power and efficiency. These innovations will be critical in handling the immense computational demands of training ChatGPT 7 and beyond.

The timeline for Project Stargate’s milestones offers insight into the ambition and urgency behind the initiative:

• Data Center Built Complete: Late 2027
• GPT 7 Build Complete: May 2028
• GPT 7 Speed Optimized Build Complete: November 2028
• GPT 7.5 Feature Complete: May 2029

Last week, the project’s location was made public: Stargate will be built in Abilene, Texas. The facility will feature 10 buildings, each 500,000 square feet in size, with plans for expansion to 20 buildings. However, the additional buildings are not planned for Texas; OpenAI is exploring other locations across the United States for these expansions. These data centers are designed to handle the intense power requirements of AI workloads. While most data centers consume 500–600 watts per square foot for compute, GPU-intensive AI workloads demand double that, ranging from 500 to 1,000 watts per square foot. Another critical consideration is Power Usage Effectiveness (PUE), which measures the efficiency of a data center by comparing the energy used for IT compute to the energy used for non-IT functions like cooling. In 2020, the average PUE for data centers was 1.6, but modern designs aim for a target below 1.4. Companies like Facebook and Google have even achieved PUE ratios as low as 1.2.

If Stargate achieves a PUE of 1.2 and operates at 750 watts per square foot, the energy consumption per square foot would be 900 watts. For a 500,000-square-foot building, this translates to 450 megawatts of power consumption per building. With 10 buildings, the facility’s total power demand would reach 4,500 megawatts (4.5 gigawatts). To put this into perspective, Texas’s nuclear power plants produced 4,926.2 megawatts in 2022, accounting for 10% of the state’s energy generation, according to the U.S. Nuclear Regulatory Commission. A data center of this size aligns with leaks out of OpenAI from a year ago that the data center would consume 5GW of power.

Powering a facility of this magnitude brings its own set of challenges. To address the enormous energy consumption, the project has explored alternative power sources, including nuclear energy. The choice highlights the need for stable, high-output, and environmentally conscious power solutions, aligning with broader industry trends toward sustainability.

The scale of Stargate comes with an equally staggering cost. The project is estimated to require $500 billion, a sum that is ten times the current global annual spend on data centers. To put this into perspective, Stargate’s budget rivals some of the most significant infrastructure and scientific endeavors in history:

• American Interstate Project: $618 billion
• Apollo Program Costs: $190 billion
• Manhattan Project Costs: $31.4 billion
• International Space Station: $100 billion
• California High-Speed Rail System: $100 billion
• Los Angeles Transit Buildout: $150 billion

These comparisons highlight how Stargate’s investment represents a bold step into uncharted territory, emphasizing the profound importance placed on AI’s role in shaping the future. Unlike the projects above, which transformed transportation, space exploration, and global security, Stargate focuses on the digital frontier—ushering in an era where computation and artificial intelligence redefine the boundaries of human potential.

Stargate represents more than just a data center; it’s a statement about the future of AI. By investing in infrastructure at this scale, OpenAI demonstrates its commitment to staying at the forefront of artificial intelligence. As this project unfolds, its ripple effects will likely shape the trajectory of AI development for years to come.