As artificial intelligence (AI) becomes a global development trend, countries around the world are increasing their investments to advance related technologies, leading to a new wave of AI arms race. In an effort to secure global dominance in the AI market, the U.S. government announced in January 2025 the official launch of the "Stargate" AI infrastructure initiative, jointly led by OpenAI and SoftBank. The name, inspired by the 1994 science fiction film Stargate, symbolizes a gateway into a new era of AI.

Project Overview

The initiative aims to invest up to $500 billion from the private sector over the next four years to build multiple large-scale AI compute data center clusters across the United States. Through significant capital and infrastructure investment, it seeks to establish a future-oriented AI physical platform, greatly enhancing U.S. competitiveness in the global AI space—especially against rising challenges from countries such as China. It also supports the development of next-generation AI models with robust computing power. Additional objectives include driving U.S. reindustrialization, creating over 100,000 jobs, strengthening national security and strategic capabilities for protecting the U.S. and its allies, and accelerating the development of Artificial General Intelligence (AGI) with real-world applications in healthcare, climate change, and more.

As one of the largest investments in AI infrastructure in history, the U.S. government has designated advanced AI infrastructure as a strategic priority. President Trump emphasized that this initiative has significant strategic importance for ensuring the U.S. wins the AI arms race and expects AI technologies to bring long-term benefits to humanity.

Core Partners

The Stargate project is a government-backed AI infrastructure program combining industry players and global capital, co-led by SoftBank founder Masayoshi Son and OpenAI CEO Sam Altman. Key participants include OpenAI, SoftBank, Oracle, Abu Dhabi’s sovereign wealth fund MGX, and others, each contributing in critical areas such as R&D, funding, cloud services, chip supply, and international financing.

OpenAI

OpenAI is the project’s core tech provider and operations lead, committing an initial $19 billion and holding a 40% stake in Stargate LLC. Its main responsibilities include leading AI R&D—especially around AGI—and managing data center operations. OpenAI will provide stable and efficient computing support while leveraging its expertise in generative AI and emerging technologies to strengthen the project’s technological edge.

SoftBank

SoftBank serves as the primary financial backer and strategic leader of the Stargate initiative, also holding a 40% stake in Stargate LLC. It committed an initial $19 billion investment and will coordinate international partnerships using its global network. CEO Masayoshi Son will chair the board, guiding overall strategic direction.

Oracle

Oracle is a critical partner in infrastructure and technology, holding an equity stake and contributing $7 billion initially. Its responsibilities include providing cloud computing tech, building data centers, and integrating hardware and software infrastructure. Oracle will ensure high-performance computing environments and work with OpenAI to optimize infrastructure for AI training and inference.

MGX

MGX, a sovereign wealth fund from the UAE, is a key source of international capital, holding an equity stake and investing $7 billion in the initial phase. Its role is to enhance the financial stability of the project and foster integration with the Middle Eastern AI ecosystem.

Other Partners

Technology companies such as Arm, Microsoft, NVIDIA, and others are early-stage tech partners supporting chip supply, cloud services, and infrastructure integration.

• Arm provides efficient chip architecture and works with NVIDIA on optimizing AI chip performance and power efficiency.
• Microsoft offers cloud technology and infrastructure support, and through partnerships with investment funds such as BlackRock, contributes to a $100 billion international financing package while assisting OpenAI in AI model deployment and commercialization.
• NVIDIA supplies AI chips critical for HPC and training needs and collaborates on optimizing data center integration.
• BlackRock and other global investment funds provide roughly $100 billion in financing, improving the project's financial flexibility and attracting more global capital.

Scope and Content

The Stargate project initially aims to construct 10 ultra-large AI data centers across the U.S. The first base, "Stargate I," broke ground in early 2025 in Abilene, Texas. The site will gradually deploy about 64,000 GB200 chips and is expected to be fully operational by mid-2026, becoming the central compute hub for future OpenAI model training.

The long-term goal is to expand to up to 20 sites nationwide, with Ohio among the next key locations. The total goal is to achieve 10 GW of compute capacity within four years—equivalent to dozens of modern data centers—supported by more than 2 million advanced AI chips. Each campus will be designed for HPC environments, integrating cloud, AI chips, and low-power tech, with advanced cooling and renewable energy sources to manage high energy demands.

Due to the massive electricity requirements, President Trump declared a national energy emergency on his first day in office to fast-track power-related project approvals. SoftBank and OpenAI are actively working with state governments and infrastructure providers to secure land, power, and construction resources. In Ohio, for example, a new cost-effective data center model is being considered to increase efficiency and reduce build costs. Ultimately, the network of data centers will form an unprecedented supercomputing grid to support global AI advancement.

Global AI Investment Comparisons

The Stargate initiative ranks as the largest and fastest-moving among global AI infrastructure efforts. While the Texas Stargate I site has already partially opened, other nations are actively developing their own AI capabilities through data centers and compute platforms, including the UAE, Norway, China, and France.

Country	U.S.	China	France	South Korea	Japan
Project Name	Stargate	Global AI Governance Action Plan	Make France an AI powerhouse.	National AI Strategy Policy Directions	AI Promotion Act
Focus	Infrastructure	R&D, Industry Integration	Infrastructure, Talent, Workplace AI	Infrastructure, Semiconductors, Green Tech	Semiconductors, Robotics
Investment ($USD)	$500B	No exact total; est. $84–98B for 2025	~$112B	~$219M	No specific amount; ~$930M private AI investment in 2025
Progress	Texas site partially operational	$8.2B national AI industry fund launched	First projects to launch in 2025	Fund & plan launched; begins 2025	AI Promotion Act passed

Project Timeline

Short-Term

Focus is on U.S.-based construction. Stargate I in Abilene broke ground in early 2025, targeting full operational status by mid-2026 with around 64,000 NVIDIA GB200 chips and 1.2 GW of power capacity. An Ohio project is in the land acquisition phase, with plans to convert existing facilities into data centers to test multi-site deployment.

Mid-Term

The medium-term goal will focus on expanding domestic infrastructure in the U.S. and deepening international deployment to drive the adoption of AI applications and enhance its industrial influence. In the U.S., the plan includes completing 10 data centers with a total computing power of 10 GW, covering Texas, Ohio, and other potential locations. Each center will be equipped with 500,000 to 1 million AI chips, creating approximately 50,000 jobs. The project also explores using renewable energy and small nuclear reactors to meet the massive energy demand.

• Internationally, the Stargate UAE project will be expanded to 2 GW to establish a core AI hub in the Middle East. The Norway project will be increased to 290 MW to support sovereign computing power in Europe. At the same time, there are ongoing negotiations with countries such as Japan and South Korea for 2 to 3 new projects, with initial capacities ranging from 200 to 500 MW.
• On the technology front, the initiative will advance AGI development for applications in climate modeling and national defense, and will collaborate with NVIDIA, Arm, and Cisco to develop high-performance chips and network architecture. To support this scale of expansion, the plan aims to raise approximately USD 300 billion in international funding and establish a transparent regulatory framework to address data privacy and ethical concerns.

Long-Term

The long-term objective is to build a global AI infrastructure network to solidify the United States' leadership in the field of artificial intelligence and drive deep social transformation. The plan envisions the construction of 20 data centers, with a total investment of USD 500 billion and computing capacity exceeding 20 GW, creating over 100,000 jobs. It aims to boost the semiconductor industry and broader supply chains, further advancing America's reindustrialization efforts.

• On the technology front, the initiative will push for breakthroughs in AGI, applying it across areas such as mRNA-based cancer vaccine development, optimization of global energy grids, and autonomous defense systems. It also aims to promote educational reform and smart city development. In terms of energy, the plan combines renewable energy and nuclear power, adopting advanced cooling technologies to ensure long-term sustainability.
• From a global strategy perspective, the initiative aims to counter competition from China and Europe by partnering with G7 countries to establish a global computing power sharing platform, further strengthening U.S. geopolitical influence. However, risks such as funding shortages, energy supply pressure, and ongoing U.S.-China tech rivalry must be continuously monitored and managed to ensure the project progresses as planned and maintains its leadership in the global AI ecosystem.

Current Progress & Status

While the project launched in early 2025 with bold goals—including investing $100B that year and achieving 10 GW capacity within four years—progress has been slower than expected. As of mid-2025, Stargate LLC had not finalized new land deals for additional data centers. According to The Wall Street Journal, key disagreements between OpenAI and SoftBank—especially on site selection—have caused major delays.

The short-term focus has since shifted toward completing a smaller demonstration data center in Ohio by the end of 2025. Meanwhile, construction on Stargate I in Texas began in early 2025 and by mid-year had partially opened to handle early training and inference loads. Oracle started delivering NVIDIA GB200 chips in June. Thousands of local jobs have been created, with more expected as operations expand.

Key Challenges

Internal Coordination

The Stargate initiative involves collaboration among multiple parties including OpenAI, SoftBank, Oracle, and NVIDIA. However, major disagreements have emerged between OpenAI and SoftBank over several key partnership terms—particularly regarding data center site selection strategies and the integration of SB Energy. Oracle CEO Safra Catz publicly stated that “Stargate has not yet officially taken shape,” highlighting the complexity and challenges of internal coordination. This has directly led to project delays and a decline in investor confidence.

Progress in 2025 shows that the initiative is working to ease tensions through high-level negotiations and alternative strategies. Still, long-term success will depend on reaching clear agreements and implementing effective coordination mechanisms. Going forward, close attention should be paid to the evolving relationship between OpenAI and SoftBank, as well as the execution and implementation progress of international projects in various countries.

Funding Gaps

The Stargate initiative is an AI infrastructure project with a total projected investment of up to USD 500 billion. However, the initial funding target of USD 100 billion has yet to be fully secured, with OpenAI and SoftBank each committing only USD 19 billion—well below the original goal. The overall USD 500 billion lacks clear funding sources, and concerns over weak investor confidence, high cost pressures, and geopolitical risks may lead to a downsizing of the project or delays in its timeline.

Progress in 2025 indicates a gradual shift toward smaller pilot projects and alternative financing strategies. However, the long-term funding gap remains a critical issue. Continued attention is needed on developments in government subsidies and international cooperation to assess the project’s overall viability.

Energy Demands

The Stargate initiative faces significant energy challenges, with each data center requiring at least 50 MW of power. The first planned facility in Abilene, Texas, is expected to consume up to 4.5 GW, and the combined demand from all 20 data centers could exceed 20 GW—placing substantial strain on the U.S. power grid. Texas’s grid is already under pressure due to high-consumption industries such as cryptocurrency mining, making it difficult to absorb the additional demand.

At the same time, OpenAI and SoftBank have diverging views on energy supply strategies. OpenAI prefers small modular nuclear reactors or hybrid energy solutions to ensure a stable power supply, while SoftBank advocates for solar and storage projects under its SB Energy unit, maintaining a cautious stance on nuclear power in light of the Fukushima disaster.

Overall, rising energy demand, strategic disagreements, and the high cost of technologies are key factors contributing to project delays and financial strain. In 2025, the initiative has sought to ease short-term pressure through leased computing capacity and increased investment in green energy. However, long-term success will depend on government policy support and technological breakthroughs, along with continued monitoring of coordination efforts with the Texas power grid and the implementation of energy solutions across international sites.

Global Expansion & Related Projects

OpenAI

Due to delays from SoftBank, OpenAI has adopted a multi-cloud approach. In early 2025, it signed a deal with Oracle worth over $30B per year for three years—securing 4.5 GW of U.S. data center capacity. It also inked a $11.9B deal with CoreWeave. Together, these arrangements already match Stargate’s initial 5 GW goal—without SoftBank involvement.

SoftBank

In August 2025, SoftBank confirmed it would acquire a Foxconn plant in Ohio and convert it into a data center for Stargate. A 50-50 joint venture with Foxconn will oversee equipment design and manufacturing, while SoftBank fully owns the land and buildings. This marks the first official Stargate campus in Ohio.

Stargate UAE

Due to slow progress on domestic projects in the U.S., the Stargate initiative officially launched its first international project, “Stargate UAE,” in May 2025. Jointly announced by the U.S. and UAE governments, the project is part of the U.S.-UAE AI Acceleration Partnership Framework, marking the first time the Stargate vision has been implemented outside the United States.

The project is located within a newly developed 10-square-mile U.S.-UAE AI park in Abu Dhabi and will house the world’s largest AI data center cluster. The facility is being built by UAE-based company G42, with OpenAI and Oracle responsible for operations, and NVIDIA supplying its latest GB200/GB300 chip systems—about 500,000 units in the initial phase. Cisco will ensure network connectivity and cybersecurity, while SoftBank provides funding and strategic support. The initial buildout will support 1 GW of computing power, with a long-term goal to expand to 5 GW. The first 200 MW cluster is scheduled to go online in 2026, powered by a mix of nuclear, solar, and natural gas energy sources.

The project aims to position the UAE as the AI hub of the Middle East, supporting applications across government, healthcare, and energy sectors. It will offer nationwide ChatGPT-Plus subscriptions to residents and serve global computing demands within a 2,000-mile radius. The initiative exemplifies a multinational alliance model in which allied nations and corporations jointly build AI infrastructure and share in the benefits of the AI era.

Stargate Norway

On July 31, 2025, OpenAI announced the launch of the Stargate Norway project in collaboration with Norwegian industrial group Aker and infrastructure provider Nscale. The project involves the construction of a renewable energy-powered AI data center in Narvik, Norway.

The first phase of the project involves an investment of approximately USD 1 billion and will deliver 230 MW of capacity, with plans to scale up to 290 MW. Technically, the facility will deploy 100,000 NVIDIA GPUs, leveraging closed-loop liquid cooling systems and renewable energy to support high-performance AI computing.

The initiative aims to provide sovereign computing power for Europe, aligning with the EU’s AI Act requirements on data sovereignty, while also fostering the development of the local AI ecosystem.

Taiwanese Companies in the Supply Chain

Taiwan stands to benefit significantly from Stargate due to its strong position in AI servers, semiconductors, and power equipment. While the project has moved slowly, several Taiwanese companies have already started supplying hardware via OpenAI, Oracle, NVIDIA, and SoftBank.

Server Manufacturing

The Stargate initiative centers on the development of high-performance data centers, driving a significant surge in demand for AI servers and creating major opportunities for Taiwanese server ODM/OEM manufacturers. Taiwan holds a leading position in the global server manufacturing market, particularly in the AI server segment, where its strong supply chain integration and manufacturing capabilities have made it a key partner in the project.

• Hon Hai (2317): As the most central AI server manufacturer in the Stargate initiative, Hon Hai is responsible for producing NVIDIA’s GB200/GB300 AI servers. It has also formed a 50-50 joint venture with SoftBank in the U.S. to focus exclusively on the design and manufacturing of AI data center equipment, while land and buildings are fully owned by SoftBank. This move is seen as a critical strategic deployment by Hon Hai within the Stargate initiative. Going forward, Hon Hai and SoftBank plan to fully align in the areas of AI computing centers and equipment manufacturing. Hon Hai is also expected to secure the majority of AI server orders under the project. With the mass production of NVIDIA’s GB200 beginning in 2H25, the company is poised to see significant operational benefits. Additionally, as the UAE project comes online in 2026, order volumes are set to increase further, providing ongoing growth momentum.
• Quanta (2382): As Oracle’s key ODM supplier for AI servers, Quanta provides high-performance server hardware and is participating in the Stargate UAE project, supplying related server equipment. The company is expected to benefit from the deployment of NVIDIA chips. Thanks to its long-standing relationship with Oracle, Quanta has secured large orders and is well-positioned to continue gaining from the project's buildout and expansion.
• MiTAC (3706): Another important server partner for Oracle, MiTAC mainly provides AI server assembly and hardware support, catering to the compute demands of the Texas Abilene data center and the Stargate UAE project. Through its partnership with Oracle, MiTAC has successfully entered the project’s supply chain, though its order volume is relatively smaller compared to Hon Hai and Quanta.
• Wistron (3231) and Wiwynn (6669): Both serve as ODM/OEM suppliers of AI servers for the Stargate initiative, focusing on the design and manufacturing of data center computing equipment. Wiwynn, a key partner of NVIDIA, supplies AI servers based on the Blackwell architecture to support high-density computing demands in the project. Wistron provides server components and system integration services, participating in both U.S. and international supply chains. As the project expands globally, order volumes for both companies are expected to grow further.

Beyond server manufacturers, other Taiwanese hardware providers are also likely to benefit indirectly. These include thermal management specialist Auras Technology (3017) and power management leader Delta Electronics (2308). Although there is currently limited momentum toward technological upgrades, and large-scale adoption of advanced cooling remains constrained by cost, both companies stand to gain from growing demand for efficient cooling solutions and stable power systems in AI data centers. As the Stargate initiative progresses, related order volumes for these suppliers are expected to continue rising.

Semiconductors

The surge in AI computing demand is expected to drive up the need for high-performance GPUs and ASICs, in turn fueling growth in the semiconductor market—particularly in advanced process nodes and packaging and testing services. Taiwanese companies hold key positions in the global semiconductor supply chain and are participating in the Stargate initiative primarily through NVIDIA and SoftBank’s supplier networks, supporting the manufacturing and backend processes for NVIDIA’s Blackwell chips. This is expected to benefit Taiwan in areas such as foundry services, advanced packaging, and IC testing.

• TSMC (2330): As the world’s leading semiconductor foundry, TSMC commands a 100% market share in AI chip manufacturing and is responsible for producing NVIDIA’s GB200/GB300 chips using its advanced process nodes. With NVIDIA orders ramping up and long-term AI chip demand driven by the Stargate initiative, TSMC is positioned to see stable and sustained revenue growth, further reinforcing its dominance in cutting-edge process technologies.
• ASE Technology Holding (3711): A major provider of advanced packaging services, ASE supplies CoWoS and InFO technologies to meet NVIDIA’s packaging needs. The company stands to benefit from the expansion of the AI chip market and the data center construction boom fueled by the Stargate project, which is expected to increase utilization rates for its advanced packaging capacity and contribute additional revenue.
• King Yuan Electronics (2449): The world’s largest dedicated IC testing company, KYEC specializes in semiconductor packaging and testing services. It holds roughly 70% market share in final testing (FT) services for NVIDIA’s AI GPUs. With NVIDIA scaling up its orders and the Stargate initiative driving up AI chip demand, KYEC’s high-end testing business is likely to grow further and support overall revenue expansion.

Power Equipment

AI data centers consume vast amounts of electricity, which is driving demand for heavy electrical equipment to support power infrastructure development across projects in Texas, Stargate UAE, and Norway. At the same time, the U.S. has launched a grid modernization initiative to improve power stability, which is expected to boost orders for transformers, switchgear, and distribution systems. This trend is creating additional growth momentum and market opportunities for Taiwanese heavy electrical equipment manufacturers.

• Fortune Electric (1519): As the only company in Taiwan capable of producing 500 kV-class transformers and the largest high-voltage transformer production base in Southeast Asia, Fortune Electric became the first Taiwanese firm to directly secure an order from the Stargate initiative. The company has received a transformer order worth approximately NT$2 billion for the Abilene, Texas project. This contract accounted for a significant portion of new orders in Q4 last year and helped drive export growth. With the highest U.S. export exposure among peers, Fortune Electric is expected to continue benefiting from transformer demand tied to AI computing center developments under the Stargate initiative.
• TECO Electric & Machinery (1504): Primarily known as a leading home appliance manufacturer in Taiwan, TECO has in recent years aggressively entered the heavy electrical segment through acquisitions. The company operates a manufacturing base in Texas and holds federal supplier status in the U.S. It recently announced a strategic alliance with Hon Hai through a share-swap arrangement, positioning TECO to potentially benefit indirectly from rising demand for heavy electrical equipment and related infrastructure spurred by the Stargate initiative. This partnership could further expand its presence and revenue streams in the North American market.

Conclusion

As one of the largest AI infrastructure projects in history, the Stargate initiative carries the strategic ambition of strengthening the United States’ leadership in the global artificial intelligence race. Its potential economic and geopolitical implications are significant. Backed by large-scale private capital investment and supported by leading global players such as OpenAI, SoftBank, Oracle, and NVIDIA, the initiative aims to establish a worldwide supercomputing network over the next four to five years—providing long-term, stable computing power for AGI development and large-scale AI applications.

However, current execution reveals considerable challenges. Internal disagreements over site selection and resource integration have caused delays in key infrastructure deployment. In addition, high capital requirements, energy supply constraints, and the complexity of construction timelines have introduced further uncertainty. While OpenAI’s multi-cloud partnerships and international expansion strategies have helped ease computing power pressure, they also highlight the joint venture’s coordination and execution hurdles.

From an industry perspective, the initiative is set to significantly expand demand for AI servers, semiconductors, advanced packaging and testing, and heavy electrical equipment—offering long-term benefits to Taiwan’s related supply chain. In particular, with the mass production of NVIDIA’s Blackwell-architecture chips and rising demand for high-density data center computing, companies such as TSMC, ASE, Hon Hai, Quanta, Wiwynn, and Fortune Electric are well-positioned to benefit.

Overall, the Stargate initiative holds immense strategic and industrial significance. If its core participants can resolve internal conflicts and ensure effective allocation of capital and resources, the project has the potential to become a flagship example of global AI infrastructure development. In the short term, however, close attention is needed on construction progress, funding realization, and energy supply—factors that will not only determine the project’s success but could also shape the future of U.S. leadership in the AI ecosystem and the broader global tech landscape.