Governments worldwide are racing to build sovereign AI systems, moving away from reliance on foreign tech companies and toward domestically controlled infrastructure. The UK launched Project Mercury, its first sovereign AI models designed to challenge hyperscaler dominance, while South Korea and the US are deepening their AI partnership through the newly signed U.S.-Korea Tech Prosperity Deal. This shift reflects growing concerns about data sovereignty, national security, and the risks of outsourcing critical technology to profit-driven private companies .

Why Are Nations Suddenly Focused on Building Their Own AI?

The push for sovereign AI stems from a combination of geopolitical anxiety and practical concerns about control. The UK has long operated as a "digital tenant," dependent on infrastructure and AI systems controlled by companies headquartered abroad. Project Mercury directly addresses this vulnerability by delivering large language models (LLMs), which are AI systems trained on vast amounts of text data, that are entirely trained and hosted on domestic infrastructure .

Beyond the UK, South Korea is accelerating its AI ambitions through government-backed initiatives. Coupang, a U.S. technology company with significant operations in Korea, recently invested $50 million in the SBVA Korea Sovereign AI Fund, matching a $50 million investment by the Korea Venture Investment Corporation (KVIC), Korea's government-backed agency supporting the venture capital ecosystem. This collaboration aligns with the U.S.-Korea Tech Prosperity Deal, which focuses on expanding technology ties between the two countries to strengthen shared national security priorities and accelerate research and development .

Experts warn that relying exclusively on foreign tech companies poses risks similar to past scandals. Cambridge University's DeepMind professor of machine learning noted that centrally deploying technology on people without engaging them creates vulnerability to repeating failures like the Post Office Horizon scandal, where a faulty IT system caused widespread injustice .

What Do These Sovereign AI Models Actually Include?

Project Mercury's portfolio spans a wide range of capabilities designed to meet both public and private sector needs. The initiative includes models ranging from lightweight "Edge Intelligence" models with 0.8 billion to 30 billion parameters for low-latency, local applications, to a 256 billion-parameter "Frontier Power" model designed for complex generative AI workloads. All models will be trained within the UK and powered by renewable energy .

The models will be deployed via Civo's sovereign cloud platform or hosted on-premise, ensuring full control over data and compliance with UK regulatory frameworks. This approach addresses longstanding concerns about data residency, security, and geopolitical risk linked to overseas-controlled platforms .

How Are Governments Supporting Sovereign AI Development?

• Direct Investment: The UK government allocated £500 million through its Sovereign AI Fund to accelerate domestic capability across infrastructure and model development, while South Korea's government-backed KVIC is driving the "Next Unicorn Project" to accelerate AI and advanced technology growth.
• Public-Private Partnerships: Coupang invested $50 million in the SBVA Korea Sovereign AI Fund, matching government investment and demonstrating how private companies can support national AI infrastructure alongside government backing.
• International Collaboration: The U.S.-Korea Tech Prosperity Deal creates formal frameworks for technology cooperation, with companies like Coupang serving as bridges for economic and technological collaboration between nations.
• Startup Ecosystem Support: Coupang's previous investment in the SBVA Alpha Korea Fund in 2023 is supporting nearly 20 Korean startups across various technologies and sectors to help them expand globally.

Beyond financial investment, companies are sharing operational expertise. Coupang is working with Texas-based robotics startup Contoro to expand AI-powered autonomous robots globally, exploring pilot programs at Coupang logistics sites in Korea and other locations. The company is sharing knowledge and operational expertise to help Contoro refine their technology for the Korean logistics environment .

"As a U.S. tech company serving millions of customers globally, Coupang is a vital bridge for economic and technological collaboration between the U.S., Korea and other countries where we do business," stated Harold Rogers, interim chief executive officer of Coupang Corp.

Harold Rogers, Interim Chief Executive Officer, Coupang Corp.

What Are the Risks of Depending on Foreign Tech Companies?

Recent setbacks have highlighted the dangers of over-reliance on US tech giants. OpenAI put its UK datacentre project named "Stargate UK" on hold this month, while a government plan to open "the largest UK sovereign AI datacentre" by the end of this year was revealed to be well behind schedule, with the site still in use as a scaffolding yard. Microsoft announced that a planned datacentre in the north of England would not come online until at least 2033 due to a shortage of power from the grid .

Experts argue that outsourcing AI development to private billionaires with zero loyalty to the British state creates systemic risk. One Cambridge University professor warned that "these corporations are clearly not aligned with the interests of our citizens," while a former UN and UK government adviser told Parliament that the UK's reliance on US tech companies including Google, Microsoft, OpenAI, and Amazon risked repeating past failures .

The geopolitical dimension adds another layer of complexity. While the US pursues what some experts describe as a "wild west" approach focused on corporate competition, China is positioning itself as backing multinational attempts to introduce global governance of AI. Chinese AI researchers are described as efficient, innovative, and willing to release their models on an open-source basis, though Beijing requires Chinese AI companies to cooperate with state intelligence work .

The sovereign AI movement reflects a fundamental shift in how nations view technology infrastructure. Rather than treating AI as a commodity to be purchased from foreign providers, governments are investing in domestic capabilities to ensure data security, maintain strategic autonomy, and reduce vulnerability to geopolitical disruption. Whether these investments succeed will shape the global AI landscape for years to come.