Artificial Intelligence & Machine Learning
,
Next-Generation Technologies & Secure Development
,
Regulation
Countries Are Pouring Billions Into Domestic AI Stacks to Escape US-China Dominance
Generative artificial intelligence emerged in Silicon Valley as a cool new technology, and now it’s being placed into the same breath of national security concerns such as power grids, ports and telecommunications. Many governments say AI is too critical to leave under foreign control.
See Also: OnDemand | Fireside Chat: Staying Secure and Compliant Alongside AI Innovation
By 2027, more than a third of the world’s nations will be locked into region-specific AI platforms built on proprietary data, infrastructure and governance frameworks, according to a Gartner forecast. That’s a sevenfold increase from the 5% using regional platforms today.
Nations are now safeguarding large language models in the same way they do critical national infrastructure, and some are making major investments in regional AI infrastructure and technology. Experts predict this trend will lead to overcapacity, higher costs and the end of AI as a globally interchangeable utility.
Most countries are still wrestling with questions related to “sovereign AI” – the technical ambition to develop domestic compute, models and data capabilities – and “AI sovereignty” – the political and legal right to govern how AI operates within national boundaries, said Gaurav Gupta, vice president analyst at Gartner. Most national strategies today combine both.
“There is no AI journey without thinking geopolitics in today’s world,” said Akhilesh Tuteja, partner, advisory services and former head of cybersecurity at KPMG.
US-China Dominance
The United States and China have emerged as leaders in AI development, with U.S.-based companies including OpenAI, Microsoft and Google leading frontier model development. “The U.S. has a market-driven approach, where they are funded by the enterprise,” Gupta said.
In contrast, China has adopted a “state-directed strategy” that emphasizes government funding, open-source communities, governance frameworks and carefully managed cross-border collaboration.
But geopolitical rivalry between Washington and Beijing has tightened export controls, restricted investment flows and politicized access to advanced semiconductor chips. The result is a global environment in which dependence on any single AI ecosystem carries risk – not just for governments, but for enterprises operating across borders.
That’s why governments – even in smaller countries – see AI sovereignty as a form of national risk management, Tuteja said.
“The geopolitics of today is expecting nations to become far more insular and self-dependent,” Tuteja said.
Pursuit of sovereign AI will cost many countries huge amounts of money and resources, cautioned the World Economic Forum. Annual global spending on AI infrastructure is expected to increase by up to $400 billion by 2030, and the report warns that some economies might fall behind as “AI sovereignty becomes increasingly conflated with infrastructure ownership.”
AI models themselves are also becoming more region-specific. European LLMs such as Mistral and Aleph Alpha are positioning themselves as alternatives to U.S.-based models. The European Union is applying a “trust-based lens, prioritizing transparency and regulations,” through the EU’s AI Act, Gupta said.
Meanwhile, Canada is building on its $2 billion Sovereign AI Compute Strategy, and India is investing in sovereign AI with a $2.4 billion funding push for supercomputing, R&D and multilingual capabilities in 2025.
In the Middle East, the UAE claimed to achieve an AI adoption to 97% nationally with its sovereign platform built with the help of a $100 billion infrastructure partnership with Microsoft and Nvidia. Saudi Arabia is supporting its state-backed AI platform Humain by investing in 6 gigawatts of data center capacity by 2034 and a new Microsoft cloud region, in hopes of becoming the third-largest player in the global AI market.
Confronting the “Closed US Model”
Smaller nations, Gupta said, are increasing their investment in domestic AI stacks as they look for alternatives to the closed U.S. model, including computing power, data centers, infrastructure and models aligned with local laws, culture and region.
“Organizations outside the U.S. and China are investing more in sovereign cloud IaaS to gain digital and technological independence,” said Rene Buest, senior director analyst at Gartner. “The goal is to keep wealth generation within their own borders to strengthen the local economy.”
Gartner projects that China and North America will lead in sovereign cloud infrastructure-as-a-service spending in 2026 at $47 billion and $16 billion, respectively. But their growth rates are projected to be in the 20% range, far below the triple-digit spending rate in other regions still in the early stages of building out their capabilities.
Europe is forecasted to surpass North America in sovereign cloud IaaS spending in 2027, partially because of a realignment of digital infrastructure away from U.S.-based cloud providers.
In France, for example, authorities recently announced a plan to discontinue the use of collaboration platforms Zoom and Microsoft Teams across 2.5 million civil servants by 2027, replacing them with a homegrown video communications platform hosted on sovereign cloud infrastructure.
The Infrastructure Challenge
The practical barriers to AI sovereignty start with infrastructure. The level of investment is beyond the reach of most countries, creating a fundamental asymmetry in the global AI landscape.
“One gigawatt new data centers cost north of $50 billion,” Gupta said. “The biggest constraint today is availability of power … You are now competing for electricity with residential and other industrial use cases.”
Power constraints have elevated utility and energy capacity to the status of strategic assets in the AI race. Countries pursuing AI sovereignty must either expand their electrical generation capacity or make difficult trade-offs between AI infrastructure and other societal needs. Gartner predicts that nations establishing a sovereign AI stack will need to spend at least 1% of their GDP on AI infrastructure by 2029.
But because most countries don’t have the resources to achieve this level of full-stack AI sovereignty, they are opting for a “layered” approach. The concept of “confidential AI” – systems that can leverage external computational resources while maintaining data privacy and security – offers a potential path for countries to benefit from global AI capabilities while maintaining control over sensitive information.
Despite the obstacles, one of the most compelling cases for localized AI development is that regional models can outperform global systems for context-specific applications.
“If you look at the overall AI technology stack, it starts with the sovereign AI infrastructure, which is the data centers and the semiconductors. But on top of it, you have security, you have models, you have applications, and you have the use cases,” Gupta said.
Controlling the Narrative
The demand for models that reflect local languages, customs, legal frameworks and societal norms is growing. But nations that build and control their own LLMs also control what those models will and won’t say.
“There has been concern in a lot of nations that if they use some of the existing closed frontier models, which are based in the West, or some of the open-source models based in China, the responses they get are not aligned to their culture and religion,” Gupta said.
The Saudi government’s AI chatbot Humain Chat, built on Arabic-first large language model, or ALLAM, is designed with “unmatched fluency in Arabic and deep alignment with Islamic, Middle Eastern and cultural nuance.”
China is reportedly imposing censorship on model training. For example, the China-based Deepseek model famously refuses to discuss Tiananmen Square or Taiwan’s sovereignty.
Unlike China and Saudi Arabia, Moscow has yet to release a globally competitive sovereign model, but Russian actors reportedly engage in “LLM grooming” to produce pro-Kremlin content via networks like Portal Kombat/Pravda and poison Western AI training data.
Iran and North Korea, similarly, have focused less on building sovereign models than on using existing LLMs for offensive cyber operations and influence campaigns, according to one Microsoft research.
Gartner warns that AI sovereignty will lead to reduced collaboration, duplication of effort and potentially supply chain disruptions.
“We already witnessed some of this during the pandemic, when there were shortages of semiconductors,” Gupta said.
Ultimately, large investments in data center infrastructure and the high demand for skilled workers could lead to overcapacity and higher operating costs for AI platforms. “As every nation goes in this direction, the buildout is happening irrespective of the supply and demand,” Gupta said.
The WEF paper argues that reframing AI sovereignty as “strategic interdependence” – rather than outright self-sufficiency – offers the clearest path to long-term competitiveness. The most successful economies, the report says, will be those that can connect with global networks and invest with precision.” Policymakers are urged to focus on comparative advantages, interoperability and alliances.
The question is whether the geopolitical forces now driving the sovereign AI race will leave much room for that kind of nuance.