Sovereign AI takes off as countries seek to avoid overdependence on superpowers
South Korea and others believe their homegrown tech sectors are strong enough to build up their own AI capabilities.
SEOUL—As China and the U.S. race to dominate artificial intelligence, countries are increasingly wary of becoming overly dependent on the superpowers for a technology that could profoundly affect their economic competitiveness and national security.
In response, a select few are seeking to build out their own AI capabilities and become leaders alongside the U.S. and China.
South Korea, with its rich and advanced tech sector, is a strong believer in the idea that smaller countries can succeed in developing a wide degree of autonomy in AI—a buzzy concept now known as “sovereign AI."
“In the age of artificial intelligence, falling behind by one day could mean falling behind by a whole generation," South Korean President Lee Jae Myung said earlier this month in an address to lawmakers. “We are currently facing an urgent, life-or-death crisis."
The overall goal of sovereign AI is to build out a localized computing infrastructure and potentially possess domestic capabilities in everything from large language model development and chipmaking to cloud storage and engineering talent. Many also hope to regulate and manage the data used and generated by AI on their own terms.
As one of the few countries outside the U.S. or China able to attempt a wide degree of AI autonomy, South Korea has emerged as a test case for sovereign AI. It has deep pockets, semiconductor know-how from homegrown firms like Samsung Electronics and SK Hynix, plus software expertise. And Lee, who took office in June, has provided the political willpower, pushing to triple the government’s budget for AI development for next year to the equivalent of roughly $6.8 billion.
The U.S. and China are the current leaders in AI technology. Beijing often expresses a desire for tech self-reliance, though it hasn’t yet focused much on exporting its best offerings. Meanwhile, the Trump administration has pledged to promote exports of a “full-stack" American AI technology package to allies, which includes Nvidia’s high-end chips and software.
By building up domestic capabilities in AI, countries hope to make their mark in a critical technology and shift away from the dominance U.S. tech giants hold over AI, said Virpratap Vikram Singh, who specializes in cyber technologies and conflict at the International Institute for Strategic Studies, a research institute based in London.
In various regions, there is “a growing hesitation" over becoming too dependent on U.S. tech behemoths, which authorities have found increasingly difficult to regulate, said Vikram Singh.
Countries also hope to build more resilience by having domestic AI capabilities and alternative service providers to turn to in case of potential disruptions, said Courtney Lang, who focuses on cybersecurity at the Information Technology Industry Council, a Washington, D.C.-based trade group.
Among the loudest champions of sovereign AI is Nvidia Chief Executive Jensen Huang, whose company stands to benefit from heightened demand for the powerful graphic-processing units used in AI computing. “It codifies your culture, your society’s intelligence, your common sense, your history—you own your own data," Huang said last year of sovereign AI.
Many countries have efforts in motion. France’s Mistral AI and Germany’s SAP have teamed up to launch a sovereign AI platform to protect European data in the public sector and regulated industries. The U.K. has set up a sovereign AI unit to facilitate investments into local startups and to become a more attractive location for AI firms. India is developing its own foundational AI model and expanding domestic computing capacity.
Saudi Arabia and the United Arab Emirates, which are among the most active and richest Middle East players in AI, recently won approval from the U.S. Commerce Department for two local companies to buy up to 70,000 advanced AI chips. The U.A.E.’s minister of economy has likened spending on sovereign AI to spending on areas like defense and cybersecurity.
Worldwide spending on AI, including in services, software and infrastructure, is expected to hit $1.5 trillion this year, according to Gartner, a market-research firm, up roughly 50% from 2024. The total is forecast to exceed $2 trillion next year.
South Korea, under Lee, has set up a roughly $102 billion “National Growth Fund" for investments over the next five years in high-tech strategic industries, including AI. A new presidential secretary for AI and future planning was created to spearhead public-private efforts.
More recently, the Lee administration unveiled a deal for South Korea to procure 260,000 of Nvidia’s advanced GPUs, touting it as the foundation for realizing the country’s AI ambitions. Some of the Nvidia chips promised to Korea will go toward a planned government-led data center. Others will go to data centers run by local software giants. The government, together with Korean firms, is also working to create a national large language model.
Korean conglomerates including Samsung, Hyundai and SK will use their share of the promised Nvidia chips to apply more AI to smart manufacturing and research. This month, the three business groups, as well as LG, pledged the equivalent of roughly $540 billion in domestic investments, including in AI data centers, chips and AI-based manufacturing.
Lesser-known names are also key to Korea’s push for sovereign AI. Chip startups such as Rebellions and FuriosaAI, which recently rejected an acquisition offer from Meta, are developing so-called neural-processing units that could potentially become lower-cost alternatives to Nvidia’s GPUs in some areas of AI computing. Korea’s top software giants, Naver and Kakao, have already rolled out Korean large language AI models of their own.
Countries looking to expand their AI capabilities face constraints, such as procuring sufficient power for data centers. A country’s ability to build out a sovereign AI ecosystem may also depend on the terms set by the U.S., which dominates many key technologies linked to AI, said Vikram Singh.
South Korea’s navigation of the challenges to foster an indigenous AI ecosystem could be a “crucial test bed and benchmark" for other nations pursuing the same task, said Chey Tae-won, chairman of South Korea’s SK conglomerate, at a recent economic conference.
Write to Jiyoung Sohn at jiyoung.sohn@wsj.com
topics
