A handful are trying to carve out their autonomy and muscle into the same league as Washington and Beijing, with South Korea loudly insisting that nimble countries can build enough homegrown tech to stand on their own feet.
South Korean president Lee Jae Myung said: “In the age of artificial intelligence, falling behind by one day could mean falling behind by a whole generation. We are currently facing an urgent, life-or-death crisis.”
Countries want their own local compute, their own models, their own chips, their own cloud, their own engineering talent, and the right to handle their data as they see fit.
South Korea has become a poster child for this effort because it actually has the money, the chips, the software firms and, with Lee in office since June, the political oomph to triple next year’s AI budget to around $6.8 billion.
Lee is already waving about a new $102 billion National Growth Fund meant to feed high-tech projects over the next five years, with AI at the front of the queue.
The US and China remain the top dogs. Beijing talks a good game about tech self-reliance, although it rarely ships its best gear abroad. The Trump administration has promised allies a “full stack” of American AI wares, complete with Nvidia’s top-end silicon and software.
Countries want to shake off the grip of US giants, said International Institute for Strategic Studies cyber boffin Virpratap Vikram Singh, who argued that many authorities now show “a growing hesitation” about depending on firms they struggle to regulate.
Information Technology Industry Council Courtney Lang told the Wall Street Journal that building domestic clout gives nations fallback options if foreign supplies are cut.
Nvidia chief executive Jensen Huang has been one of the loudest cheerleaders for sovereign AI, which is hardly surprising given the demand it creates for his GPUs.
“It codifies your culture, your society’s intelligence, your common sense, your history, you own your own data,” Huang said.
Efforts are underway across the globe. France’s Mistral AI and Germany’s SAP have teamed up on a sovereign AI platform to keep European public-sector data tidy. The UK has established a sovereign AI unit to funnel funding to local startups. India is cobbling together its own foundational model while boosting its domestic compute.
Saudi Arabia and the United Arab Emirates have scooped approval from the US Commerce Department for two local firms to buy up to 70,000 advanced AI chips. The UAE’s economy minister compared sovereign AI spending to defence and cybersecurity.
Beancounters at Gartner say that worldwide AI spending is set to hit $1.5 trillion this year, roughly 50 per cent more than in 2024, and is forecast to top $2 trillion next year.
South Korea struck a deal to buy 260,000 Nvidia GPUs, calling the haul the backbone of its sovereign AI push. Some chips will feed a state-run data centre while others will go to local software outfits. Seoul is also building a national large language model with domestic firms.
Samsung, Hyundai, SK and LG have promised about $540 billion in domestic investment, targeting AI data centres, chip development and smart manufacturing.
Chip startups Rebellions and FuriosaAI are working on neural-processing units that could become cheaper alternatives to Nvidia gear for some AI tasks. Meanwhile, software giants Naver and Kakao have released their own Korean-language models.
SK chairman Chey Tae-won told a recent economic shindig that South Korea’s attempt to stitch together a homegrown AI ecosystem could become a “crucial test bed and benchmark” for other nations.