The race for technological sovereignty is becoming the new driving force for global growth. As countries around the world seek to build their own infrastructure to ensure they, too, can use generative artificial intelligence (Gen AI) to enhance productivity, drive innovation, improve customer experiences, or generate blog posts, social media content, marketing materials and even computer code, they are helping create new markets.
Little wonder, then, that new national ecosystems of AI chips, hardware, software and services are emerging not just in the developed world but also in Asia and the Middle East.
The importance of AI capabilities at nation-state levels has drawn much attention. Indeed, resource-rich Gulf states were among the first to jump on the AI chips bandwagon soon after ChatGPT, the pioneering Gen AI chatbot and virtual assistant, was unveiled two years ago.
They began buying state-of-the-art AI chips and laid out plans to build data centres within their borders.
“In the last 12 months, we have seen India, Japan, France, Canada and [in] Southeast Asia, Singapore [and] Malaysia, speak up about the importance of investing in sovereign AI capabilities,” Jensen Huang, the CEO of AI chip giant Nvidia, said last month.
Until recently, the bulk of the high performance computing market was the US and, to a much smaller degree, China.
See also: Australia’s social media ban for under 16s to become law
Now, “for the very first time, because of Gen AI, computing is impacting literally every industry and every single country,” Huang noted.
As such, “governments around the world are increasingly becoming aware of the potential held in their own stores of national data,” he said.
“No country can afford to outsource its intelligence,” Nvidia’s founder noted. “Every country has its native natural resources — its languages, culture, and intelligence. Countries realise their data is part of their natural resource, not just copper mines or gold or lithium.”
See also: UBS-Credit Suisse integration opens up new tech for bigger plans
Most countries just want to ensure they can harness their natural resources or data and refine it to produce artificial intelligence. To do that, sovereign states need AI factories, like data centres, within their borders.
Fortunately, data is not the only thing these countries have in abundance. Many countries also have excess power capacity or energy to power data centres that use chips for AI training or inference.
Take Malaysia, for example. It has 40% excess power capacity, equivalent to 15GW, as well as competitive power tariffs. It also has lots of inexpensive land, adequate water resources, supportive government policies and proximity to Singapore and unlike some countries wooing data centres, it has low earthquake risk.
Sovereign AI allows nation-states to join the growing global AI ecosystem. “This new industry, producing intelligence, is one of the greatest opportunities we’ve ever seen,” Huang said in an interview last month.
“You’re going to see countries around the world continue to use public clouds, but also build regional data centres as well as publicly-supported infrastructure, so that each one of the countries will be able to cultivate and advance its own industries,” said the Nvidia CEO noting growing adoption of AI in healthcare, logistics, transportation, as well as manufacturing including heavy industries.
High stakes
Unfortunately, most of the AI spending comes from a handful of large tech companies. Hyperscalers or large cloud infrastructure providers like Amazon, Microsoft, Google and others are spending nearly US$50 billion ($64.5 billion) a quarter — or an annualised run rate of US$200 billion — on AI chips and other hardware, software and services.
Sink your teeth into in-depth insights from our contributors, and dive into financial and economic trends
So much so that a growing concern for investors has been where the next big source of demand for AI will come from when big spenders like Amazon’s AWS, Microsoft’s Azure, Google Cloud, as well as Meta Platform and EV pioneer Tesla, start to slow their spending on state-of-the-art AI chips and other infrastructure.
In late July, Sundar Pichai, CEO of search engine Google’s parent Alphabet, said, “the risk of underinvesting (in AI chips) is dramatically greater than the risk of overinvesting” for tech giants. “At this point, I’d rather risk building capacity before it is needed, rather than too late,” Mark Zuckerberg said recently.
The stakes are high and in the winner-takes-all world of tech right now, it pays to outspend your peers. If you were worried about overinvestment from cloud hyperscalers, fret no more.
The fastest-growing segment of AI chips and hardware is Sovereign AI. If nation-states continue to grow their AI spending at the pace they have been recently, soon it wouldn’t matter if Meta Platform or Google start to slightly moderate their own pace of AI spending.
Sovereign AI is part of the second wave of the ongoing artificial intelligence boom. The first wave was mostly driven by the private sector, with companies across the tech spectrum, from global tech giants to AI start-ups.
The catalyst for the second wave is the recognition that every region and every country needs to build their own AI ecosystem with Sovereign AI. That is why you see a huge boom in data centres across Asia, from Johor Bahru to Jakarta and Ho Chi Minh City to Hangzhou.
Think of Sovereign AI as a country’s approach to harnessing AI for its own socioeconomic, cultural and geopolitical context. That radically differs from corporate AI, whose main objective is to generate profits and gain market share.
Sovereign AI includes deploying AI technologies to protect national sovereignty, security, economic competitiveness and societal well-being.
Sovereign AI’s underlying drivers are the deployment of tier-2 cloud enterprise, education and research infrastructure, notes Pierre Ferragu, tech infrastructure analyst for NewStreet Research in New York.
Sovereign AI allows countries to produce artificial intelligence using their infrastructure, data, workforce and business networks.
Sovereign AI also helps reduce reliance on foreign AI technologies by developing domestic AI capabilities and ensuring access to the country’s critical data, expertise and infrastructure.
Asians do not want their data stored on remote servers in Darwin or Dallas. They would rather have their data as close to them as possible. Building a strong Sovereign AI will protect the countries from potential supply chain disruptions and also help reinforce national sovereignty.
Sovereign AI includes data infrastructures or foundation models, such as large language models (LLMs), developed by local teams and trained on local datasets to promote inclusiveness with specific dialects, cultures and practices.
Moreover, speech AI models can help preserve, promote and revitalise indigenous languages.
LLMs aren’t just for teaching AIs human languages but for writing software code, protecting consumers from financial fraud, teaching robots physical skills and much more.
Through its Sovereign AI initiative, Nvidia is helping countries around the world build sovereign AI capabilities, including ecosystem enablement and workforce development. This creates the conditions for engineers, developers, scientists, entrepreneurs, creators and public sector officials to pursue their AI ambitions in their own home countries.
Among the biggest proponents of Sovereign AI is Arvind Krishna, CEO of tech firm IBM Corp, who has long argued that governments should invest in AI infrastructure. Krishna has encouraged the Indian government to set up national AI computing centres nationwide.
Global initiatives
Sovereign AI spending globally this year is expected to reach US$30 billion this year, rising to US$70 billion in 2027. Nearly US$15 billion is expected to be spent on AI chips in 2024.
NewStreet Research estimates that up to US$115 billion has been committed to Sovereign AI efforts over the next few years with the top 11 countries — the US, Canada, Japan, China, India, Germany, Brazil, Saudi Arabia, France, Israel and Singapore — setting aside US$74 billion for their initial Sovereign AI investment plans.
Here is what is happening on the Sovereign AI front.
Japan’s National Institute of Advanced Industrial Science and Technology is building its AI Bridging Cloud Infrastructure 3.0 supercomputer with Nvidia.
Japan is using its Sovereign AI to upgrade the skills of its workforce, support Japanese language model development and expand AI adoption for natural disaster response.
Saudi Arabia’s Public Investment Fund (PIF) has set aside US$40 billion for an AI Fund with top VC firm Andreessen Horowitz.
Tokyo will spend US$1.5 billion on sovereign AI, including money to build supercomputers.
The US has set aside US$8.6 billion to boost the use of AI by government departments and agencies. Canada included US$2.4 billion for Sovereign AI this year in its federal budget.
Germany is spending US$5.1 billion on its AI Action Plan and Jupiter supercomputer, while India has earmarked US$1.3 billion for AI spending.
Taiwan is building an LLM called the Trustworthy AI Dialogue Engine, or Taide, funded primarily by the government as part of its Sovereign Model Strategy. Its main purpose is to counter the influence of Chinese AI tools like Baidu’s Ernie bot, which provides politically biased information. Taiwan uses Meta’s opensource Llama 2 model, fine-tuning it and using licensed content from local media and government agencies.
Last December, Singapore announced plans to build its own LLM, partly prompted by the “strategic need to develop sovereign capabilities in LLMs. Southeast Asia’s local and regional cultures, values and norms differ from those of the West, where most LLMs originate”.
The LLM project also supports its National AI Strategy 2.0. Nvidia has supplied Singapore’s National Supercomputing Centre with its H100 GPU chips.
India has ensured its Sovereign AI Plan will not compete against private sector AI players. Delhi’s game plan is to collate and make available Indian data for creating AI models and public-private partnerships to develop infrastructure to train and deploy AI in India and perhaps abroad. India is also pushing Sovereign AI initiatives to promote workforce development, sustainable computing and private-sector investment in domestic computing capacity.
Last month, French President Emmanuel Macron called on Europe to create public-private partnerships to buy more AI chips to train AI and help push its share of those deployed globally from 3% to 20% by 2035.
Earlier this year, the Netherlands unveiled its Gen AI Vision plan, which includes further development of the country’s open LLM, GPT-NL. The country is keen to pursue investments in large-scale scientific and technological infrastructure, including supercomputers, at the national and European Union levels.
UAE, Qatar and Saudi Arabia, aside from investing in their own sovereign AI, are ploughing money into AI ventures around the world.
UAE’s MGX, an AI-focused fund, is taking a stake in Open AI, the firm behind ChatGPT. It is also co-investing with Microsoft, Global Infrastructure Partners and asset management behemoth Blackrock to invest US$100 in data centres around the world.
Not every country can afford its own Sovereign AI. In a recent report, Barclays noted that “given the significant monetary and energy outlays required by leading-edge AI clusters, only 15 nations globally have fiscal budgets able to accommodate Sovereign AI”.
Huang, who co-founded the graphic chip behemoth in 1993, believes the growing push for AI sovereignty could ultimately see America lose its dominance over the global tech industry as governments around the world aggressively use public funds to support the building of the infrastructure needed to use AI.
Most likely, the US and other developed nations will build their own Sovereign AIs even as their private sector leverages AI to keep its competitive edge.
Assif Shameen is a technology and business writer based in North America
