TThat week, Microsoft announced it would invest 3.2 billion euros ($3.5 billion) in Germany over the next two years. Microsoft Vice Chairman and President Brad Smith said the US tech giant will use the funding to double its artificial intelligence and data center infrastructure capacity in Germany and expand its training programs.
The move follows a similar announcement in November 2023, when Microsoft said it would invest 2.5 billion pounds ($3.2 billion) in UK infrastructure over the next three years.
Both countries welcomed the investment as an important step that will enable them to compete on the world stage when it comes to AI. However, this investment dwarfs investments made by cloud service providers in other US-based regions, particularly in the US. As AI becomes increasingly important economically and militarily, governments are taking steps to ensure control over the technology they rely on. But some experts warn that small economies may be better off sharing the power of AI with countries that are already in the lead, rather than trying to compete.
The rise of AI nationalism
In 2018, then-tech investor Ian Hogarth published an influential essay titled “AI Nationalism,” in which he argued that AI would become increasingly powerful and its economic and military importance. As AI rises, governments will take steps to promote AI nationalism. Domestic AI industry. (Mr Hogarth went on to play a key role in pushing the UK to lead the world in AI safety regulation).
Hogarth’s essay proved prescient. Many countries are increasing public funding for AI research and commercialization, and the urge to protect domestic companies is shaping regulations. For example, Germany and France led last-minute efforts to ease key aspects of the continent’s landmark AI regulations over concerns that they could hinder domestic AI champions, but ultimately It ended in failure.
However, although the essay mentions computational power, or “computing,” as one of the AI production factors that governments can and should try to influence, it is important to note that computing and how central it would become to policy. Many of the most important actions taken by governments related to AI nationalism are aimed at securing access to computing or depriving rivals of access.
For example, in 2022, the United States imposed export restrictions to prevent countries such as China from accessing cutting-edge semiconductor chips and the equipment needed to make them. The Department of Commerce will update this restriction in 2023, and in January 2024, the Department of Commerce will require cloud computing providers to monitor who uses their services and are found to be using them in cyber attacks. proposed rules that would require blocking access to anyone who does so.
Countries such as the United States and China offer subsidies to encourage domestic semiconductor chip production. Many countries have announced their intention to build government-owned computing clusters. Finally, governments need to ensure that large cloud computing providers headquartered in the United States and China have adequate power supplies and eliminate delays associated with planning permission, allowing them to build more on their territory. We’re trying to encourage people to build data centers.
size matters
In a statement accompanying the announcement of Microsoft’s investment in the UK, Chancellor Rishi Sunak said the announcement was “a turning point for the future of AI infrastructure and development in the UK”.
But research firm Semianalysis estimates Microsoft will invest less than 3% of the roughly $120 billion it will spend on data centers and other AI infrastructure over the next two years in Germany and the United Kingdom. Most of Microsoft’s funding will go toward investments in the United States, said Dylan Patel, principal analyst at Semianalysis. Patel said U.S.-based companies prefer to build data centers in the country, primarily because the country has an abundant supply of renewable energy backed by natural gas and permits. This is because regulations are also lax.
Efforts to secure access to sovereign computing by building state-controlled computing clusters are even more futile. The UK will invest £900m in government computing clusters, but this figure pales in comparison to the tens of billions of dollars US cloud providers spend on AI infrastructure each year.
“Many places probably won’t be able to afford to build the scale of computing needed to run the kinds of projects they want to do,” said Robert Trager, director of the Oxford Martin AI Governance Initiative. ” states. “For those countries, it may be wiser to train the model on a cloud system in a physically separate location.”
Countries often depend on other countries for strategically important technologies. For example, the British spy agency decided to store top secret information on Amazon Web Services, even though Amazon Web Services is based in the United States, because it has the “required scale or functionality.” This is reportedly because a UK provider could not be found. More broadly, many countries have decided not to build their own nuclear weapons, relying instead on security from nuclear-weapon states.
Given that it may be virtually impossible for some countries to catch up, power sharing may be necessary to ensure some countries do not fall behind. Yes, Trager says. “We need a multinational governance framework that gives voice to many stakeholders,” he says. “Without that, there will be competition for these resources that will affect global power.”
