This week, Microsoft made a significant announcement regarding its investment strategy in Germany, revealing plans to allocate €3.2 billion ($3.5 billion) within the next two years. The tech giant from the U.S. intends to enhance its artificial intelligence and data center capabilities in Germany, as well as expand its training initiatives. Brad Smith, the Vice Chair and President of Microsoft, shared these details.
This move mirrors a prior commitment made in November 2023, where Microsoft disclosed its intention to invest £2.5 billion ($3.2 billion) in infrastructure in the U.K. over a three-year period.
Both Germany and the U.K. view these investments as crucial steps towards bolstering their global competitiveness in the field of AI. Despite their significance, these investments pale in comparison to the substantial financial commitments made by U.S.-based cloud service providers, particularly within the U.S. itself. As the economic and military importance of AI continues to rise, governments worldwide are taking measures to secure control over this pivotal technology. Nonetheless, some experts caution that smaller economies might benefit more from collaborating with established leaders in AI rather than directly competing with them.
The Emergence of AI Nationalism
In 2018, Ian Hogarth, a tech investor at the time, penned an influential essay titled ‘AI Nationalism,’ predicting that governments would increasingly support their domestic AI industries as the power and significance of AI grows in various sectors. This foresight has materialized as many countries have ramped up public funding for AI research and commercialization, leading to regulatory frameworks aimed at safeguarding domestic companies. For instance, Germany and France spearheaded efforts to amend critical aspects of the continent’s landmark AI regulation, fearing potential constraints on their domestic AI champions.
While computational power, known as “compute,” was highlighted as a key factor in AI production that governments should influence, the pivotal role of compute in AI progress and policy was underestimated. Numerous governmental actions linked to AI nationalism have focused on securing access to compute or restricting rivals’ access to it.
For instance, in 2022, the U.S. imposed export controls to prevent countries like China from accessing cutting-edge semiconductor chips and related manufacturing equipment. Subsequent updates to these restrictions and proposed regulations by the Department of Commerce underscore the emphasis on monitoring cloud compute usage to prevent cyberattacks. Countries such as the U.S. and China have incentivized domestic semiconductor production through subsidies, while also exploring the establishment of government-owned compute clusters. The endeavor to attract major cloud compute providers to build data centers in their jurisdictions involves ensuring sufficient power supply and streamlining regulatory processes, with these providers predominantly based in the U.S. and China.
Significance of Scale
In a statement accompanying Microsoft’s investment declaration in the U.K., Prime Minister Rishi Sunak hailed it as a pivotal moment for AI infrastructure and development in the country.
However, of the projected $120 billion expenditure by Microsoft on data centers and AI infrastructure in the next two years, the investments in Germany and the U.K. constitute less than 3%. The bulk of Microsoft’s funding is directed towards investments in the U.S., as per insights from Dylan Patel, Chief Analyst at SemiAnalysis. U.S.-based companies tend to prioritize domestic data center construction due to factors like abundant renewable energy resources supported by natural gas and more flexible permitting regulations.
Efforts to establish state-controlled compute clusters to ensure sovereign access face significant challenges. While the U.K. plans to invest £900 million in a government compute cluster, this amount pales in comparison to the substantial annual investments made by U.S. cloud providers in AI infrastructure.
Robert Trager, Director of the Oxford Martin AI Governance Initiative, points out that many countries may lack the resources to build the requisite scale of compute infrastructure for their projects. Therefore, partnering with cloud systems based elsewhere could be a more pragmatic approach for these nations.
In strategic technology domains, states often rely on external sources. For instance, U.K. intelligence agencies opted to store classified information with Amazon Web Services due to the lack of a U.K.-based provider offering the necessary scale and capabilities. Similarly, many countries opt not to develop their own nuclear arsenal, relying instead on security assurances from nuclear-armed states.
Trager emphasizes the potential necessity for power-sharing arrangements to prevent certain countries from lagging behind in AI capabilities. He advocates for multinational governance frameworks that offer diverse stakeholders a voice to mitigate competition over these influential technological resources, which shape global power dynamics.
