Europe cannot afford to miss the artificial intelligence (AI) train. This technology is already significantly affecting the economy, scientific progress, and even military capabilities of nations. Currently, the USA and China are leading the way in this crucial domain. Until recently, Europe seemed content to follow in the footsteps of these two superpowers, but its strategy is set to change dramatically.
According to a report by CNBC, the European Union plans to invest €10 billion in the construction of thirteen data centers dedicated to AI. Additionally, it will allocate €20 billion for a network of “Gigawatt Class” facilities. These facilities will be among the largest and most ambitious projects within the realm of AI, consuming an immense amount of energy. To put it into perspective, a single gigawatt is equivalent to one billion watts, enough to power a small city.
Currently, sixteen European countries have expressed interest in hosting these facilities. As reported by CNBC, Munich, Germany, will be home to the first of these large data centers. Each Gigawatt-class installation is projected to cost between €3 billion and €5 billion and will feature at least 100,000 cutting-edge GPUs for AI, likely utilizing NVIDIA H100 chips. While the initiative looks promising, it raises an important question: how will the participating countries manage the electricity supply for such demanding facilities?
The Costs of Competing with the US and China
The US government, under Donald Trump’s leadership, is determined to spearhead advancements in AI technology, with its initiative dubbed the ‘Stargate Project’ slated to cost $500 billion. Funding will come from a combination of high-profile investors, including the Japanese investment group SoftBank, OpenAI (the creators of ChatGPT), Oracle, and the Emirati investment firm MGX.
Diverse entities will support the construction of an advanced network of data centers over the next four years to ensure US leadership in the AI sector. The spearhead of these facilities is already under construction in Abilene, Texas, and it promises to be colossal. This initial data center of the ‘Stargate’ initiative is anticipated to host more than two million chips designed for AI, as detailed by OpenAi.
The ‘Stargate’ infrastructure should be fully operational before the end of Trump’s current term.
The announcement of the ‘Stargate’ project has raised questions about how the US government plans to tackle the considerable electricity demand posed by these new facilities. As AI data centers are infamous for their high energy consumption, some tech companies are investing in nuclear power plants to meet their electricity needs.
Currently, the energy supply equation remains unresolved. The ‘Stargate’ infrastructure is required to be operational before Trump’s presidential term concludes, making it nearly impossible to establish a new nuclear power plant within such a timeframe. Nevertheless, OpenAI and Oracle have reached an agreement to deliver an additional 4.5 GW of energy to their data centers. Interestingly, SoftBank, despite its investment in the ‘Stargate’ project, is not involved in funding this energy expansion.
Another significant variable in this equation is China. Analysts from CBM consultancy anticipate that China will ramp up its investments in AI and semiconductors in response to the United States’ growing dominance in these fields. This competitive dynamic makes sense, as both countries are vying for global supremacy. Consequently, any advances made by one superpower are likely to provoke a robust response from the other. As we head into 2025, we can expect heightened activity from both the US and China on technological and geopolitical fronts—and Europe cannot sit idly by.
Image | Christina Morillo
In summary, Europe’s ambitious investment in AI technology signals its intent to catch up with the competitive landscape dominated by the US and China. This development is crucial, not only for economic growth but also for ensuring that Europe remains at the forefront of global technological advancements.