The European Union (EU) wants to more than triple the size of its datacentre capacity in the next five-to-seven years to reduce its dependence on artificial intelligence (AI) compute capacity located in other regions of the world, it has emerged.
This intention is detailed in a copy of the EU’s draft 12-page AI continent plan, which sets out the actions it claims must be taken to both capitalise on the European Union’s existing strengths in the field of AI and to reduce its reliance on overseas clouds.
“The EU currently lags behind the US and China in terms of available datacentre capacity, relying heavily on infrastructure installed in other regions of the world, which EU users access via the cloud [and] this dependence on non-EU infrastructure is a concern for European industry but also for public authorities,” said the draft document, seen by Computer Weekly.
“To adequately serve the AI and general computing needs of businesses and public administrations across the entire EU, and to ensure competitiveness and sovereignty, it is essential for the EU to increase its current cloud and datacentre capacity in a geographically balanced manner.”
To ensure the EU has access to enough sovereign cloud capacity to support its AI ambitions, it is consulting on the creation of legislation – dubbed the Cloud and AI Development Act – to accelerate the pace of new datacentre developments.
“Today, the average time to obtain a permit and the related environmental authorisations for building a datacentre in Europe often lies upwards of 48 months … [and] the datacentre industry struggles to identify suitable sites, and to obtain sufficient energy to power their facilities. The Cloud and AI Development Act will address these obstacles,” said the document.
It will do this by ensuring datacentre projects that meet certain energy and water efficiency usage requirements benefit from simplified permitting procedures so they can be fast-tracked.
“For highly critical use cases, including AI applications, sovereignty and operational autonomy can only be preserved by relying on highly secure EU-based cloud capacity. The Cloud and AI Development Act will ensure that the public sector in the EU can rely on such capacity for these use cases, thus laying the basis for the public sector to adopt AI in an environment of trust,” the document continued.
The document also outlined an intention by the EU to improve competition within the continent’s cloud market. “More generally, to enable the entry into the market of a more diverse set of cloud service providers, the Cloud and AI Development Act will explore establishing a common marketplace for cloud capacity and services,” it added.
These actions are all part of a push by the EU to establish its own “distinctive approach to AI by capitalising on its strengths and what it does best” so that it can position itself as an “AI continent”.
These strengths include having access to a “substantial pool of engineers and skilled professionals, a large single market with one single set of safety rules” that ensure the AI technologies being produced and used by its members are high quality and trustworthy.
The EU is home to a flourishing community of 6,300 AI startups, with more than 600 of these focused on generative AI (GenAI), said the document, but action must be taken to ensure these firms – as well as the EU’s population of AI researchers – have the resources they need to turn their early AI wins into long-term success stories.
On this point, the document said there needs to be a scaling up of the continent’s public AI compute infrastructure by establishing “gigafactories” containing energy efficient, high-performance compute capacity that can be connected by networks to create “AI factories”.
The EU has already committed to providing €20bn of AI infrastructure funding to partly cover the creation of five AI gigafactories within the European Union, but it is also calling on interested parties to commit to creating public-private partnerships to accelerate the buildout of these facilities.
“These gigafactories will foster scientific collaboration around powerful and unique infrastructures, bringing together researchers, entrepreneurs and investors to tackle ambitious projects in areas like healthcare, robotics and scientific discovery,” the document stated.
Establishing the EU as an AI superpower will also require AI innovators to have access to better quality data, as well as for more work to be done in-continent on the development of AI algorithms – and to encourage their adoption within “strategic sectors”.
More needs to be done to grow the EU’s “strong AI talent base” by “further developing excellence in AI education and research” by making it easier for skilled workers in this field to legally migrate across the continent, the document added.
John Buyers, global head of AI at law firm Osborne Clarke, said the EU document shares a lot of common ground with the UK government’s recently introduced AI opportunities action plan.
“[The] key themes of the EU’s plans echo those of the UK government’s AI opportunities action plan such as the focus on boosting AI innovation, the importance of data for AI systems to work on and the need for a major increase in compute capacity,” he said.
“There is a real emphasis on easing the burden of regulation and removing barriers to innovation … [and] the EU is explicitly recognising the crucial importance of AI for economic growth and security.”
