Yesterday, the EU published a new “AI innovation package” to boost policy coordination and investment in AI. This political analysis of the package was originally sent to clients in advance of the announcement, earlier in the week.
Stay ahead of the news – find out more about our AI updates and get a free trial of our Emerging Tech service.
As we wrote last month, EU rhetoric on supporting AI innovation has tended to outpace the reality of funding constraints and disparate schemes.
The Commission is now looking to change this narrative and put its money where its mouth is. The initial plan to open up European supercomputer resources to innovative AI start-ups (announced on 13 September last year and originally expected in February) has not only been brought forward by a month to tomorrow but has also been expanded to become just one part of a much broader, Commission-led “strategic investment framework in trustworthy AI”.
The core of this new AI initiative remains the same: to open up the capacity of the European High-Performance Computing Joint Undertaking (EuroHPC) to EU start-ups.
Initially, the brainchild of Member States, the public-private partnership was set up as an autonomous vehicle from the Commission in 2020 as a means of pooling EU budgets with contributions from private stakeholders and individual Member States (along with six of the non-EU participants in the Horizon Europe programme: Israel, North Macedonia, Norway, Montenegro, Serbia and Turkey).
By widening supercomputer access (including pooling resources on computing power and data), the Commission is looking to bring the EuroHPC programme back into its orbit by expanding its responsibilities and providing greater strategic direction to its operation.
To this end, the Commission will issue the Joint Undertaking with a new objective of operating “AI factories”. In essence, this is a re-branding exercise: existing public supercomputer resources from the EuroHPC will be paired with their nearest data centres and opened up to AI start-ups as well as dedicated researchers and data specialists to advance the application of AI in a range of industries.
However, of particular significance is the Commission’s decision to pledge a further €2.1bn to the EuroHPC to help it maintain and upgrade its AI-enabled supercomputer capacity and research expertise as the technologies develop.
This represents the biggest Commission-led AI investment to date (Horizon Europe receives €1bn a year, but across a broader range of science and technology research projects, while the ill-fated Strategic Technologies for Europe Programme [STEP] has seen both its funding and scope slashed to focus on defence investments).
As a result, EuroHPC – until now a niche corner of European policymaking jettisoned to Luxembourg – will become a much bigger part of Brussels’ digital innovation strategy.
The expanded EuroHPC is likely to sit alongside the Horizon Europe and Digital Europe programmes, which are each being directed to jettison €500m of existing EU funding towards AI-specific innovation programmes. A further €200m (€100m each from EU-level budgets and from Member States, with the Commission’s support) will go to supporting start-ups/scale-ups in the space and to pooling linguistic data in support of the so-called “ALT-EDIC” – a Member State-led initiative to improve the linguistic capabilities of LLMs in smaller Member State language communities.
Nevertheless, the Commission’s AI initiative is not simply a renewed attempt at creating a centralised EU AI investment drive to sit alongside the forthcoming AI Act. In our view, it also brings AI development to the centre of the Commission’s wider economic policy and industrial strategy in two key ways:
- First, it aims to facilitate research into creating distinctly “European” AI, in line with the EU’s wider strategic autonomy objectives. Both the forthcoming 2024 Digital Europe Programme workstream to develop an EU LLM – a large language model capable of addressing all European languages – and the wider industry access to EU-level data from the European Common Data Spaces as part of the new “AI factories” play into the narrative of creating distinctly “European” AI R&D.
- Second, it aims to bring the development of new AI applications in line with the EU’s industrial strategy by focusing public-private partnerships and industry coordination on existing strategic industries for the EU, such as healthcare, biotechnology, robotics, batteries and manufacturing, and climate technologies.
As such, the AI initiative is designed to be presented as a “reset moment” in the Commission’s approach to AI research and innovation, which not only brings investment and coordination measures further under the Commission’s jurisdiction but also facilitates the interaction of AI research with wider EU-level strategic direction.
Whether the results can match the ambition remains to be seen.



