Germany frets over EU funding shortfall for AI training hubs
The European Commission's flagship plan to develop large gigafactories for computing will cost billions
The German government fears the European Commission hasn’t secured enough financing to support a massive build-out of AI computing power, according to an email sent by a national diplomat to other envoys and seen by Euractiv.
The so-called gigafactories are planned as hubs for training large AI models within the EU, as the bloc vies to keep pace with the US and China. To get them up and running, the facilities will require billions of euros in private sector investment alongside significant public money.
The Commission has pledged to find €3.4 billion to fund EU AI gigafactories, but its funding proposal sent to the Council for approval is so far just above €1 billion.
“The funds provided by the Commission are not sufficient in the current MFF [Multi-annual Financial Framework] to adequately finance the AI Gigafactories on the Commission side for five years,” reads an email sent on 27 October by Germany’s representative to the EU after the Council research working party on 24 October.
The spectre of inadequate funding could mean that investors view the whole project as having an “increased default risk”, the email goes on to warn, before suggesting that the effect of EU funds relief on private companies “may not be as great as suggested”.
At the meeting in late October, EU countries discussed an amendment to the EU’s existing public-private partnerships for the bloc’s supercomputing facilities – known as the EuroHPC – which will be tasked with building the AI Gigafactories.
The Commission responded to the financing doubts by saying that the project is still in its “first phase”. It also said it plans to seek additional support from the European Investment Bank and InvestEU fund. Additionally, it’s considering reallocating unused funds from the EU’s budget to pay for the AI hubs.
Germany declined to comment on their concerns when contacted by Euractiv.
(nl, jp)