Lagarde urges EU funds to absorb first losses on AI data centres and chip plants: KIRA proposes EUR 25 billion

By
CTOL Staff Reporter
1 min read

Lagarde urges EU funds to absorb first losses on AI data centres and chip plants, while KIRA proposes EUR 25 billion at EU level.

At a chip plant financing meeting, the crucial page specifies who takes the first loss. On 20 November 2026, Christine Lagarde urged European public money, including the EU budget, to occupy that position for chip plants and data centres. Her speech described a public-private partnership that would pull private investment into projects requiring European scale. A banker can lend against an enacted guarantee. A preference expressed from a lectern still leaves the banker pricing the loss.

We would pass on AI infrastructure financing that assumes Lagarde's first-loss appeal will turn into a dedicated EU budget allocation this year. We expect the 2028 to 2034 budget deal, which leaders aim to agree before the end of 2026, will keep that money flexible. The Earmark Gap is the distance between an available public risk-sharing mechanism and a specified amount set aside for a particular industry. Lagarde identified who should take the loss; budget negotiators still have to decide how much to reserve.

EUR 1.3 trillion sounds like a bill sent to Brussels. In the KIRA Center's September 2026 strategy, it is investment toward a 15% share of global AI compute by 2030, rather than a government invoice. KIRA proposes EUR 100 billion of public contributions, EUR 25 billion at EU level and EUR 75 billion from member states. EUR 100 billion over EUR 1,300 billion is 7.7%, and the EU share of EUR 25 billion over EUR 1,300 billion is 1.9%. The proposal gives private investors most of the bill and asks public institutions to make that exposure easier to bear.

An explicit allocation has a compelling economic case. The European Commission says, "Each EUR 1 of EU guarantee has supported EUR 15.20 in total investment". If the proposed EUR 25 billion were entirely a guarantee and achieved that ratio, EUR 25 billion multiplied by 15.20 would support EUR 380 billion of investment toward the 2030 compute ambition. KIRA proposes starting with InvestEU and continuing under the 2028 to 2034 budget, while member states supply their EUR 75 billion through guarantees, purchases and possible equity. A dedicated EU line would make the public share visible to lenders and accelerate investment in European capacity.

History supplies a more persuasive argument than the speeches. The European Investment Bank's EFSI programme, which ran from 2015 to 2020, used a EUR 21 billion EU budget guarantee; it now lists EUR 540.3 billion in investment related to approved projects. The European Court of Auditors says EFSI "contributed substantially to addressing the investment gap" and records a multiplier of 16.08 at the end of 2022. Guarantee calls stood at EUR 309.4 million by the end of 2023. That history makes it rational to believe an EU first-loss commitment can draw in private capital while keeping guarantee calls modest, as EFSI did through 2023.

The same audit makes the comparison less comfortable. Auditors found the EUR 503 billion investment figure overstated by EUR 131 billion, or 26%, and questioned how much financing was genuinely additional. The adjusted EUR 372 billion over EUR 503 billion, applied to the 16.08 multiplier, gives about 11.9. In a 2030 scenario that ratio turns EUR 25 billion into about EUR 297 billion, or 23% of the EUR 1,300 billion compute ambition. An EFSI portfolio can absorb one loan loss while other loans perform. A chip plant concentrates capital in equipment that loses value as chips age. The earlier success makes a new guarantee feasible, but it sets a demanding standard for the amount and the assets it covers.

The generous multiplier still produces a difficult financing calculation. At the Commission's EUR 15.20 ratio, EUR 1,300 billion divided by 15.20 calls for EUR 85.5 billion of guarantees, about 3.4 times the EUR 25 billion KIRA proposes at EU level. That is a scenario, not a forecast of the returns on AI hardware. KIRA's national EUR 75 billion mixes loan guarantees with purchase promises and possible equity, so adding it all to an EU guarantee would misstate the protection available to lenders. The political pitch is about leverage; underwriting depends on what each euro promises.

The proposed European Competitiveness Fund illustrates why budget flexibility has admirers. Its EUR 234 billion sits alongside EUR 175 billion proposed for Horizon Europe, a combined EUR 409 billion, with Digital Leadership among its policy windows. Money assigned broadly can support different technologies as demand changes. Reserving a first-loss layer for AI compute or chip plants would be a sharper commitment, including to projects whose economics have yet to emerge. We expect negotiators to prefer the broad window, even if developers would prefer a line with their own names on it.

Meanwhile the EU is pursuing another way to reduce AI infrastructure risk: buying output. EuroHPC's 30 July 2026 gigafactory plan would make public authorities anchor customers for compute time and expects more than EUR 20 billion of private investment. That is different from the Commission's EUR 20 billion InvestAI gigafactory fund. A purchase agreement supports prospective revenue, whereas first-loss protection absorbs initial losses when the investment fails. Confusing the two would flatter a project's apparent security and leave its lenders carrying risk they had assumed was public.

Execution is already colliding with public budgeting. On 5 November 2026, EuroHPC extended the Lot 1 deadline to 4 February 2027 to allow member states to complete budget procedures and submit financial commitments. Lot 2 remains due on 3 December 2026. EuroHPC says the bulk of its public funding for gigafactories will come from the next budgetary period. A dedicated first-loss fund would require policymakers to make another explicit allocation, while developers and their financiers would carry the timing risk.

Lagarde herself supplied an awkward qualification to the grand buildout. In the same speech, she said "the case for paying that price is hard to make" while foreign compute and frontier AI models remain accessible. The strategic desire for domestic infrastructure therefore competes with the financial option of purchasing it elsewhere. The Earmark Gap persists because officials can pursue some capacity through purchases and general guarantees before agreeing that taxpayers should absorb the first losses on a named class of projects. That distinction rewards consortia with funded customer commitments over those relying on a future budget promise.

We would pass on investments priced as though a dedicated EU first-loss allocation for AI compute or chip plants already exists, and favour financing supported by actual purchase contracts or agreed guarantees. We would be wrong if, by 31 December 2026, the Council's 2028 to 2034 budget agreement explicitly reserves any positive dedicated euro amount for first-loss guarantees covering AI compute or chip plants.

You May Also Like

This article is submitted by our user under the News Submission Rules and Guidelines. The cover photo is computer generated art for illustrative purposes only; not indicative of factual content. If you believe this article infringes upon copyright rights, please do not hesitate to report it by sending an email to us. Your vigilance and cooperation are invaluable in helping us maintain a respectful and legally compliant community.

Subscribe to our Newsletter

Get the latest in enterprise business and tech with exclusive peeks at our new offerings

We use cookies on our website to enable certain functions, to provide more relevant information to you and to optimize your experience on our website. Further information can be found in our Privacy Policy and our Terms of Service . Mandatory information can be found in the legal notice