TL;DR
Deutsche Bundesbank Executive Board member Fritzi Köhler-Geib used an Oct. 1 Brussels roundtable on transformative AI to argue that Europe should develop frontier AI as part of a broader digital-sovereignty strategy. She called for coordinated investment in areas including photonic chips and proposed a European “AI Airbus,” but the supplied statement does not set out a specific launch plan or commitments.
Deutsche Bundesbank Executive Board member Fritzi Köhler-Geib called on Europe to pursue transformative artificial intelligence alongside investment in chips, cloud services and other technologies, proposing a European “AI Airbus” to develop a large language model on European infrastructure. She delivered the remarks at a Brussels roundtable on Oct. 1; the Bank for International Settlements published the introductory statement on Oct. 8.
Köhler-Geib argued that frontier AI should be treated as part of a wider effort to strengthen Europe’s digital sovereignty, rather than as a stand-alone technology project. She said AI depends on other capabilities, including chips, cloud computing, robotics and quantum computing, and that coordinated development could make use of connections between those fields.
As an example, she pointed to photonic chips, which process information using optical signals. Köhler-Geib said European research and innovation in the area could help provide computing capacity for training and deploying advanced AI models, while using less energy than conventional chips. She described the technology as ready for mass production, but her statement did not identify manufacturers, production volumes, costs or a financing decision.
She said Europe should pool public demand, improve the path from research to commercial products and support application-focused research in selected areas. Public procurement could help give prospective chip producers a market, she said, alongside private investment, including venture capital. She cited the Scaleup Europe Fund as one existing effort to improve venture financing.
Köhler-Geib also described the Bundesbank’s current tender for EU-native cloud services and said the bank was seeking European providers for a large part of its multi-cloud strategy. She referred to cooperation with EU-native cloud providers in Eurosystem integration programs, including work on bank reporting, and said demand coordinated at EU level could have greater effect.
Europe’s AI Plans Depend on Infrastructure
The proposal matters because building or training advanced AI models requires more than model research: it also depends on computing power, data centres, chips and cloud infrastructure. If European institutions and companies rely heavily on suppliers elsewhere, they may have less control over key technology and fewer opportunities to capture the economic gains from the sector. Köhler-Geib presented these as strategic and economic risks, not as outcomes already measured in the statement.
A joint European effort could, in principle, connect public purchasing with private investment and give technology firms a clearer route to scale. That approach would also require agreement on funding, governance and participation across countries and industries. The speech makes the case for coordination, but does not announce an agreement or establish that any proposed project will proceed.
Her emphasis on public demand gives the discussion a practical dimension. Government and central-bank procurement can create customers for services and technologies, but whether such contracts can support competitive European suppliers depends on price, performance, security and the scale of demand. The Bundesbank’s tender is a described example; the statement does not report its outcome.
As an affiliate, we earn on qualifying purchases.
From Digital Sovereignty to Joint Projects
Köhler-Geib opened with an anecdote about a discussion of European digital sovereignty. She recalled that ChatGPT had given a negative answer to whether Europe could achieve it, while a European AI model offered reasons it could. She used the contrasting responses to argue both for developing European AI capabilities and for considering AI within a broader sovereignty agenda. The anecdote was an illustration in her remarks, not an assessment of either model’s capabilities.
She also referred to a paper she said she had published earlier in 2026 with Claudia Plattner, president of Germany’s Federal Office for Information Security, and Kristina Sinemus, Hesse’s minister for digitalisation and innovation. According to Köhler-Geib, the paper contains 37 implementation-oriented measures linking areas such as AI, chips, cloud, robotics and quantum computing.
For the proposed “AI Airbus,” she drew an analogy with the European aircraft manufacturer Airbus. She recalled asking former European Central Bank president Jean-Claude Trichet why Europe had created Airbus but not equivalent companies in other industries. In her account, Trichet pointed to Europe’s earlier gap with the United States in aircraft production, the absence of a European global champion and political leaders’ willingness to remove obstacles.
Köhler-Geib said an AI venture could build on an existing open-source model, host it on European infrastructure and advance its training. She called for at least one more European-trained large language model alongside Mistral, which she said should also receive further support. These were recommendations in a speech, not confirmed details of a formed company.
„In order to succeed in joining the ‘Premier League’ of frontier AI, we must approach this topic within the broader context of digital sovereignty.“
— Fritzi Köhler-Geib, Deutsche Bundesbank Executive Board member
large language model AI infrastructure
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
No AI Airbus Plan Was Announced
The statement does not say whether governments, investors or technology companies have agreed to form or fund an “AI Airbus.” It gives no proposed ownership structure, timetable, budget, location, participating firms or technical targets. It also does not specify which open-source model such a company would use, how its training would be advanced or how it would relate to Mistral.
Details about the photonic-chip proposal are also absent, including which start-ups might manufacture the chips, what investment would be required and when production could begin. Köhler-Geib said the technology was ready for mass production, but the statement provides no independent technical or commercial evidence for that assessment. The results of the Bundesbank’s cloud tender and the scale of any EU-level purchasing arrangement were not reported.
As an affiliate, we earn on qualifying purchases.
Funding and Procurement Decisions Ahead
The next steps described in the speech are proposals rather than a published implementation schedule. Relevant questions include whether public bodies will coordinate demand, whether private capital will support European chip production and whether an AI venture will be established. Köhler-Geib’s remarks provide no dates for decisions on those issues.
The Bundesbank’s cloud-services tender is an ongoing procurement process cited in the statement; its award and resulting supplier arrangements may show how the bank’s EU-native ambitions are put into practice. At EU level, any broader effort would require decisions on procurement, financing and the structure of cooperation. Until those details are disclosed, the “AI Airbus” remains a policy proposal, not an announced operational project.
As an affiliate, we earn on qualifying purchases.
Key Questions
Who called for a European “AI Airbus”?
Fritzi Köhler-Geib, a member of the Deutsche Bundesbank’s Executive Board, proposed the idea in introductory remarks at a Brussels roundtable on Oct. 1, 2026.
What would the proposed “AI Airbus” do?
Köhler-Geib described a European corporation that would build on an existing open-source model, host it on European infrastructure and advance its training. She did not provide a company plan, budget or launch date.
Why did she discuss photonic chips?
She said photonic chips use optical signals and are more energy-efficient than conventional chips, potentially supporting data centres for advanced AI. Her statement did not name manufacturers or specify costs, output or a production schedule.
Has the AI Airbus been formed or funded?
No formation or funding decision is reported in the statement. The AI Airbus is presented as a proposal, and details such as ownership, investors and participating countries remain unspecified.
Source: primary
