The EU’s New Commission: What the First 100 Days Mean for AI Industrial Policy
Black Forest Labs’ Unicorn Status, New Investments in Aleph Alpha, and ECB Sounds the Alarm on AI Bubble
Welcome to the second issue of our EU AI industrial policy newsletter. If you’re new here, we track European debates on competitiveness, sovereignty, and AI.
Here’s what we’ll cover today:
What to expect from the first 100 days of the new Commission on AI
Black Forest Labs could be the EU’s latest unicorn: What counts as a European start-up anyway?
Schwarz Group and Deutsche Bank invest in Aleph Alpha, signaling a shift in the geography of large-scale AI in Europe
The European Defense Investment Plan drops its “Made in Europe” requirement
In other news: the increasingly murky chips economy, ECB warnings about global financial instability and an AI asset price bubble, OpenAI exec says Mistral is no serious competitor, a growing network on European AI partnerships, and discrimination in Denmark’s AI-enabled welfare state
What to expect from the first 100 days of the new Commission on AI
Von der Leyen’s Commission was waved through the Parliament hearings without a hitch. Threats from left-of-center groups to stall nominations over controversial, far-right nominees from Italy and Hungary never materialized. Some interpret this as a sign of weakness in the Parliament, traditionally the most progressive of the European institutions.
The new Commission will begin its term on December 1, starting the countdown for its first 100 days. During this period, we expect significant movement on a number of files aimed at turning the EU into “a global leader in AI.”
What to expect: a continued focus on resources for generative AI, promoting AI adoption through government procurement, and more concrete specifics around the new EU Cloud and AI Development Act.
In her mission letter, Executive Vice-President for Tech Sovereignty Virkkunen has been tasked with developing several AI initiatives within her first 100 days in office. Part of this effort is the recently launched AI Factories initiative, which aims to extend access to supercomputing capacity, data, and storage services for AI start-ups, SMEs, and industry to develop “cutting-edge generative AI models.” The Commission has received seven proposals from member states and associated countries to build an “AI Factory” around an existing or a new public supercomputer adapted to AI needs. The selection of five winners is expected to be announced by December.
Virkkunen has also been tasked with establishing a European AI Research Council—dubbed a “CERN for AI”—to pool European resources, including compute and researchers, to support AI research in Europe. She will also develop an “Apply AI Strategy” aimed at boosting industrial applications of AI and improving the delivery of various public services through AI.
We expect early signals on the timeline and scope of the proposed new EU Cloud and AI Development Act, which seeks to enhance European High Performance Computing (HPC), AI, and quantum capabilities and infrastructure, and the review of the Public Procurement Directive. In practical terms, this means hearings, evidence gathering, and calls for input are likely to begin in Q1 2025.
Issues to monitor in the first 100 days:
The fundamental question is whether investments in large-scale AI can in fact deliver the kind of economic growth, productivity gains, and increased efficiency that the Commission hopes for (the evidence base remains mixed). As we’ve previously argued, the proposed investments are not only meager compared to what Big Tech is spending, but the Commission’s emerging industrial strategy on AI also fails to account for Big Tech’s stronghold across the AI technology stack.
How well the Commission understands the global AI market—and Big Tech’s monopolized positions in data, talent, and the cloud—will likely shape the mission and design of the European AI Research Council, a proposal still in its conceptual phase. A “CERN for AI” could become a space to shape and steer “what AI is produced, why it is produced, and who it is produced by,” as Cecilia Rikap suggests. However, proposals such as purchasing 200,000 of the latest Nvidia chips to the tune of €13 billion euros highlight the risk of the initiative being captured by those pushing an AI arms race narrative—arguing that Europe must invest billions to compete for AI leadership while ultimately maintaining the status quo.
A key issue to monitor is the fine print of the Apply AI Strategy, which seeks to drive efficiency and productivity in both industry and the public sector. Beyond questions surrounding the evidence base for these premises, blanket AI adoption in government services raises concerns about infrastructural dependencies, single points of failure, and resource waste. In sensitive social domains, is also poses significant risks for harm.
Across all these initiatives, a critical issue to watch closely is how transparently decisions about the allocation of significant public resources will be made. Lessons from the European Chips Act, show there is much room for improvement. In her political roadmap, von der Leyen promised that the Apply AI Strategy would be developed with “civil society, industry, and Member States.”
Black Forest Labs: what counts as a European start-up anyway?
The German image and video AI startup Black Forest Labs is in talks for a $200 million funding round that could raise its valuation to $1 billion. As we highlighted in our October analysis of the EU AI startup market, the question of what counts as a European startup is far from straightforward. Should we assess Black Forest Labs based on the location of the company’s headquarters (Freiburg, Germany) or its legal place of registration (Delaware, U.S.)? Should it depend on its rumored sources of funding (U.S. venture capital, including Andreessen Horowitz, General Catalyst, and Lightspeed—many of the same backers behind existing EU AI champions like Mistral)? Or perhaps its partnerships (Elon Musk’s xAI)? Or even its infrastructural dependencies (Nvidia chips and an unknown cloud provider)? These factors illustrate the complex interplay of territory and companies.
What this means for AI in Europe:
Just as terms like “sovereignty” are often used interchangeably with state sovereignty, data sovereignty, or strategic autonomy—with unclear and sometimes contradictory implications for what constitutes sovereign technology—it is equally challenging to define a “European” AI startup. As boosting a “European” AI startup ecosystem has become a policy priority, addressing these questions will become increasingly urgent.
Schwarz Group and Deutsche Bank invest in Aleph Alpha, signaling shift in geography of large-scale AI
Aleph Alpha, Germany’s previous candidate for a European AI Champion, has seen changes in its ownership structure. The German financial giant Deutsche Bank injected fresh capital alongside the company’s long-term billionaire patron, the Schwarz Group, founded by the owner of the Lidl supermarket chain. Instead of providing new funding to the German company, Deutsche Bank acquired the shares of two early investors, the German VC funds 468 Capital and Lakestar, who opted to exit their positions.
What this means for AI in Europe:
The new funding round leaves the company’s valuation unchanged from a year ago, a sign that is typically not encouraging for a startup. Aleph Alpha has previously pivoted from training its own large language models (LLMs), instead moving upstream to help other companies deploy existing models. “It’s hard to make it make sense economically. It’s not hard to wire a lot of money to Nvidia,” CEO Jonas Andrulis told Sifted in October.
This shift has implications for the geography of large-scale AI in Europe, positioning France—and particularly Mistral AI—as the central force behind Europe’s aspirations for AI leadership. However, Mistral continues to face significant challenges, as chips and compute remain dominated and largely controlled by U.S. Big Tech. While Mistral’s trajectory continues to defy gravity, its path to revenue and profitability remains unclear, making the company a potential candidate for a takeover.
Changes in the European Defense Investment Plan
The European Defense Investment Plan (EDIP), the Commission’s planned instrument for supporting research and development in defense, has dropped its “Made in Europe” requirement, according to the Financial Times.
What this means for AI in Europe:
This reflects the impact of Trump’s election on Europe. While some have doubled down on the need for European “strategic autonomy” in response to the U.S. election, France’s shift on the plan’s “Made in Europe” requirement suggests the possibility of a more conciliatory approach.
In other news
The Financial Times has been reporting on the increasingly murky chips economy, raising concerns about the emergence of a new debt market. Wall Street’s largest financial institutions have loaned more than $11bn to a group of “neocloud” companies—cloud computing providers supporting tech groups building AI products—collateralized by their use of Nvidia AI chips. This ties the financial stability of these companies even more closely to the volatile ebbs and flows of the AI hype cycle.
More broadly, the European Central Bank has warned of potential global financial instability and an AI asset price bubble, as an increasing share of the stock market depends on the performance of hyperscalers.
OpenAI has officially opened its second EU office in the French capital. In an interview with Sifted, an OpenAI executive, stated that Mistral’s focus on open-source and smaller models means it is not a serious competitor to the Microsoft-backed U.S. giant.
In other Paris news, Mistral has released a series of updates to its product portfolio, including the integration of Black Forest Labs’ Flux Pro model for image generation into its platform, Le Chat. This partnership is part of a growing European network of collaboration, which already includes Finnish Silo and German Aleph Alpha.
Mistral’s flagship line of text-only models, Mistral Large 24.11, will soon be available through U.S. cloud platforms such as Google Cloud and Microsoft Azure, in addition to its existing accessibility on the AI model platform Hugging Face. The Schwarz Group has also entered into a partnership with Google to offer Google Workspace on its cloud platform. Not everyone was a fan.
Meanwhile, Amnesty International has published an investigation exposing widespread surveillance and discrimination against marginalized groups in Denmark’s AI-enabled welfare state.