At TechBBQ in Copenhagen, Europe’s artificial intelligence conversation moved beyond which models are fastest or which startups are raising the most capital. Founders, investors and technology operators repeatedly returned to a more strategic question: who controls the models, infrastructure and data powering Europe’s AI economy?
TechCrunch reported that discussions throughout the Nordic conference focused on how Europe can gain more control over the technology powering AI, reflecting growing unease about depending heavily on technology controlled by companies in the United States and China.
UA.News, which summarizes and links back to TechCrunch’s reporting, likewise highlighted concerns about European ownership of AI models and the infrastructure on which they operate rather than continued dependence on foreign technology providers.
Model Access Turned Sovereignty Into an Engineering Problem
The debate gained urgency after Anthropic’s Mythos and Fable models became unavailable to users outside Europe earlier this year. TechCrunch said the incident encouraged people in Europe’s technology ecosystem to think more seriously about owning the systems underneath their AI products rather than renting that capability from foreign providers.
One startup executive shared that the disruption had seriously affected his software team, illustrating a risk that increasingly matters to founders: a model provider’s commercial or geopolitical decision can suddenly become an engineering dependency.
That turns AI sovereignty into something more practical than political rhetoric. For startups, control can mean model portability, access to European compute, ownership of critical data and infrastructure capable of continuing to operate if a third-party provider changes its terms or availability.
TechBBQ’s ‘Agency’ Theme Extended to Human Control
Ellen de Brever, angel investor and head of partnerships at the Novo Nordisk Foundation Cellerator, said TechBBQ’s “Emerging from Agency” theme was particularly relevant because discussion had shifted from what AI can accomplish toward what humans are actually willing to let AI do.
She argued that the central debate was not about producing smarter machines, but about human judgment and who remains in control of decision-making. UA.News similarly noted that conversations focused on human judgment and who would retain control over decision-making processes.
Privacy Is Part of the Sovereignty Debate
Signal President Meredith Whittaker pushed the discussion toward data infrastructure. During a panel on privacy and AI, she criticized the integration of assistants and agents directly into operating systems and described the current AI environment as creating a “data collection apparatus”.
Whittaker argued that there remains a “huge market need for privacy”, particularly as sovereignty becomes a more prominent concern for companies and governments.
Meanwhile, Emad Mostaque, co-founder of Stability AI and Intelligent Internet, warned about power becoming concentrated among a small number of AI laboratories. He defined sovereignty as the ability to resist power being exerted over you and argued that control of AI could eventually translate into considerable political and economic influence.
Europe’s AI Debate Is Moving Down the Stack
TechBBQ also covered growth capital, women’s health and cooperation between Nordic and African technology ecosystems, but AI sovereignty became a thread connecting many conversations.
For European founders, the deeper signal is that owning an AI company is not necessarily the same as controlling its technology stack.
A startup may be European while depending on American models, foreign cloud infrastructure and externally controlled APIs. As AI becomes embedded deeper into products and business operations, those dependencies become strategic vulnerabilities.
The conversation emerging from Copenhagen therefore points toward a broader European technology opportunity: building the models, compute, privacy infrastructure and developer tools that give European companies credible alternatives when dependence on foreign AI becomes too risky.
