If you lost your LLM tomorrow, would you still have a business?

If a company relied solely on a single AI model provider, such as Anthropic or OpenAI, it would face significant risks if that supplier were to cease operations or limit access. For instance, in June, Anthropic temporarily disabled its two newest models, Claude Fable 5 and Mythos 5, in response to a US Department of Commerce decision restricting access for foreign nationals. Turned off completely, the models were restored after 17 days, highlighting the impact of such a shutdown on any business dependent on them.

As a European AI company, one would first need to consider how to mitigate the risk of losing access to crucial AI models. Rather than solely focusing on the product itself, the priority would be understanding what would happen if that supplier suddenly became unavailable and how quickly an alternative could be found. This awareness is vital in an era where AI sovereignty is increasingly recognized as a key consideration for businesses.

European companies must proactively identify and plan for potential dependencies on AI providers. This involves creating a contingency strategy that outlines what alternative models could be used if the current ones are no longer accessible. Simply relying on the availability of Open weights or alternative models like Mistral, Gemma, DeepSeek, and Qwen is not enough.

Founders need to take immediate action to identify backup options and estimate the time required to switch models, as this can vary significantly depending on the complexity of the application and the availability of relevant expertise.

While developing alternatives to major AI providers poses challenges, it is crucial for European companies to address these obstacles. Money, infrastructure, and skilled personnel are essential components of building an independent AI ecosystem. The process of raising capital and securing a data center with sufficient, uninterrupted power supply can take years, particularly in Europe where obtaining the necessary grid connections is a lengthy and bureaucratic process.

Despite efforts, such as the establishment of AI Factories and Antennas by the EU, significant hurdles remain, particularly when it comes to obtaining physical infrastructure.

The availability of skilled talent also plays a critical role in this equation. While Europe boasts a pool of talented AI researchers, many of them are lured to other regions, such as California, by lucrative opportunities. To avoid this talent drain, European companies must create an environment that attracts and retains top AI minds. This involves offering competitive compensation packages, fostering a culture that values innovation, and providing researchers with the resources they need to thrive.

In summary, building an independent AI ecosystem in Europe requires a multi-faceted approach that encompasses financial resources, physical infrastructure, and a strategic retention of top talent. By proactively addressing these critical factors, European companies can reduce their reliance on external AI providers and ensure business continuity in the face of potential disruptions.