Seedbox.AI backs re-training AI models to boost Europe’s competitiveness
Germany’s Seedbox.AI wants to boost AI efficiency, but faces chip shortages, funding gaps, and talent competition from Big Tech.
Germany’s Seedbox.AI is betting on re-training large language models (LLMs) rather than competing to build them from scratch. Co-founder Kai Kölsch believes this approach could give Europe a strategic edge in AI.
The Stuttgart-based startup adapts models like Google’s Gemini and Meta’s Llama for medical chatbots and real estate assistant applications. Kölsch compares Europe’s role in AI to improving a car already on the road, rather than reinventing the wheel.
A significant challenge, however, is access to specialised chips and computing power. The European Union is building an AI factory in Stuttgart, Germany, which Seedbox hopes will expand its capabilities in multilingual AI training.
Kölsch warns that splitting the planned EU gigafactories too widely will limit their impact. He also calls for delaying the AI Act, arguing that regulatory uncertainty discourages established companies from innovating.
Europe’s AI sector also struggles with limited venture capital compared to the United States. Kölsch notes that while the money exists, it is often channelled into safer investments abroad.
Talent shortages compound the problem. Seedbox is hiring, but top researchers are lured by Big Tech salaries, far above what European firms typically offer. Kölsch says talent inevitably follows capital, making EU funding reform essential.
Would you like to learn more about AI, tech, and digital diplomacy? If so, ask our Diplo chatbot!