Germany’s Seedbox.AI Navigates Challenges in EU AI Landscape
Germany’s Seedbox.AI is strategically positioned to engage EU lawmakers, focusing on artificial intelligence development alongside established industry players. Co-founder Kai Kölsch emphasizes that the company has been active in AI since before OpenAI’s ChatGPT gained traction in 2022, reports 24brussels.
Seedbox.AI specializes in customizing large language models (LLMs) from major firms like Google’s Gemini and Meta’s Llama for specific applications, including a chatbot for medication inquiries and an AI assistant for real estate appraisals. Kölsch sees Europe’s opportunity in capitalizing on existing LLMs rather than attempting to recreate them. “We want to drive the [AI] field forward,” he states, advocating for improvements in technology efficiency.
However, the road ahead is fraught with challenges. A primary concern for European AI firms is the lack of access to the same computing infrastructure as their American counterparts. Kölsch suggests that access to specialized AI chips is crucial for training and running models. With the EU establishing AI factories in Stuttgart to better link startups with necessary hardware, Seedbox.AI is collaborating with these resources to train models in all 24 official EU languages. Kölsch believes that acquiring advanced chips will enhance their capabilities.
Despite the establishment of larger gigafactories for AI, Kölsch stresses the need for strategic consolidation to effectively train new foundational models. The European Commission’s division of the initial “AI CERN” concept into five separate gigafactories complicates the situation further.
Kölsch also expresses concerns about the EU’s regulatory landscape, particularly the AI Act. He advocates for a delay, citing its chilling effect on established companies. “Doing nothing is the worst thing you can do right now,” he warns, as calls for regulatory reassessment grow amidst legal uncertainties leading to delayed implementation across member states.
Capital investment emerges as a pivotal issue for European tech firms. Kölsch notes a significant disparity in venture capital inflow, with European startups receiving six to eight times less funding than their American counterparts, despite comparable economic sizes. He attributes this gap to stringent regulations preventing institutional investors from engaging with high-risk startups and the fragmented financial sector across the EU’s 27 countries. “It’s absurd that we want to finance Europe and at the same time European pension funds are investing in US bonds,” remarks Andreas Schwarzenbrunner from Speedinvest.
Although there have been improvements in funding ecosystems for young companies, accessing necessary financing remains a struggle for European tech firms. Kölsch emphasizes that external threats from US Big Tech, capable of deploying significant resources to acquire promising startups, further exacerbate the issue. He mentions that Seedbox.AI is attracting interest from companies like AMD and Nvidia, suggesting that European firms are lagging in technical innovation.
An additional challenge pertains to talent acquisition. Currently employing 15 staff members, Seedbox.AI is seeking to recruit more AI engineers, a task compounded by the competitive climate among tech companies. The EU’s initiative in 2023 to promote STEM education aims to cultivate a future workforce, but Kölsch points out the difficulty in competing with enticing salary offers from American firms, which have reportedly presented earnings exceeding €100 million to top researchers.
As European tech grapples with these multifaceted challenges, the overarching need remains clear: the ability to attract talent and investment in an environment where capital flows favor established giants. “Talent just goes to where the capital is,” concludes Kölsch, stressing the urgency for systemic changes within the EU.
