EU’s $56M OpenEuroLLM Investment Isn’t Enough, Experts Say

The European Union has announced plans to create an open-source large language model with $56 million in funding to bring together top researchers from a consortium of 20 companies, universities, and EuroHPC centers across Europe.

The goal is to develop a model that works with all EU languages and offers an alternative to AI models from the US and China. Called OpenEuroLLM, the project will build the model using powerful supercomputers, such as Mare Nostrum in Spain and Leonardo in Italy.

Can the new European LLM become a serious contender in the bipolar US-China AI dominance battle? We asked experts to weigh in on the project’s potential impact and plans.

Table of Contents Table of Contents

1. Key Takeaways
2. Is $56M Enough? Experts Say It Falls Short
3. More Than Money: What the EU Needs to Compete in AI
4. Europe's Path to AI Sovereignty
5. Challenges Ahead: Can OpenEuroLLM Keep Up With Global Rivals?
6. The Bottom Line
7. FAQs
8. References

Show Full Guide

Table of Contents

1. Key Takeaways
2. Is $56M Enough? Experts Say It Falls Short
3. More Than Money: What the EU Needs to Compete in AI
Show Full Guide
4. Europe's Path to AI Sovereignty
5. Challenges Ahead: Can OpenEuroLLM Keep Up With Global Rivals?
6. The Bottom Line
7. FAQs
8. References

Is $56M Enough? Experts Say It Falls Short

While $56 million is small compared to the approximately $40 billion the Microsoft-backed OpenAI is reportedly seeking in a new funding round, it’s still 10 times more than the roughly $6 million China’s DeepSeek has said it spent to train its breakthrough model. (However, recent research indicates that DeepSeek has invested about $1.6 billion in infrastructure.)

Komninos Chatzipapas, founder of HeraHaven.AI, a small AI startup in Cyprus, told Techopedia:

“The budget definitely seems low, especially given the goal of developing foundation models (training AI models from scratch) rather than building on top of something, such as LLama, and adding the multilingual aspect.”

Training a model across so many languages is even more challenging because of the limited availability of high-quality multilingual data online, he said. At this point, there’s no open-source model capable of performing in that many languages.

“I think the direction we need to take is to help companies in Europe that have already proven they can do this very well,” Chatzipapas says. “Mistral AI is one example which, despite being overshadowed by DeepSeek, has developed lots of high-performing open-source and closed-source AI models. Their funding currently comes from investments from US companies.”

Developing a new consortium of a couple dozen companies seems odd, and this might have been a rushed effort, which might also explain why they did not allocate a lot of funds to it, he added.

More Than Money: What the EU Needs to Compete in AI

John Yensen, president at Revotech Networks Ltd., told Techopedia that the EU’s $56 million investment in OpenEuroLLM is a step forward but far from enough to establish AI leadership.

“First off, from my perspective, for the EU to complete, it needs larger funding, centralized coordination, and lastly stronger infrastructure. The AI firms that are leaders today have received billions of funding dollars, which has allowed them to access top talent and cutting-edge computing power. Spreading limited resources across 20 organizations may slow progress rather than accelerate it.”

The EU is thinking way too small with its $56 million investment, according to AI engineer Vincent Schmalbach. For comparison, that’s roughly what big AI companies spend in a week. But throwing more money at a big group of companies isn’t the answer either. Schmalbach told Techopedia:

“Here’s what would actually work: set up three to five massive GPU centers across Europe where small teams can access serious computing power. Put them in places like Paris and Munich connected to universities that already have good infrastructure. Focus on giving promising startups and researchers the tools they need, not on building one massive project.”

Even if the EU only manages to set up one or two of these centers, it would be a huge step forward. Sometimes, you need to think bigger to actually get something done, Schmalbach said.

Europe’s Path to AI Sovereignty

However, cybersecurity and AI expert Mithilesh Ramaswamy said the mission of OpenEuroLLM is not to compete with other models but an attempt to equip European companies with the tools to build AI in their own languages and local context, which they can own, modify, and use. The alliance also commits to preserving linguistic and cultural diversity.

“I find this exciting and empowering,” Ramaswamy said. “Speaking as a senior developer who’s been building AI features, this is exciting, especially given its tradition of successful cross-border collaboration and top-tier research institutions.”

The open-source model will also encourage innovation beyond the project itself, enabling startups and small and medium enterprises to build on this foundation and expand Europe’s AI ecosystem, he said.

“Europe’s leadership in data protection and ethical guidelines could position OpenEuroLLM as a preferred model for customers wanting features,” Ramaswamy added.

Challenges Ahead: Can OpenEuroLLM Keep Up With Global Rivals?

Still, Jacob Barnes, founder of FlowSavvy and an AI and development expert, said it’s going to be tough for the EU to compete because OpenAI’s ChatGPT has very deep pockets, and DeepSeek has much cheaper operational costs on its side to help maintain that competitive advantage on price. He told Techopedia:

“The success of their OpenEuroLLM will likely hinge on the EU’s moves to build a strong open-source community, and right now, it’s promising – open models are getting exponentially better.”

If they can find ways to encourage contributions from the top researchers and developers, there’s a real chance to see OpenEuroLLM not just compete but potentially lead the charge, he added.

“If they double down on areas where the EU holds strong values, like ethical AI development and data privacy, they could see even more success,” he noted.

And Ben James, founder of 404, Bittensor Subnet 17, said that while it’s encouraging that AI investment is on the EU’s radar, a much larger capital infusion will be necessary for Europe to compete with the immense funding driving advancements in the US and China.

“Moreover, to be a real contender in the global AI race, the EU should streamline coordination among its member states, encourage collaboration, and implement policies that foster innovation, retain talent, and promote broader adoption of AI technologies,” he added.

The Bottom Line

For the EU to successfully establish leadership in AI, it’s imperative that stakeholders define what leadership means in this context. Todd Ruoff, CEO at Autonomys, told Techopedia:

“What are the goals? If the goal is to create alternatives to proprietary models that already exist in the US and China, then $56 million is a symbolic gesture rather than a serious play. AI Innovation on a global scale is dependent upon computational power, talent, and access to high-quality datasets – none of which can be secured at scale with a fragmented funding approach.”

According to Ruoff, the strength of the EU lies in its regulatory vision, but regulations alone don’t drive innovation. A well-organized AI system needs to be set up to encourage open-source collaboration without being slowed down by too much bureaucracy.

“A more effective approach would be to channel resources into a smaller set of AI research hubs with long-term funding, a clear roadmap for commercialization, and direct incentives for enterprises to adopt and contribute to these systems,” he said.

“Without this, European AI will continue to be an academic exercise rather than a competitive force in the global market.”

FAQs