Does Europe have a credible path to competing in AI, or has the window already closed? Theodoros Evgeniou, INSEAD Professor of Technology and Business, discussed Europe's AI opportunities at an INSEAD Tech Talk with Constantijn van Oranje-Nassau, envoy at Techleap.nl and former European Commission advisor, as well as Tim Gordon, co-founder of Best Practice AI. 

To the opening question, Van Oranje-Nassau's view was that the large language model race is effectively over as a competitive opportunity. “The LLM space is very crowded and very hard to catch up with, given the capital requirements,” he said. The more realistic opportunity is industrial AI, where Europe still has something that the United States and China don't: decades of unexploited manufacturing data.

The conversation kept coming back to the harder question: What does Europe actually want? Van Oranje-Nassau was candid about how much ground has already been ceded. Cloud infrastructure went to US firms and manufacturing to China. “We allowed this to happen,” he said, highlighting the speed at which China is moving – both in automation and development – as a challenge Europe has not yet fully understood. 

Whether European leaders are grasping the urgency produced the sharpest response. While senior civil servants in the Netherlands are not permitted to use ChatGPT, the Dutch government is building its own model. Van Oranje-Nassau challenged the rationale, “If you're not even able to use the tools, how can you decide how to regulate them?” But he isn't without hope – the pace of European defence mobilisation has surprised even its critics, and genuine startup activity is emerging across the region. The window hasn't closed, but it won't stay open indefinitely.