Europe has spent much of the artificial intelligence revolution doing what it knows best: writing the rules while others build the technology. An Italian company now wants to challenge that division of labour.
As Leo Marchandon reported for Reuters on 25 June, Milan-founded Domyn is leading EUROPA, a consortium selected to develop a fully open-source artificial intelligence model using Europe’s supercomputing infrastructure.
The ambition is considerable. The proposed model will exceed 400 billion parameters, with organisations able to reproduce and operate it on their own infrastructure. For governments and businesses increasingly uncomfortable with technological dependence, that is an attractive proposition.
But there is an important distinction between owning the technology and knowing what to do with it.
Europe already possesses universities, researchers, industrial companies and computing resources capable of supporting significant technological development. What it has struggled to create is an ecosystem that consistently converts those assets into globally competitive products.
Domyn’s background makes the initiative particularly interesting. Rather than concentrating exclusively on consumer chatbots, the company has developed systems for regulated sectors, including finance, government and heavy industry.
These are areas where control over data and infrastructure can matter as much as model performance.
For Italy, this could be a more promising opportunity than simply attempting to reproduce Silicon Valley.
Italian manufacturing is characterised by specialist companies, complex processes and knowledge that rarely appears in publicly available datasets. Artificial intelligence capable of operating securely within those environments could deliver substantial economic value.
Yet the largest obstacle may not be technological.
European companies must be convinced that adopting European AI offers advantages beyond political symbolism. The systems must be reliable, commercially accessible and sufficiently adaptable to justify the investment.
Public institutions can finance research and provide computing capacity. They cannot guarantee that businesses will choose the results.
And a model containing hundreds of billions of parameters is not automatically better suited to a manufacturing company than a smaller, cheaper system designed for a specific task.
The measure of Europe’s success should therefore be neither the size of its models nor the nationality of their developers. It should be the creation of an industry capable of turning technological independence into productivity, employment and competitive advantage.
Italy has an opportunity to contribute to that industry. Whether it can also become one of its principal beneficiaries remains an open question.
The ITALIC View
Europe does not need an artificial intelligence model merely to demonstrate that it can build one. It needs companies that can turn research into useful products and customers willing to pay for them. Italy’s industrial knowledge may prove more valuable to that mission than another impressive technological announcement.