Three Openings for Europe in the Global AI Ecosystem

Europe has repeatedly missed out on tech-driven booms over the past 30 years, a critical factor in its lagging economic performance. As artificial intelligence (AI) emerges as the next major technological wave, we cannot afford to let this opportunity pass. The key question is: what strategy should the European Union (EU) pursue to secure its place in the global AI ecosystem?

Some European governments and companies think the way to do that is by developing low-cost copies of ChatGPT in local languages for local markets. That would be a mistake. A more effective approach to making AI in Europe could be founding a CERN for AI, developing more domain-specific AI systems, and creating a robust AI assurance industry.

Europe’s economies need rejuvenation. Aging populations, an intensifying climate crisis, a growing shortage of affordable housing, persistently high energy costs, and many other factors weigh on already stretched national budgets. While AI’s exact global economic value-add is unclear, the scale and growth could be substantial. PwC estimates the potential economic contribution to the global economy to grow to $15.7 trillion by 2030, rivaling the combined GDP of the ten largest EU economies.

Recognizing the immense economic potential of AI, the European Council called for the build-up of European capacity in this “key technolog[y] of the future.” Countless opinion leaders – from CEOs to politicians to academics – have also shared this sentiment. But, the exact value that AI can add to Europe’s economy will depend on a strategy that can secure the bloc a strong place in the global AI ecosystem.

Half-heartedly chasing European GPAI development will fail

Too many European companies and governments are making a superficial effort to develop general-purpose AI (GPAI) in local languages. The excitement is understandable; GPAI systems can help with a wide array of daily tasks in response to simple queries – and who wouldn’t want assistance in their native tongue? However, the bigger-is-better investment dynamics of GPAI development mean these efforts may lack the funding to compete with the current market leaders. Google’s latest Gemini Ultra model cost a staggering $191 million to train, starkly contrasting European efforts such as the Dutch government’s 13.5 million euro investment into a Dutch GPT. Moreover, the immense costs of development are increasingly challenging to justify solely on linguistic grounds, considering that, for instance, ChatGPT already supports 22 out of 24 official EU languages.

As a result, attempting to compete with American AI systems or similar GPAI models will not be a viable European strategy without unprecedented levels of government support. According to the latest estimates, global AI chip production will need to increase by a factor of 25 by 2030 to maintain current growth in GPAI sizes. Moreover, global power production must increase by an amount equal to 25% of Europe’s energy use every year.

In other words, the EU has to go big or go home. The scale of investment is too big for any single European company or even country to manage. As the Commission President rightly identified, only a CERN-like European AI Research Council initiative of resource pooling could move the required capital. And even so, the Union would likely need swift investment, single market, and bankruptcy rule reforms to successfully develop GPAI. In short, the EU should consider whether it can create the conditions to play in the big leagues of AI based on infrastructural, financial, and political factors.

Two underappreciated European strengths to secure AI market share

Rather than only chasing general-purpose AI systems, the EU should instead consider diversifying its innovation portfolio by investing in another kind of powerful AI: domain-specific systems. In specific, highly specialized technical industries with high-quality proprietary data, these narrower models often outperform GPAI for niche but critical tasks. The benefits can be life-changing for millions of citizens. For example, domain-specific AI models have already accelerated vaccine development. Moreover, not only are domain-specific AI systems more effective in their niches than GPAI, but they are also substantially cheaper to train. Alphafold3, a flagship domain-specific AI system in the life sciences, only cost 1 million USD to train, a hundredth of GPT4’s cost.

Fortunately, the EU can leverage its existing strengths to develop leading domain-specific AI in crucial industries. Domain-specific AI’s unique developmental challenge is finding large and diverse training and testing datasets in their respective domains. Multiple European companies are already at the cutting edge of their respective domains, such as Novo Nordisk in life sciences and SAP in enterprise software, equipping them with uniquely high-quality data.

Ensuring European champions have the necessary computing access and human resources to develop AI could help cement their global industry leadership. Moreover, if the upcoming European Data Union Strategy announced by Commission President von der Leyen delivers on its goal of streamlining data sharing, other European companies would also be able to benefit from the wealth of domain-specific data generated by Europe’s market leaders.

In addition, the EU is uniquely well-positioned to develop a booming AI assurance industry. Companies that provide the software, hardware, and services for other businesses to build and deploy well-maintained and legally compliant AI systems have a fast-growing market niche. AI assurance firms can already secure a billion-euro market globally, with expected market growth rates in the hundreds of billions globally by 2030. Operating under the world’s first binding international AI treaty, EU businesses can leverage their experience with AI risk management to develop audit, risk advisory, and other assurance tools. European AI assurance companies also have other key assets that may prove valuable in the industry. Access to mobile and world-class AI talent, a strong traditional audit industry, and leading research universities provide aspiring EU assurance businesses with human resources and avenues for intra and cross-sector cooperation.

There is still time for Europe to shift approaches to secure a competitive spot in global AI development. But we must stop wasting resources on half-hearted and fragmented efforts. Instead, the continent needs to build on its strengths and focus on efforts to collectively develop competitive GPAI models, domain-specific AI systems, and a globally leading AI assurance industry.