Draghi disappoints on digital

Most of the slowdown in productivity growth in the European Union, compared to the United States, is because Europe is falling behind in the digital revolution. Europe does less well than the US in creating tech companies, deploying digital services and R&D investment in digital innovation. High regulatory compliance costs, weaknesses in leveraging digital data as a production factor and insufficient private investment in start-ups are additional contributing factors. 

Former European Central Bank president Mario Draghi’s report on the future of EU competitiveness, commissioned by the European Commission and published on 9 September, goes straight to the heart of these woes. But his policy recommendations – which will inform the Commission’s strategy for the next five years – are less convincing. 

His analysis starts from a very narrow measure of the digital industries, representing only 5.5% of EU GDP. That fails to recognise that digital is a general-purpose technology that is not confined to a particular sector. Consequently, Draghi’s recommendations focus on digital hardware and infrastructure, including telecoms, chips and cloud infrastructure – despite the fact that the report actually presents evidence that growth in digital content services (euphemistically labelled as “over-the-top” telcos) are a far more important economic driver. 

Draghi recognizes that EU telcos do not have the financial resources to keep up with their US and Asian counterparts, largely because of the EU’s own telecoms rules. Reducing regulatory constraints in pricing, national segmented markets and cross-border merger restrictions would help significantly to overcome this disadvantage and enable catching-up with the US. But most US digital R&D and investment comes from a handful of big-tech firms, not telcos. 

On cloud computing services, for which the EU market is dominated by US hyper-scalers, Draghi blames high EU electricity and real-estate costs. But US players pay the same costs when they build server farms in the EU. Draghi’s report eventually concedes that the main disadvantage for EU players is poor offerings of complementary platform services, including software and business services.

He proposes consolidation of small EU cloud providers into larger entities, and additional financing to expand cloud capacity, but that will not solve the lack of complementary services. It would be better to learn from successful privately financed US start-ups such as SnowFlake and Palantir, which have succeeded in carving out significant market shares in data value-added services, building on top of hyper-scaler cloud infrastructures. Their EU counterparts could try something similar, without public money or policy changes.

Lack of local hyper-scalers also explains the EU’s weak position in AI training and deployment, which in turn represents a threat to EU leadership in many digital services and robotics. Draghi recommends EU public investment in AI model training and deployment infrastructure, starting from the existing European High-Performance Computing network. But again, this focus on hardware loses sight of the overall business-services ecosystem that is required to build and operate successful AI models and services. Draghi acknowledges the exponential growth in AI infrastructure and model training costs, but does not seem to worry that this will soon put AI investments out of reach of EU budgets. Developing EU private equity and venture capital markets to finance AI start-ups would be a more promising approach. 

Draghi offers an alternative AI policy route that revolves around sectoral AI models that could be adapted to the needs of EU industries. But they do not require huge investments in AI infrastructure and model training. Specific AI application models could build on existing and pre-trained open-source AI models, to be complemented with proprietary industry data. That is perfectly feasible with smaller AI infrastructure and is already happening. There is no need for additional public financing. Policy measures to facilitate access to proprietary datasets might be needed, however. Voluntary data contributions have not been very successful to date. 

The report touts regulation-compliant EU cloud infrastructure as an advantage for European AI start-ups. But strict enforcement of EU copyright and privacy regulations would make it very difficult and prohibitively costly to assemble the huge datasets required for AI model development. Draghi rightly advocates harmonised and simplified implementation of the EU general data protection regulation. That would go a long way towards bringing down compliance costs for digital services start-ups. 

Draghi’s emphasis on hardware and telecom infrastructure thus repeats an old EU digital policy mantra that contributed to the EU missing out on the digital services revolution. The same is true for his focus on public investment rather than accelerating private investment. He does advocate pro-competitive regulatory changes rather than the precautionary approach that pervades most EU regulation but still it is disappointing that an economist of Draghi’s standing underplays the importance of economics and business models that drive digital innovation and ensure the success of firms.