Europe is Winning the AI Small Data Race

US tech might need to catch up if the European strategy pays off.

Source: European Commission

In the artificial intelligence contest, US firms dominate through massive proprietary data lakes and compute farms. Europe is pioneering a radically different approach focused on compiling accessible and useful small data.  

The continent’s small data program represents a direct counter to the “winner-takes-all” market dynamic of the US, where dominant tech companies hold the vast majority of market share. It aims to extract maximum value from common data sharing standards for 14 sectors. The European strategy unlocks access to a 130-million health patient network. It enables farmers to share data points across 400 data-sharing initiatives on soil and crop information. And it allows financial institutions to securely exchange standardized customer data across borders. 

Potential benefits are immense. Vehicle telematics data could upend traditional insurance by personalizing premiums based on driver habits rather than demographic factors.  A new maintenance-as-a-service industry could be based on real-time device feedback. Traditional one-size-fits-all financial services can be replaced by services that leverage cross-sectoral datasets — from energy consumption patterns for green mortgages to real-time machinery performance.  

Europe’s data ecosystem promises significant advantages in deploying the next generation of digital initiatives. The digital euro project will leverage novel payments infrastructure to propel financial diversification. Smart infrastructure investments will be able to utilize sectoral data spaces to optimize everything from 5G networks to precision agriculture. Europe’s data economy is expected to be close to 6% of GDP by 2030, growing by close to 50% from 2023. 

While critics often point to overregulation holding back European innovation, the data strategy leverages a complex web of regulations. The Data Governance Act establishes trust-building mechanisms for voluntary data sharing. The General Data Protection Regulation’s privacy protections set data minimization principles to ensure focused and high-quality datasets. The PSD2 payment rules allow open banking infrastructure that mandates secure access to financial data, and the Digital Markets Act’s anti-gatekeeping provisions prevent data monopolization. 

The main driver behind these opportunities is the overarching Data Act, which comes into effect in September. It establishes trust-building mechanisms for voluntary data sharing, while encouraging data minimization to ensure focused, high-quality datasets.  

New research shows that dataset size and compute rapidly reach diminishing returns, and future effectiveness gains are to be found in few-shot learning techniques. Europe’s 14 sectoral data spaces are purpose-built to provide curated, high-quality, domain-specific datasets, ensuring AI practitioners can fine-tune state-of-the-art models rapidly and cost-effectively — allowing EU businesses to remove the main competitive advantage that US businesses have — mass data. 

The European small data vision does not represent a panacea, and the continent finds itself behind in leveraging machine learning and data analytics. Not only is EU investment in AI just a fraction of that of the US and China, but the current digitalization disparity between EU member states may only widen, deepening the gaps in innovation and economic competitiveness.  

Most painful may be the compliance trap. The regulation enabling the small data market is also capable of hampering it, as companies spend more resources navigating legal requirements than innovating. European rightsholders are demanding mandatory copyright payments for data mining. Critics accuse the landmark AI Act of being premature, and it is bogged down in debates about what its provisions mean. The Act’s implementation is being delayed by at least a year, adding regulatory uncertainty.  

But US AI efforts also display weaknesses. The US suffers the failure of a federal privacy regulation, which leaves a patchwork, unwieldy, and unfriendly environment for sharing data. By 2025, 20 states have implemented data privacy laws, each with different requirements and standards for data handling, consent, and sharing.  

Europe’s small data strategy is generating new markets that are limited under American volume-dependent approaches. The Data Act protects small companies from unfair contractual data sharing conditions. It prevents lock-in and exploitation, but also encourages competition through provisions for easily switching cloud service providers.  

Much depends on Europe’s ability to drive and confirm the commercial utility of the Data Act in the upcoming year. But the foundation for the small data revolution has been set. 

---

By Ēriks Selga. Ēriks Selga is a Fellow with the Tech Policy Program at the Center for European Policy Analysis. He is a former Digital Innovation Baltic Fellow at CEPA and is currently a Council member of the Latvian Competition Council. Article and pictures first time published on CEPA web page. Prepared for publication by volunteers from the Res Publica - The Center for Civil Resistance.