Reforms to boost AI in the EU: an overview of TBI's ideas
European leaders are wrestling with digital sovereignty. Meanwhile, a new paper from the Tony Blair Institute for Global Change lays out how the EU can stay competitive in AI by fixing rules, funding the essentials, and pushing adoption with real incentives.
The approach is practical: simplify regulation, unlock capital, build compute and energy capacity, speed up adoption, and export the European digital stack with intent. Here's what matters for public leaders and why it's actionable now.
The problem Europe has to solve
Markets are fragmented. Regulations are complex. Scaling across borders takes time and money most firms don't have. Add chronic underinvestment and some of the highest energy costs worldwide, and you get a system that slows AI down instead of speeding it up.
The paper also challenges a common myth: digital sovereignty as self-sufficiency. Trying to build everything domestically is costly and unrealistic. As the authors put it, true sovereignty is about leverage and choice - not owning every piece of the stack.
Solution 1: Make the single market work for AI
First, fix the rulebook. The current digital framework makes it expensive to innovate and scale. The AI Act shows how hard it can be to move fast across 27 countries.
The paper proposes using a Digital Omnibus to simplify reporting, harmonise enforcement, and streamline rules across member states. It also calls for fast-tracking a Savings and Investment Union and a 28th regime to speed cross-border scaling, plus a continent-wide stock exchange to deepen liquidity.
Labour markets need modernising, too. Mobility, flexible contracts, and faster hiring are essential if Europe wants to keep builders and operators - not watch them leave.
Solution 2: Build compute and secure affordable energy
Compute is the new infrastructure. The EU's current efforts (like EuroHPC and gigafactory plans) are steps, but scale and private investment will decide outcomes. The paper argues for conditions that attract large private capital into AI infrastructure.
Energy is the other constraint. The proposal: accelerate an energy union and launch a continental program to coordinate new nuclear, speed up permitting for renewables, expand grids, and build storage. Without cheap, reliable energy, AI won't scale where it matters.
Solution 3: Speed up adoption with data, clarity, and talent
Adoption hinges on data access and trust. The authors back a trusted, interoperable ecosystem rooted in open-source innovation. They want the upcoming Data Union Strategy to open strategic datasets, while procurement and compute-credit incentives drive demand where it counts.
They also propose sector-specific AI adoption packages with clear rules and fast-track approvals. That reduces ambiguity for hospitals, factories, banks, and public agencies that are ready to deploy but stuck waiting.
On talent, the message is blunt: align universities with a modern innovation economy. Reform career paths, increase autonomy, pay competitively, and enable fluid movement between academia and industry. A fair caveat: extra autonomy and funding should be earned - reward institutions that deliver, and push laggards to improve before they get more.
Solution 4: Export the European digital stack with intent
Europe needs a coordinated strategy to export its digital government stack - identity, payments, registries, and standards - backed by finance and technical expertise. That includes regional tech hubs staffed by technologists and diplomats who can help partner countries adopt European solutions.
The paper suggests a European Investment Acceleration Mechanism to help investors work through regulatory requirements on strategic projects. Critics say reforming existing tools like Global Gateway is hard, but there's no credible alternative on the table - so make the current system work better and move.
What public leaders can do now
- Set up cross-border regulatory sandboxes aligned with the AI Act's intent, not just its legal minimums.
- Pilot compute-credit programs for SMEs and public bodies to kickstart real AI use cases.
- Publish sector-specific guidance with expedited approval paths (healthcare, manufacturing, finance, public services).
- Map strategic datasets and define clear access rules with audit trails and privacy by design.
- Create national task forces to cut energy permitting times for renewables, grids, and data-center co-location.
- Deploy tech-diplomats and solution engineers in priority regions; pick 2-3 flagship government-stack exports and fund them end to end.
- Modernise procurement: outcome-based contracts, open-source preference where viable, and faster vendor onboarding.
- Tie university autonomy and funding to measurable research outputs, spinouts, and industry collaboration.
- Back a deeper capital market: support the 28th regime concept and a continent-wide stock exchange to help scale-ups list at home.
Why this matters
Europe has the talent, institutions, and values to lead. But without simpler rules, more capital, serious infrastructure, and faster adoption, outcomes will lag.
The path forward is clear: make the market work, build the base layers, reward performance, and take Europe's best ideas abroad - with scale and speed.
Resources
Upskill your public-sector team
If you're building practical AI skills across roles, explore curated training by job function: Complete AI Training - courses by job.
Your membership also unlocks:
