VISION REVISED — France's worst unrest in years and Marine Le Pen's widening poll lead show the vision was too thin on member-state politics, so it now pictures sovereigntist governments bargaining hard inside the Union; final summit text keeps Ukraine's accession untimed, leaders regret the US diesel waiver, and UK talks do not move the picture.

Vision for 2036

This living baseline was revised on 19 October 2026; it is not a prophecy. It pictures the European Union on 19 October 2036. The horizon moves with the publication date: always ten years on. Nothing here is the necessary path of European history.

Institutions and enlargement

The Union of 2036 rests on the same skeleton: a Commission that proposes and administers, a Council of governments, a directly elected Parliament, and a Court whose judgments bind. A treaty rewrite remains possible this decade; this text does not assume one is ratified. The tools are passerelle and bridging clauses, accession treaties with probationary safeguards, and political pledges to limit new vetoes without treaty change.

Enlargement is still the largest open constitutional question. The mid-2020s queue will not move as one convoy. One or more accessions by 2036 remain plausible; so does further delay. Agriculture, cohesion money, free movement, and foreign-policy decision rules remain the brakes that matter. Any new member is likely to face a decade of heightened scrutiny, suspended rights if backsliding returns, and staged entry into sensitive policies. Keeping unanimity on tax, foreign policy, and treaties would still leave a larger Union slower. Extending qualified-majority voting would make it more capable, and harder for a government that loses the vote. Partial movement on both questions remains the central case. The open fight is whether it arrives through the package now circulating, a thinner compromise, or not at all.

The euro area is the same size or larger. Leaving the Union is not the central case. Differentiated integration — Schengen, the euro, defence projects — is how the Union lives with states that will not move together. Pre-accession participation in selected programmes is part of that differentiation, not a substitute for membership.

Economy, technology, and artificial intelligence

The single market is still the Union’s main source of weight. Some income gaps have narrowed; others have stuck. Industrial policy is more explicit than before the early-2020s shocks. Subsidies and joint projects for chips, clean technology, and defence production are ordinary tools. They have not produced one industrial state. State-aid fights continue, and the Commission is still the referee.

By 2036 artificial intelligence is ordinary infrastructure in firms, offices, and campaigns. The AI Act’s outline — prohibitions, risk classes, transparency — has held, with guidance doing the daily work and enforcement uneven, as with data protection. Europe is a hard place to deploy high-risk systems and a secondary place to train the largest models. Assume neither a growth miracle nor a collapse of employment. The argument is who keeps the gains, which regions lose routine work, and whether schools and social insurance keep up. “Digital sovereignty” names several policies, from cloud procurement to export controls, not a completed state.

Ageing, defence, and energy all draw on the same public budgets. Joint borrowing may be repeated for a named crisis; it is not an ordinary fiscal union. Budget fights over enlargement start long before accession day.

Defence and the security posture shaped by Ukraine

Security in 2036 is still shaped by the war that began in 2022, whether that war is then active, frozen, or ended. This text does not choose an ending.

Spending and production are higher than in the early 2020s. Ammunition, air defence, and drones are industrial problems. Stockpiles are only partly rebuilt. NATO is still the alliance most members rely on. The EU does more on money, procurement, military mobility, and industry. There is no European army, only closer national forces, more joint orders, and an expectation — ahead of the equipment — that Europe can act if the United States hesitates.

Ukraine is inside that order as a member, a late-stage candidate, or a heavily armed partner. Which of those depends on the war, on reforms, and on whether governments will pay. Roadmaps raise the political cost of indefinite delay; they do not decide the war or the budget. Russia is not a partner and not a closed file. Sanctions, separated energy systems, and deterrence remain the tools. A settlement is asserted more often than it is shown.

Climate and energy

The live argument in 2036 is about 2040. The 2030 targets are a mixed record. Electricity is cleaner: more wind and solar, nuclear where it was kept or rebuilt, more wires and storage. Gas is smaller and not gone, above all in industry and in cold, still weeks. Bills are no longer set by the 2022 shock. Grids, buildings, and factories still spread the cost unevenly.

The carbon border adjustment is a normal trade tool, disputed abroad and used at home. It has not greened world industry on its own. Heat, flood, drought, and coasts take more spending than early speeches allowed. Farming and land use are still among the hardest files. Climate policy and rural politics have not made peace.

Rule of law, democracy, and the home front

Some governments, members and candidates, still weaken courts, public media, or the running of elections. Cases, funding conditions, and political pressure have mattered, and they have been slow. Some backsliding has been stopped. Less has been reversed than the treaties suggest. A member that defies the Court and keeps its vote is still a real risk. So is repair after an election. Safeguards aimed at new members do not by themselves repair an old member that digs in.

Campaigns run on synthetic media. Platform duties, public-service journalism, and readers who pause have done more than any ban on lies. Resilience means a government can lose, a reporter can publish, and a court can embarrass a minister.

National politics is still what most people follow. Prices, schools, hospitals, and austerity budgets move more votes than any treaty, and protest over them recurs, sometimes violently. Trust in the Union varies by country and usually trails attachment to travel, trade, the passport, and the currency. Sovereigntist parties lead or join governments in several large member states; a founding member run for a term by a party that wants a "Europe of nations" is a central possibility. Such governments stay in and bargain harder over net contributions, slow enlargement and Ukraine money, test the primacy of EU law at home, and pull migration and climate costs back to national choice. The Union absorbs this as a slower, more conditional bargain. The Parliament matters most over money or a crisis.

The United States, China, and the neighbourhood

The Atlantic link is a working alliance that quarrels over trade, technology, and burdens. Military dependence on the United States is lower than in 2022 and higher than communiqués say. A deep American withdrawal would break this baseline. So would a return to easy reliance.

China is a major market and the main systemic rival. Trade continues. De-risking has been selective: some investment screened, more care over critical inputs and sensitive technology, no general split. Full separation would land on households, and governments have not accepted that. Climate cooperation exists and is thin.

The neighbourhood is where the Union matters most and looks least tidy. Accession, association, and migration deals sit together. The United Kingdom is close and limited: useful on security and on some markets, outside the institutions. South of the Mediterranean, energy, border control, and aid buy cooperation and create dependence. Deals reduce some irregular arrivals. They do not erase the reasons people move.

In 2036 the Union is a regulatory power, a large market, and a regional security actor exposed to the world economy. It is strongest when it can condition access to its market or its funds, and weakest where force it does not control decides the result.

How to read this revision

Each claim can be kept, narrowed, or dropped by a later weekday edition. If the page starts to sound like fate, it has failed.