Europe's turn to build the vertical-AI winners

Europe did not miss the wave in the built environment

Not in the built environment. The two largest exits in construction AI so far are both Norwegian. Spacemaker, an Oslo company of 115 people whose software explored how to get the most out of a building site, was acquired by Autodesk for $240 million in November 2020. Five years later the pattern repeated one level deeper: Consigli, also from Oslo, had automated parts of buildings engineering itself, and AECOM paid about $390 million for it in November 2025. A global engineering consultancy buying an AI company, not the other way around.

The capital keeps arriving. In June 2026 Endra, a Stockholm company that automates mechanical, electrical and plumbing design, raised a $50 million Series A led by Andreessen Horowitz, taking it to roughly $75 million raised within two years of founding. One of Silicon Valley's best-known funds now underwrites Swedish engineering software.

It's a trend: contech funding more than doubled

A trend, and a measured one. Cemex Ventures, which publishes an annual count of construction-technology venture funding, recorded $3.1 billion across 325 deals in 2024 and $6.57 billion across roughly 337 deals in 2025. The money more than doubled in a year, carried above all by activity in seed and Series A rounds: the stage where the next generation of winners is decided. Cemex's own reading of 2025 is that investors became more selective and steered the capital toward digitalization, automation and AI: fewer experiments, more systems that carry real work.

The gap is measurable too. An analysis by Derek Montes in January 2025 counted 19 AEC technology unicorns worldwide: 11 in North America, 7 in China and India, one in Europe. So far, Europe produces the companies and America keeps them. Both Norwegian winners now sit inside US groups, and the standalone European giant in this vertical has not been built yet. That gap reads less like a verdict than like an opening.

Europe has the projects, the codes and the people

The projects, the codes and the people. Start with the codes, because they are the least obvious asset. European engineering is unusually codified: the Eurocodes and their national annexes, agency regulations for road and rail, standardized structures for describing materials and works. A Swedish infrastructure deliverable must hold against Trafikverket's regulatory documents, AMA's requirement texts and thousands of pages of project-specific tender documents. For decades this depth has been read as European sclerosis: rules as friction.

For AI, it is the opposite. A system that produces or reviews engineering documents needs correctness criteria, and in Europe those criteria are written down, versioned and public. The rulebook that exhausts a human reviewer is exactly the ground truth a machine can be held against. Standards depth is not the obstacle to automating engineering work; it is the raw material. The same code clause can mean different things in different editions, and a national annex overrides the volume it annexes. Depth like that punishes skimming, and rewards a machine that reads everything.

The demand side is just as loaded. Sweden alone fixed a national transport plan of 1,171 billion kronor for 2026–2037 in April 2026, while document volume per project keeps growing and human engineering capacity stays flat. And Sweden is not alone: Norway adopted a transport plan of 1,308 billion Norwegian kroner for 2025–2036, and Germany's parliament approved a 500 billion euro infrastructure fund in 2025. The pain, the talent and the codes sit in the same cities.

The regulated road-and-rail seat is still open

Every exit so far comes from the buildings side. Spacemaker optimized building sites. Consigli automated buildings engineering. Endra designs the mechanical, electrical and plumbing systems inside buildings. The mechanism, software that carries a whole engineering task rather than assisting at the margins, is proven in the discipline next door. Regulated road and rail infrastructure, the most document-heavy and most codified discipline of all, has no equivalent winner yet. Its deliverables are pure document work, investigations, plans, technical descriptions and review rounds, produced against the densest rulebooks in the industry.

That opening will not stay open long, and it will not be filled from somewhere else: the codes are national, the languages are small and the buyers are here. The timing looks like Oslo in 2020. Yesper is the AI civil engineer for construction and infrastructure. That winner is what we are building, from Stockholm.