AI roll-ups: the best succession solution for Europe

For the first time, more German owners plan to close their business than to hand it over. The businesses are healthy; the successor is missing. An AI roll-up solves the problem from the other side: not one buyer per company, but one buyer for ten who rebuilds the processes.

Europe has a succession problem that is bigger than any downturn, and it is almost never named as such. KfW's succession monitor for 2025 shows for the first time that more German owners plan to close their business than to hand it over: about 569,000 intended closures against 545,000 intended transfers by the end of 2029. One in four mid-sized companies wants to shut down once the senior generation leaves, around 114,000 a year. The most common reason is not the state of the business. It is the age of the owner.

That is the decisive point. These are not sick companies disappearing. These are businesses with customers, employees and margins, vanishing because nobody is there to take them over. In June 2026 the European Commission renewed its recommendation on business transfers after 32 years and warned that millions of companies will be affected over the next decade. Eurochambres estimates that around 150,000 businesses a year across Europe fail to find a successor.

Why the classic answers are not enough

All the existing solutions start at the same place: they look for one successor per company. The daughter, the plant manager, the founder from the succession exchange, the management buy-in. Each of them needs a person willing to take over an entire company with all its risks, and with the capital to do so. In 2024 the German chambers of commerce counted almost 10,000 owners who wanted to sell but only just over 4,000 prospective buyers. The maths does not work, and it gets worse with every cohort.

There is a second problem that never shows up in the statistics. Even when a successor is found, they usually take over a company whose processes live in the founder's head and in spreadsheets. The clerk who has checked invoices for 22 years knows every exception. The founder knows every customer. None of it is documented. The successor buys a company that only works as long as the old heads stay.

What an AI roll-up does differently

A roll-up reverses the logic. Instead of looking for ten successors for ten businesses, one investor acquires the ten and combines them into a group. This is not new. Private equity has done it for 30 years under the name buy and build, and the numbers support it: a study by BCG and HHL Leipzig found an average return of 31.6 percent for buy-and-build deals against 23.1 percent for standalone acquisitions. Almost one in two PE-backed companies in Europe made add-on acquisitions between 2019 and 2024.

What is new is what AI does to this logic. So far the gain of a roll-up lay in purchasing, overhead and financing. The actual work stayed as it was. An AI roll-up starts where a service company's largest cost block sits: in the work itself. Capturing documents, writing quotes, scheduling, answering queries, keying data into the ERP. These are tasks, not jobs, and AI can now take on a large share of them. In a single business with 40 employees the rebuild is not worth it. Nobody is there to build it. In a group of ten businesses with the same processes it pays ten times over.

In the United States this has become an asset class of its own. General Catalyst has earmarked around 1.5 billion dollars to buy accounting firms, call centres, property managers and IT service providers, and builds the AI before it buys. Thrive Capital launched Thrive Holdings with more than a billion dollars for the same approach and brought OpenAI on board as a partner. Long Lake, a platform from this circle, reached 100 million dollars of EBITDA in under two years according to industry reports, and in 2026 agreed to acquire American Express Global Business Travel for 6.3 billion dollars.

Why Europe is the better playing field

The American vehicles buy in markets that are already consolidated. Europe, and the German-speaking region in particular, has three properties that make the approach stronger here.

First, the market is fragmented like no other. Cleaning, building services, logistics providers, inspection services, property management: in almost every sector there are thousands of owner-managed businesses with one to five million euros of EBITDA that have never seen a buyer. Second, the supply is there. The closure wave means owners want to sell and no longer wait for the top price; asking prices have risen only about ten percent in real terms since 2019. Third, the skills shortage is AI's strongest ally. In a business that has not found a clerk for three years, nobody needs convincing that the machine should capture the invoices.

The rule: decompose, buy, combine

An AI roll-up fails at the same point as every AI project: when a tool is placed on top of old processes. MIT measured in 2025 that 95 percent of corporate AI pilots deliver no return. Whoever buys ten businesses and then rolls out Copilot owns ten old companies with a new licence bill.

The order decides. First the work is decomposed: every role in the target business is broken into its tasks, and for every task it is decided whether a person or a machine takes it on. That is my Task Architect method. It runs before the signature, not after. Then you buy, and you buy the business whose task profile fits the platform, not the one that happens to be cheap. Then you combine, on one process standard that applies to all. The exceptions that only one person used to know become rules in the system.

What remains is what makes the business valuable: the customers, the people, the name, the location. What changes is the work. The owner gets a buyer who otherwise would not exist. The employees get the tasks they were hired for instead of keying in receipts. And Europe keeps businesses that would otherwise have quietly disappeared.

What I do with this

For a long time I stood on the other side and built AI companies for corporations. With Generation Tech Partners I now buy the businesses that need this technology most and have the least access to it. Subsidy programmes and succession exchanges do not close the gap. Buyers close it, buyers who know how to rebuild the work.

Who pursues the model worldwide is listed in the map of AI roll-ups: platforms, capital providers and studios with source and confidence.

Sources: KfW Research, Nachfolge-Monitoring Mittelstand 2025 (Fokus Volkswirtschaft No. 526, January 2026) · European Commission, Recommendation on business transfers, 23 June 2026 · Eurochambres, 10 Suggestions to Support Business Transfer in the EU, 2025 · DIHK, Report Unternehmensnachfolge 2024 · BCG and HHL Leipzig, The Power of Buy and Build, 2016 · PwC based on Gain.pro, add-on activity of European PE portfolios 2019 to 2024 · PitchBook on General Catalyst, Long Lake and Amex GBT, 2026 · MIT, The GenAI Divide, 2025