This week a founder from Germany reached out to me. He had heard me on a podcast about AI rollups.
He and his team are building an AI-native managed IT service for fast-growing companies. And they are asking themselves the question that a lot of AI-native services founders are asking right now: do we grow organically, or do we roll up MSPs and transform them?
I have been in touch with MSPs since I was twelve. My father built techbold: 18 acquisitions, around 180 employees, more than 1,200 customers, sold to accompio in 2026. At our kitchen table the only topic was MSPs, MSP rollups and how acquisition conversations go. I then built M&A strategies for MSPs myself. And in the last two weeks I talked to over 30 MSP managing directors.
So before I answered his question, I gave him my ground truths.
Ground Truths on MSPs
A) Internally, the MSP’s own work is already automated very strongly. The whole ticketing topic, whether they have their own RMM or use NinjaOne or whatever their tech stack is. So your AI won’t unlock much margin inside the MSP. And where an MSP has no management structure, that’s not a reason to walk away. Those are often the best targets, because that’s where the value gets created: through the processes, standards and tooling the buyer brings in. Most of techbold’s 18 acquisitions were firms like that. The lever there is still standardization, not AI.
B) Externally, Microsoft is absorbing the low end of the portfolio. The partner program gets more and more unpleasant for MSPs, margins get lower and lower, and more of what MSPs resell and manage today gets offered by Microsoft directly. So that part of the revenue shrinks regardless of what you do.
C) That’s why we had the shift from IT service provider, to cloud provider, to MSP, to MSSP, and now the next shift is being announced: MIPs, Managed Intelligence Providers. Pax8 and these huge platforms for MSPs are saying an MSP will become the agent orchestration layer for its customers, the last mile delivery of AI. MSPs will be the multiplier of the labs. That’s the bet, and VCs like General Catalyst and Thrive have already invested in AI-native MSPs.
D) I believe very few MSPs will make the jump to an MIP (Managed Intelligence Provider), because the talent an MIP needs and the talent an MSP has today are not the same. Someone who sets up root servers in their sleep isn’t automatically the person who brings AI into a company. For that you need people who understand how sales, accounting and processes work inside a firm, and being able to code is not the requirement. I expect that competence at the firms that implemented the accounting and ERP systems: an agent that can see pricing, payroll or margin in the ledger has to be governed by the people who built the data model, and that is what managed intelligence really is. Not at MSPs, and not at AI-native teams without operating experience.
E) Where I do see the next step for MSPs is security. More and more AI-generated code is being written, and more and more security breaches happen. Securing what the customer has internally sits right where the MSP already is.
So the question is: what do you actually buy when you buy an MSP? And now to rollup versus no rollup.
Photo Credits: techbold Technology Group AG
The Rollup Math
MSP is a market that is being consolidated from all sides right now. But the hot market is for platforms from around €10 million revenue, not for small MSPs. The groups buy finished pre-consolidators, not single firms.
He said small MSPs are their scope. They are looking at companies with 8 to 10% EBITDA margin.
I understand that, but why is it relevant for them to know, that big platforms are buying up the market right now? Because if they become a pre-consolidator (buying companies too small for a normal PE), they will only be bought by such a group in 4 to 5 years if they reach platform size. That’s the rollup argument, and at the same time the risk. It only pays off if you get there.
And small targets are not valued on EBITDA multiples. They are valued on contribution margin, in German DB1: revenue minus direct delivery cost, so the techs, the licenses and the hardware. In the DACH deals I see, that’s somewhere between one and two years of DB1, depending on how recurring the business is. Almost all deals in pre-consolidation rollups are asset deals: you take over the order volume, the customer contracts and the employees. The majority is paid in cash at closing, and the rest is variable over several years, depending on the DB1 you make with the customers you took over. The variable part can swing a long way in both directions.
Let’s take an MSP with €3 million revenue and 35% DB1, so around €1 million DB1 a year, with mostly recurring business. The price is somewhere between one and two years of DB1, and only part of it is cash at closing. The target isn’t cheaper than an EBITDA multiple would make it. The method is different: less cash is due at closing, because the rest depends on whether you keep the customers.
Now the numbers nobody wants to hear. MSPs grow 2 to 5% a year. Integrating the back office takes maybe 5 points out of cost, and automating workflows with AI on top doesn’t add much, because the MSP’s own work is already tooled up and lots of AI implementation shops are already doing exactly that for MSPs. In the deals I know, the margin goes from 10% to 13 or 15%. And the exit multiple in five years is the real unknown. My guess, and it is a guess: 6 to 8x EBITDA in 2031 for a group with €5 to 8 million EBITDA and 40% recurring revenue.
Most MSP rollups I know started pre-consolidating in 2015, with targets between €1 and 5 million revenue. And that’s why now, about 10 years later, the groups come that buy up the pre-consolidators. If you start a pre-consolidator now, the question is whether you reach platform size in time, and what a platform is worth when you get there.
The Seller
Rollups don’t work because of the multiple. They work because of who is selling.
Around 40% of the sellers techbold bought from were in their sixties and wanted to solve their succession. The price was important, but it wasn’t the deciding factor. It was about loyalty to their employees and to customers they had looked after for 20 years.
The second group were technician-founders in their forties who had built a company bigger than they wanted to run, and wanted to get back to the technology they loved.
If you understand what these two groups want, you understand how to buy.
The First Weeks
In this business the bond is between the customer and the employee, not between the customer and the brand. The techs are the retention. So you win the employees over before signing. Right after signing you call the top 50% of the customers: same contacts, same prices, same portfolio. The brand you can change within a few weeks, and nobody will care. An SMB customer is busy with their own business and doesn’t want any change from their IT.
The first weeks decide everything.
What Stays and What Goes
If you look at an MSP right now, it offers 4 to 5 product types. A bit of cloud. The majority is probably billable hours, network installation etc. The ones that are a bit further have cybersecurity. And the rest is reselling products.
I asked the founder what he thinks will fall away. His view: the whole ticketing topic will eventually be covered by Microsoft, software reselling and licensing and device procurement are not interesting, and the whole workplace topic will disappear. What will stay is security, NOC and SOC.
Exactly. So why would you go in with 4 forward deployed engineers to automate the whole workplace topic? Instead you could go in and ask: how can we shift your product portfolio so that you only do high-margin business? So that you have organic growth because you offer products where you don’t compete directly with all other MSPs, but go much more into security, where the market grows by itself.
And I think that’s the crazy thing. If you target the same market, but with a different product, the market grows by itself and you can create momentum with that growth. If you come with a product that has existed for 30 or 40 years, the market doesn’t grow anymore. Then it’s a displacement market, and organic growth there means 2 to 5% a year.
But it doesn’t happen in three months. The way to a security provider with a 24/7 SOC and ISO 27001 took techbold 12 to 18 months, and that was fast. An own SOC only pays off from around 100 to 150 employees. A 30-person MSP can only resell white-label MDR.
That’s why the central SOC across the group is one of the strongest arguments for the rollup. It’s the scale effect a single MSP never gets.
What the Bridge Costs
And here is what nobody puts on the slide. To get from a 30-person MSP to a group that can carry its own SOC, you need five to seven deals, two to three years and €10 to 15 million of capital before the strongest argument produces a single euro.
That’s where most of these cases fail. Not on the idea. On the funding.
Go-to-Market Is the Key
The MSP market is a displacement market. There is hardly any good customer with a good basket that doesn’t already have an IT service provider. You have to go in and say: we are better because of XY. It’s not a growing market by nature, except maybe the 3 to 4% of companies that are new.
How do you win in it? Cold outreach, a meeting, an IT audit. techbold did more than 600 IT audits in the Austrian Mittelstand in eight years. The audit shows the weaknesses and the quality of the current provider, and the offer and the collaboration are built on that.
It’s about whether you can realistically sell a vision that you can take 10 or 20% of the market with your story. You need that on both sides. If you don’t roll up, you need the story that you are the alternative to all the MSPs. If you roll up, you need to show the MSPs a vision: you won’t manage to make the change alone, but we will do it with you. I think both make sense, but then I would make 90% of the team go-to-market and M&A experience, and less product.
The Swamp
His last question was about the German SMB market. Most of them are still very far away from AI. Is it interesting, as their IT service provider, to go in and push their AI transformation?
Right now most new companies are founded with one slogan: we automate your processes with AI. All the individual software service providers became AI workflow automation firms. Lots of young people became AI automation firms.
The question is: do you also want to go into this swamp, or do you just want to profit from it?
If you are an MSP rollup, you actually have extreme customer power. So the question is whether you offer it as another product, white-label it and let someone else do the fulfillment. Doing the fulfillment internally would be too much loss of focus, I think. Or you say: that would dilute our go-to-market pitch a lot and move us away from security. That’s a strategic question that needs some thought.
And there is a segment problem he has to know about. He builds for fast-growing companies. A rollup of classic MSPs brings him SMB customers with on-prem and Microsoft. Going organically into the German SMB market, he didn’t think was realistic, and I don’t either. Through a rollup, I think the SMBs are actually the better business, because those customers stay ten years and longer. But it’s a different sales motion and a different product.
So my answer to him wasn’t rollup or no rollup. I am bullish on the operator who already has the sellers and the capital. I am bearish on a tech team starting a pre-consolidation from a deck in 2026.
If I did it, I’d start with the sellers, not the software.
Challenge me, I’d be curious to hear your thoughts.