Vertical AI rollups - New business models, and can they work in AEC-tech? Pt. 1

Roll-ups: some of my readers might have heard of it.

The concept is very compelling, mostly because it has the potential to generate a ton of "shareholder value", as we like to call it.

The concept is also not novel. In fact, it's been around for quite some time in Private Equity, Venture Capital's cousin.

So why am I writing about it?

Well, lately it caught the attention of "Startupland", and it's become a strategy that several founders are now pursuing and that (some) VCs are funding.

Take General Catalyst. The VC fund has recently secured $8 billion, with approximately $1.5-2.5 billion designated for their Creation Fund. The Creation Fund operates on a distinctly different philosophy compared to a traditional VC fund: it begins by incubating software companies with strong AI-driven foundations with seasoned entrepreneurs, and subsequently promotes expansion through strategic acquisitions (roll-ups), with the platform's proprietary software/AI being systematically implemented within the acquired organizations. In its initial phases, the Creation Fund supports acquisitions through equity investments, purposefully avoiding debt, but as the platform develops scale and operational resilience, debt financing becomes an available option to further enhance growth. Their ultimate objective is to establish industry-defining enterprises capable of generating exceptional returns through IPOs.

Given the recent interest from founders and VCs alike, I felt compelled to write something about it. Obviously, with an AEC-first perspective ;)

Before we dive in, let me first provide a super-short primer on roll-ups, for those unaware of the concept.

A primer in roll-ups

At its core, a roll-up or "consolidation play" is an investment strategy where a financial sponsor (typically a PE firm) systematically acquires multiple smaller companies within the same industry or market segment and combines them into a single, larger entity with enhanced scale, capabilities, and market positioning. HVAC, dental clinics, veterinary shops, etc., have all been subjects of roll-up strategies, particularly in the USA.

The strategy operates on a simple yet powerful premise: the whole can be worth significantly more than the sum of its parts (1 + 1 = 3).

In fact, roll-ups create value through multiple, reinforcing mechanisms. Multiple expansion represents a key driver, as larger companies typically trade at higher EBITDA or revenue multiples than smaller ones, creating immediate value through so-called "multiple arbitrage." This is particularly powerful when starting with a substantial platform company (e.g. trading at 7x EBITDA multiple) and acquiring smaller competitors at lower multiples (e.g. at 3x EBITDA multiple), as it gives the acquirer an immediate valuation bump in the newly acquired asset.

Operational synergies provide another value lever, as consolidated entities can eliminate duplicative costs in areas like administration, marketing, and R&D, while leveraging combined purchasing power with suppliers. Market position enhancement occurs as larger entities often command better terms with customers and suppliers, while potentially reducing competitive pressures within the industry.

Access to capital improves as consolidated entities typically enjoy better access to capital markets and financing options than their smaller predecessors. Strategic positioning becomes enhanced as scale provides opportunities to invest in capabilities, technology, or market expansion that would be out of reach for smaller entities.

The most fertile markets for roll-up strategies share specific characteristics, amongst which are high fragmentation with many small acquisition targets, weaker competitors ripe for consolidation, and business owners facing succession challenges who have incentives to sell.

So, what does this all have to do with VC, and why are startups going after it? Blame AI and vSaaS.

The AI-powered evolution: vertical AI SaaS buyouts

In the past months, I've witnessed the emergence of startups following a new approach, called by some investors "vertical AI SaaS buyouts".

What's the rationale behind it?

Very simple, but a tad different from the thesis PE funds have so far pursued.

Many industries where vertical AI software should theoretically capture significant value have proven resistant to technological adoption: startups develop fantastic products capable of significantly reducing costs and/or greatly increasing productivity, yet might struggle (to monetize) against entrenched cultural resistance and adoption barriers.

The vSaaS buyout thesis elegantly circumvents this problem through a two-pronged approach: first, build the software platform internally, optimized for operational efficiency; and second, acquire operating companies that become both customers and transformation targets. This way, you capture significantly higher value than by just selling software. Moreover, another reason for going after this strategy is that pure software plays will only get more competitive and commoditized as AI increases in power - significantly reducing SaaS value extraction for future players.

Let's have a numerical example. Consider a small legal services firm specializing in contract review that typically charges clients $2000 per contract and maintains a 25% gross margin - their costs per contract are $1500, yielding $500 in gross profit. The majority of these costs come from attorney time spent manually reviewing documents, a process that's both time-intensive and prone to human error. Now, imagine that this firm implements an AI contract analysis system. The technology can pre-review documents, flag potential issues, and handle routine elements automatically. This reduces the attorney time required by 50% while actually improving accuracy and consistency. The firm's costs per contract drop from $1500 to $750, and gross profit jumps to $1250 - a staggering 62.5%! They can even significantly undercut competitors and still maintain the same margin. Not bad!

Besides the pure value capture, there are other advantages to pursuing this roll-up strategy. First, as an operator of the business, you have access to primary data to train your AI models - data that might not be straightforward to come by as just a software vendor. Second, you know exactly what you need to build, which workflows to tackle, how to finetune your product, etc. since you're literally using it on a day-to-day basis to deliver value to customers.

All in all, in the age of AI, the opportunity is huge, particularly if the tech you build can reduce/replace human labor, therefore significantly benefitting the bottom line. In fact, it's specifically the AI dimension that adds unprecedented potential to the roll-up model.

Still, I argue this doesn't make sense in every industry, and that some are better suited than others to this model.

Not all industries are created equal

Not all markets are created equal when it comes to the "AI rollup" potential. Apart from the obvious characteristics (fragmented market, succession problems, lack of technology implementation, cheap multiples, etc.), the "AI-rollup" playbook is particularly powerful in specific cases.

Ultimately, these opportunity spaces can be identified by answering a pretty simple question: what are those sectors in which AI can automate a significant portion of key (emphasis here) workflows, hereby significantly reducing COGS/increasing margins? Moreover, AI shouldn't merely reduce costs but potentially generate new and/or higher revenue streams, e.g. when AI enables the delivery of products or services that substantially outperform what competitors offer.

Therefore, the nature of the core workflows of the sector is what ultimately determines whether it's a right fit or not for this strategy, in my opinion. While you can add software to improve both front and back-office processes, the greatest value creation occurs when you can insert technology directly into the core work of the business itself. It's industries where AI can actually "do the work", rather than merely support it, that offer exponentially greater transformation potential.

It follows that companies still reliant on manual, pen-and-paper workflows represent a low-hanging fruit. Particularly attractive are businesses where human knowledge work constitutes a major portion of operating expenses - these represent cost centers that can be streamlined through intelligent automation. For example, processes involving document review, data extraction, text-based communication, and materials creation can be dramatically improved through AI implementation. Businesses with employees primarily focused on rote process execution rather than high-touch advisory work are especially promising. Therefore, the ideal acquisition targets are ultimately companies with white-collar desk employees performing manual, repetitive, rules-based work. Companies with low margins where headcount expenses represent a significant share of revenue have the most to gain from labor-cost reduction through automation. Think insurance third-party administrators, or tax preparation services/accounting.

As this is an AEC-focused newsletter, you might wonder: has this model been adopted in AEC-tech yet?

Who's rolling up in AEC?

For the reasons outlined above, I think that most of the AEC industry is not a great space to "AI-rollup" in. Actually, let me clarify: it's not a great space to pursue a VC-backed AI-first roll-up strategy. A "traditional" PE roll-up will work just fine.

In fact, particularly for all those businesses with physical operations (home services, renovation, construction, etc), the extent to which a pure AI/software play can realistically provide a margin uplift is limited. You might successfully automate peripheral/backend processes - like quoting, scheduling, marketing, etc. - but the core physical activities will still require human involvement. Tech can help identify inefficiencies, recommend specific improvements, track implementation progress, etc - it's a valuable augmentation, but hardly one that allows you to 2-3x margins. It can still help increase revenues, e.g. by better allocating resources. However, this is the same tech and playbook that PEs already employ in their roll-up strategies (and less like transformative AI): it's tech that can already be purchased off the shelf, and relatively simple to build. For a significantly higher margin uplift, you would most likely need to develop hardware on top of pure software - a topic for another day.

That said, some VC-backed players exist in our space. They're not operating in "core" AEC verticals - rather in adjacent home services, for the most part. The model seems to have gotten particular interest in the USA.

Cabana Pools (fka Splash Ventures), is rolling up in the pool cleaning industry. By acquiring small, regional pool service providers, the company consolidates the fragmented market and builds a scalable, technology-driven platform. Cabana Pools integrates AI-powered tools for optimizing scheduling, route planning, and customer service, enhancing operational efficiency and customer satisfaction. The company uses smart sensors and data analytics to monitor pool conditions in real-time, allowing for proactive maintenance.

Founded in 2019 and based in San Francisco, Electric Sheep is pioneering the automation of outdoor maintenance tasks using AI and robotics (I know: I just told you it would be a topic for another day). The company's flagship products include autonomous mowing robots powered by their proprietary AI agent, ES1. ES1 utilizes large world models to enable reasoning and planning in unstructured outdoor environments, allowing robots like the RAM mower and Verdie, designed for edging and trimming, to operate seamlessly alongside human crews without on-site engineering support. Electric Sheep acquires traditional landscaping companies and integrates its technologies to enhance operations. This approach has led to significant growth, with revenues increasing eightfold since implementation. The company has raised a total of $21.5m to date.

PipeDreams (2020) is a tech-powered home services provider specializing in acquiring and enhancing plumbing and HVAC businesses. By integrating proprietary AI-driven technology, PipeDreams streamlines operations, optimizes scheduling, and automates inventory management, leading to increased technician earnings and improved customer satisfaction. The company retains the brand identity and employees of acquired businesses, respecting their established community presence. They claim they have achieved 100% organic growth within 24 months in one of their acquired businesses, which would be pretty amazing! They have raised a total of $39m to date.

Roofer.com , founded in 2022 in Dallas, leverages AI and drone technology to provide roofing services. They combine advanced AI algorithms with drone-captured high-resolution imagery to conduct precise roof scans, identifying damage, assessing remaining roof life, and determining necessary repairs or replacements. The company strategically acquired Bearded Brothers Roofing & Restoration, a major Texas roofing contractor known for significant projects like the San Antonio Marriott Riverwalk and thousands of residential installations. Currently, Roofer primarily focuses on residential re-roofing services while rapidly expanding its enterprise division that caters to multi-family apartment complexes and commercial properties. They have raised $7.5m to date.

On the other side of the pond, we have 1komma5° and the recently announced Reshape Energy.

1KOMMA5° is a German company founded in 2021, specializing in making buildings carbon-neutral by integrating renewable energy technologies such as solar panels, heat pumps, and charging infrastructure. At the core of 1KOMMA5°'s technological innovation is its proprietary energy management system, Heartbeat AI. 1KOMMA5°'s growth strategy is heavily centered on the acquisition of established energy solution providers across Europe. By integrating these companies, 1KOMMA5° not only expands its geographic footprint but also enhances its service offerings and technological capabilities. This approach addresses industry fragmentation by consolidating expertise and resources, enabling the company to offer comprehensive, one-stop solutions for energy renovations. They have raised over €450 million.

In a similar space, Reshape Energy , another German company founded in May 2024, is advancing energy renovations in commercial real estate through a strategic acquisition approach. The company acquires and integrates businesses across the building energy services value chain, creating a one-stop shop solution that simplifies energy optimization for property owners. This strategy addresses the traditionally fragmented and inefficient market, enabling streamlined processes and comprehensive service offerings. In the German market, Reshape Energy has acquired an energy consulting firm and a commercial solar planning, installation, and maintenance company. By integrating these services, Reshape Energy empowers property owners and tenants to reduce costs, enhance property values, and achieve sustainability targets. They have raised €5 million.

While some startups have successfully scaled through roll-up strategies in AEC-adjacent sectors, I believe that the AI component of the model (and its potential for workflow automation and margin uplift) remains limited in this space, making it a suboptimal candidate for this business model (VC-backed, vertical-AI rollup) in my opinion. Again, that's not to say that a "traditional" rollup does not work - quite the opposite!

Still, one question still remains open: should VCs even deploy capital for roll-ups to begin with?

Is this even a VC-backable opportunity?

As discussed, the opportunity seems huge, but size alone is not a strong argument for capital flowing into the space. Are roll-ups a good asset class for a Venture Capital investor, and should VC pile in? This is the question to answer, so let me share my 2 cents here.

What is happening in this case of "AI" rollups is that you're combining a high-multiple software business with a low-multiple service operations within a single entity. It obviously follows that you're not building a tech company, but rather a technology-enhanced service business. Therefore, you will not be rewarded with tech-like multiples, but need to anticipate eventually being valued as a premium tech-enabled business. These high-margin service companies are more likely to get multiples closer to service peers than software counterparts, because your software becomes captive to your service operations, sacrificing its potential as a neutral industry platform.

It's also crucial to understand whether it's better to buy small or large, which informs fundraising and the fit for early-stage VCs. Buying a large first business (the platform) gives many benefits, as we've seen. But specific to this case, a large target provides immediate access to proprietary data assets, and significant cost structures across which to amortize AI development investments (very important!). Moreover, it gives an instant multiple uplift whenever you buy a new add-on, and accelerates the path to meaningful scale and quicker returns while cutting the number of acquisitions required. Early-stage capital won't generally allow you to do that.

Besides, an important thing to keep in mind is that for this playbook, money is needed both for tech development and for acquisitions. So, who's providing it?

It's very hard (impossible?) to make VC equity money work for acquisitions. Think about this: VCs give x amount for y% of the company - let's assume $30m for 15%. If all that capital is used for acquisition purposes, the VCs would have paid almost 7x the book value of the acquired companies (since they only effectively get 15% "ownership" in the acquired companies through their investment, while paying the full amount). If the startup achieves 20x growth, the VCs would only benefit from a "meager" 3x - it makes no sense.

Hence, debt makes the most sense for acquisitions. You would need a lot of it.

However, first off, debt (at scale) is no amateur hour - it requires someone in the founding team with a deep understanding of the instrument. Second, servicing debt generally does not leave much room for R&D and its ongoing investments, unless the initial investments in AI/tech (financed by equity) are generating enough additional cashflows (read: unlock high enough margins + higher revenues) to pay both debt and the ongoing technological development (whether this is possible seems might be speculative, for now). The assumption is also that R&D spending funds the development of differentiated and highly innovative (read: hard to build) products. Otherwise, this approach would have no sensible differentiation and could be easily done by a PE firm with minimal investments - anecdotally, I know of some PE firms doing this already, by outsourcing their IT overseas (e.g. to India).

Last, let's not forget that since markets will value you based on cashflows (remember, you will get a service-business multiple with a premium, not a software multiple), the incentive will always be that of boosting EBITDA. This leads me to my last point: maintaining a strong competitive advantage vs. software vendors whose only business model is building and selling SaaS to your (traditional) competitors will become increasingly more difficult, and over time the financial motivation to develop and maintain superior software solutions weakens considerably. When the same tech becomes available off the shelf, there is no reason for continued R&D spending. And if you'll ultimately need to adopt the best external AI solutions anyway, why allocate resources to internal development instead of focusing on implementing and integrating the best third-party AI tools? We're going full circle to PE.

Therefore, making this model make sense for venture capital as an asset class is not straightforward.

Perhaps an answer could be found in more creative approaches to solving the problem.

Learning from RE: the PropCo-OpCo (and is it the solution?)

One such approach could be the PropCo-OpCo structure, as also suggested by Evan Lyseng.

The PropCo-OpCo model, widely popularized in real estate, separates physical assets from operations, creating two distinct entities with different capital structures, risk profiles, and investor bases. This bifurcation allows each entity to attract the most appropriate funding sources for its specific characteristics and might be the answer we were looking for.

Imagine a structure where the "PropCo" equivalent holds the acquired service businesses (funded primarily through debt and structured with traditional PE economics), while the "OpCo" equivalent develops and owns the AI technology, funded by venture capital with traditional venture economics (equity, etc.). In this model, the revenue of the OpCo is directly tied to the PropCo's performance. Since the OpCo's technology is (theoretically) capable of unlocking higher revenue and better margins, it's fair to structure a deal by which the OpCo monetizes based on a percentage of the top line and/or the margin it unlocks.

From a founder's perspective, this structure is more complex, and it results in effectively running two separate companies, but it potentially solves several challenges. For example, the PropCo-OpCo model also addresses several operational tensions inherent in the traditional single-entity approach. As previously mentioned, service-focused businesses naturally prioritize consistent cash flow to service debt, while technology ventures require ongoing R&D investment to maintain competitiveness. When combined in one entity, these competing priorities often lead to underinvestment in technology over time. Separation allows each component to operate according to its optimal strategy without compromising the other.

Ultimately, this separation would also allow the technology component to maintain its identity as a neutral platform rather than becoming captive to specific service operations. The OpCo licenses its technology to the PropCo while remaining free to pursue independent growth opportunities across the broader industry, allowing it to maintain the potential for software-like multiples for the technology component.

This is where it stops making sense for me (help me out if I am missing something).

If you're building a platform that can monetize through a percentage of customer revenue to begin with (as in the PropCo-OpCo model), why create two separate companies and introduce unnecessary complexity? I can't find a compelling argument for this approach. At the same time, structuring an exclusivity deal between the OpCo and PropCo creates its own problems: the OpCo would face severe limitations with its addressable market (effectively tied to a single customer's revenue), resulting in restricted growth potential. Additionally, the extreme customer concentration (100% dependent on one independent client) would make the OpCo virtually unsellable to anyone except the PropCo itself. The supposed benefits of separation quickly dissolve.

Hence, I am not convinced this is the answer: there might be another way to look at the problem.

Artificial roll-ups (or is it just vSaaS?)

Perhaps an "artificial" AI roll-up strategy (using the term very freely here) in AEC might be possible, but it needs to be extremely intentional with the product being built and the ICP.

Specifically, I think it can work by targeting small and/or one-person A&E shops, and maximize their billable hours (e.g. by reducing administrative burdens), developing an AI system that essentially functions as a "junior employee" handling routine tasks - ideally not just front and back-office, but some core workflows in the long run too, developing a full-stack product - while the human professional focuses on complex problem-solving and client relationships. Therefore, you still operate as an industry-neutral software provider, but rather than charging traditional SaaS fees, if you can prove that your solution unlocks significantly more billable hours/higher revenue, you can charge on a "value-based" model (e.g. a percentage of the invoices). Even more so if you can add a revenue generation/lead generation/marketplace component on top of the software. Therefore, this platform creates value by unlocking and augmenting professional capacity across thousands of small practices, rather than attempting to consolidate them, while still benefiting from the upside of the value created.

Did I just get full circle on vSaaS here? Maybe. Ultimately, it's vSaaS I am talking about - but with very specific nuances.

First of all, the value-based pricing model fundamentally changes the business relationship. Traditional vertical SaaS typically charges fixed subscription fees regardless of outcomes, while this model directly ties compensation to the value created (e.g. percentage of invoices), creating a true partnership where growth is directly linked to your customers' success. This means that there's a significant difference in how customers perceive the offering: you're not a tool or a software/IT expense, but rather a revenue-generating asset. This also has implications for your scaling mechanics: while vSaaS typically needs to expand features to grow ARPU or move upmarket to larger customers, this model naturally scales revenue as you help small practices grow their businesses, without necessarily needing to chase enterprise clients. The depth of workflow integration is also different. Standard vertical SaaS often focuses on digitizing specific processes, but this approach requires much deeper domain expertise and more sophisticated AI capabilities than simply digitizing paper processes.

Ultimately, what makes this particularly interesting is that it creates a pathway to capture significantly more value from fragmented professional services markets without the operational complexity and capital requirements of actually acquiring and integrating firms. You're essentially creating many of the economic benefits of consolidation through software rather than ownership.

Conclusion

While AI rollups present an innovative approach to industry transformation, I remain skeptical of their viability as venture-backed (emphasis here) opportunities. Apart from a potential fundamental mismatch between VC economics and rollup capital requirements, let's not forget that executing successful rollups remains challenging even without technology considerations. Harvard Business Review research found that approximately two-thirds of roll-up strategies historically have been value-neutral or outright destructive.

This means a founder pursuing this strategy faces a compound challenge: they must simultaneously master the complex art of identifying, acquiring, and integrating suitable companies while developing sophisticated software that meaningfully transforms operations.

This is not rookie territory, and perhaps one of the strongest arguments against pursuing this model. Therefore, while AI rollups may create substantial value, I think traditional PE structures with selective technology enhancement likely represent a more appropriate capital source than venture funding for most consolidation opportunities (particularly in service industries).

Perhaps, in the AEC sector, deeply integrated technology platforms that empower small practices through value-based pricing models may create "virtual consolidation" without the complexities of ownership, potentially offering a more capital-efficient path to transformation than traditional roll-up strategies, while maintaining true software economics (and multiples).

That's it for part one of the "New business models, and can they work for AEC-tech?" series. Stay tuned and subscribe to my newsletter so you don't miss the next one in the series.

More appetite for this type of content?

Subscribe to my newsletter for more "real-world" startups and industry insights - I post monthly!

Head to Foundamental's website, and check our "Perspectives" for more videos, podcasts, and articles on anything real-world and AEC.

#AIRollups #ConTech #AECtech #VentureCapital #ConsolidationStrategy #AITransformation #BusinessModel #ValueBasedPricing