Why most AI-enabled roll-ups will fail (and the 4 rules of those that will win)

Follow the discussion on LinkedIn and X.com

As we’ve highlighted in some of our recent posts on AI value creation and “Hidden AI”, there are increasingly diverse set of ways to win given the technology advancements made available to founders today. One of the models that has come into heavy focus (and scrutiny) is AI-powered roll-ups. There are a lot of reasons to be excited about this model, after all, it’s one of the most direct pathways to tie the outcomes of AI to your own bottom-line, while implementing AI as swiftly and fully as you can. That said, it may also be the most controversial of AI value creation strategies with some claiming that it’s simply PE dressed up in VC’s clothing. With that, we wanted to share additional perspective given our experience and research on the space.

As many know, we are early investors in Equal Parts, an AI-powered aggregator of insurance agencies. While many have recently sprinted full force into AI roll-ups, our thesis behind what would ultimately become Equal Parts was born in 2021, far before AI was part of the calculus. AI has supercharged this thesis making a thesis that we already loved even more powerful, however, the truth is that very few properly understand how most of these roll-up strategies actually create value.

In PE, you aren’t just counting on revenue multiples, you need to drive real EBITDA and/or earnings growth to create value. As we see many of the pitches that come in under this umbrella, many are simply heralding the top line growth of these companies with incremental margin expansion from AI implementations, without determining whether they are actually creating any LONG-TERM improvement in the earnings potential of the company. It is our opinion that those that don’t have very little chance of generating venture returns and, perhaps more realistically, will struggle to deliver PE level returns. The reality is that PE people are pretty fricking smart and a bunch of newcomers approaching that market with hubris and big budgets seem unlikely to deliver the execution or financial prowess to succeed (more often than not, we actually are seeing a lot of these companies burn through capital and then sell for less than their paid in capital raised as companies like Thrasio ultimately did.

But there will indeed be winners and we believe Equal Parts has a hell of a shot to be one. There are some industry specific conditions that make insurance agency roll-ups particularly attractive (and its also helpful to back a proven beast like Mike 😊), but there are 4 essential characteristics that we believe enable success for AI-driven roll-ups.

The 4 Rules of AI-Powered Roll-up Success

1. White collar services

This may seem obvious, but the principal advantage of AI is not in making assets cheaper, but in reducing the labor cost associated with labor-intensive businesses. With that, identifying industries where AI-affected labor represents the most significant component of the industry’s cost structure is critical to determining whether the model has enough juice to squeeze. For those who read our “Hidden AI” piece, you can see that identifying the cost structure of a given business and determining where AI may have the greatest leverage on the overall cost profile is critical to success.

The businesses with the highest labor cost components don’t just have lots of bodies, they have expensive ones. Using AI to replicate the behavior of off-shored individuals earning a few dollars an hour doesn’t generate significant enough leverage when you look at the competing computing/token costs. What is far more interesting is identifying expensive “white collar” services that can be trained to deliver superior results at a superior price point via AI.

2. Recurring Revenue

Expensive labor costs are just the first step in this process, with recurring revenue seeming to be requisite for long-term success. We are seeing plenty of de novo operators automating white collar services such as graphic design, legal services and tax prep and those businesses could very well be highly successful (albeit, it’s too early to tell and it’s hard to discern whether customers will constantly bid out services to the lowest cost provider, ultimately creating a race to zero marginal profit as competition enters into what are now early greenfield categories). But if you are buying a company, you need to ensure that company has long-term terminal value, otherwise the CAC:LTV dynamics of this growth model don’t work. Services that need to constantly acquire customers to maintain revenue limit predictability of a given company (as we saw with many of the Amazon merchant aggregators) and the #1 rule of a roll-up strategy is to ensure that you don’t take losses on cash that you put out the door. There is undoubtedly a certain degree of financial engineering necessary to any roll-up model and having visibility of future revenue streams from the companies purchased is critical not only ensuring that any acquisition target is successful, but to properly finance this as a strategy for growth.

3. Revenue side synergies

The vast majority of AI-powered roll-up stories are heavily focused on cost-side synergies. I don’t blame them. AI is really good at identifying and exploiting lower cost strategies. The problem is that those cost advantages are largely ephemeral. Any business that wants to have LONG-TERM success in this category needs to assume that competition will implement similar AI interventions (at some point), ultimately competing any initial advantages down to the industry’s new cost structure. While the cost advantages delivered via AI may produce some initial dopamine given the initial margin lift produced, competition encroaching on your market as they gradually adopt AI will ultimately lead to a declining margin structure – something that is often lethal to a business.

Perhaps even scarier, if you are investing in a business that has a declining cost structure, there is ample risk that prices for your services may decline significantly over time, leading to significant ACV declines from your existing customers. We’re already seeing this play through in the TPA market as well as other white-collar services as emerging AI-powered entrants compete prices asymptotically down to the point of zero margin. For a startup trying to show growth to investors, it’s incredibly difficult to combat declining ACVs and we’ve seen the cost curves in some of these spaces collapse pricing per activity by >50% in a given year (meaning that a given company needs to double its customer base just to stay flat). This is one of the reasons why we see insurance agencies as particularly suited to benefit from the long-term effects of AI – insurance premiums naturally rise with inflation (creating pre-baked revenue growth every year), but commissions are ultimately set at industry standards (generally 10-25% based on volume and type of coverage). Customer ACVs don’t fall as AI enters the space, just costs making these businesses more profitable over time (NOTE: this could ultimately lead to asset prices going up making acquisitions increasingly difficult, but such conditions would also dramatically improve the implied market value of our company as well).

Revenue-side synergies, however, are far more sustainable. When aggregation via AI can enable companies to produce incremental revenue lift (while simultaneously benefitting from AI-powered cost advantages), then you have a compelling and durable profit margin. Ideally, these revenue side synergies demonstrate increasing returns to scale that enable incremental acquisitions to demonstrate incrementally higher yield. Notably, we see this in insurance, where larger premium volumes enable agencies to not only achieve higher commission rates, but also gain access to additional products to cross-sell, further lifting the both the revenue and margin profile of not just the target, but the rest of the underlying agencies in the portfolio. We wrote about this dynamic in our Prepared Mind Deep Dive on the space and it’s one of the core considerations for why insurance agency roll-ups are so effective.

4. Roll-up as a wedge into something bigger

While the 3 rules above provide a blue-print for success, the most exciting companies in this category will leverage the initial advantages of this model to achieve something beyond the roll-up model itself. As you can see, so much of my ideology around these types of opportunities (as well as in AI and broader investing) is rooted in competitive dynamics. Ultimately, building an at-scale player can yield good returns (especially for early investors), but leveraging that position of strength to slingshot into an even bigger prize is where the real win is. I won’t unveil what that is for Equal Parts, but if successful, the company has the potential to monopolize the entire retail insurance market, a feat that would make it the largest insurance agency on the planet (NOTE: the largest insurance agency in the US is currently valued at ~$150b). Developing the next Acrisure (a $25-30b insurance agency aggregator started in the last 15 years) isn’t too shabby, but make no mistake, Mike and the Equal Parts team have their eyes set on an objective capable of producing 10x that.

As always, every company is its own snowflake and some of these rules may or may not be relevant to every given company, but I’m hopeful this initial list of criteria provides some clarity to how we think through these type of opportunities at Equal. Building great companies is NOT meant to be easy and as the great Charlie Munger once said “Anything that is too easy, should raise your suspicions.” Building a great company with this model requires tremendous nuance and ruthless execution, making this model far less applicable and far more difficult than most think. That said, for those who can figure out how to apply these rules to develop a category monopoly, we see a tremendous opportunity for value creation.

As always, our phone lines are open for those who are ready to climb that mountain.