Carried No Interest on AI and PE software roll-ups: 'Transactional SaaS is dead'

Summary

Transactional SaaS is dead. That's Joe Lamont's thesis from Trilogy, and it cuts to the heart of why private equity software roll-ups face an existential reckoning as AI reshapes the economics of software itself.

For the past eight years, Vista Equity Partners and TOMA Bravo proved that buying a software company at seven times ARR, growing it six times over, and selling at ten times ARR generated the strongest returns in private equity across any industry. The model was built on public multiples and a bet on low interest rates. When rates rose in 2022, public multiples on enterprise software compressed hard. Carried No Interest argues that compression was almost entirely interest rates rather than a technological shift.

AI is different. It's a structural threat to the factory-as-moat logic that built the roll-up model. If it takes 90% less time to build a competitive software solution, the perceived moat dissolves. Transactional SaaS, where you buy an old code base and flip it, will be dramatically worse. AI-native software will be worth more than any SaaS company has ever been because the customer can do radically more with it.

The gap between theory and practice is wide. Mark Leonard, the founder of Constellation Software, signaled deep uncertainty on a recent investor call. His capital allocators aren't seeing a massive AI threat yet. Joe Schmo spinning up a vertical market software company and crushing incumbents hasn't happened. Multiples on software PE acquisitions haven't compressed because of perceived AI risk. Leonard essentially said: I don't know what happens next. Constellation's stock dropped 8% after the call.

Value accrues differently under AI depending on the buyer's model. Traditional PE buyers like Vista and TOMA Bravo make their returns on exit multiples, selling for a higher ARR multiple than entry. That multiple is a function of macro sentiment, not how AI-first the organization is. Constellation and Trilogy own for cash flow. They care about free cash flow generation, not the exit multiple. For them, AI is an accelerant: higher net retention from AI-first product features, lower operating expenses from AI-driven efficiency. If they can prove they're making more cash from existing businesses, the value compounds regardless of where public multiples land.

The replatforming question remains unsettled. No one is yet ripping out Salesforce wholesale in favor of custom AI solutions, despite CEO commentary suggesting they might. Convenience and switching costs still matter. A gym valet manager might be able to build valet software with one part-time person now, but the hassle still justifies paying a vendor a few thousand dollars a month. The natural pressure, though, is on pricing. If a vibe-coding service provider can build and support custom vertical software for $20,000 upfront and $100 per month with an AI-first, low-cost team, the price of incumbent vertical market software falls. That doesn't kill those businesses immediately, but it resets what they can charge.

Software PE owners and operators need to find new ways to drive value. They need to build AI-first features that weren't possible before cheap LLM APIs, features that increase net retention, and lower OpEx through AI-driven operations. Some are succeeding: one founder running an AI roll-up play noted that he can buy legacy businesses, tell existing teams they won't face layoffs, make them dramatically more efficient with AI tools, and scale revenue at the same time. But it requires execution against an uncertain future and against the structural headwind that the cost to build software is falling fast.

Lamont is right that the value of AI will accrue much more to the free cash flow buyer than the ARR buyer. Constellation is better positioned than Vista if it executes on AI-driven product and operational improvements. The question is whether Leonard and his capital allocators can move fast enough to prove it.