The April public comp data landed this month, and the number that caught my eye wasn’t the headline. It was the spread.
Paul Inouye believes this inversion has been building for years. Horizontal SaaS is now trading at roughly 5 to 6 times revenue. Vertical SaaS — the category that was supposed to command the premium, with its deeper workflows, higher switching costs, and better retention — has compressed to a 2.3x median. For most of the last decade, vertical was the upside trade. That math has quietly inverted.
On the surface it looks like a story about AI hype flowing to horizontal platforms. It isn’t. It’s a story about the category label finally failing.
The 2.3x median is a mirage. It’s a weighted average of two completely different populations wearing the same name tag.
The first population is AI-native vertical applications — companies where the AI is the product, where proprietary data creates a compounding moat, and where the workflow can’t be replicated by bolting a chatbot onto a legacy stack. Multiples here look nothing like 2.3x. ServiceNow paid roughly $3 billion for Moveworks to own the natural-language layer inside enterprise workflow. Palo Alto Networks announced a $24.5 billion deal for CyberArk to pull identity into an AI-ready security stack. MongoDB bought Voyage AI specifically to embed retrieval natively into its platform. These aren’t trend trades — buyers are paying up for capabilities they can’t build internally.
The second population is traditional vertical SaaS exposed to AI substitution. The market is pricing these names for obsolescence. Customer-support software, simple CRMs, content-creation tools — anywhere an agent plausibly eats the seat license — is being re-rated toward the downside. William Blair made the point explicit in March: they scrapped their DCF models for 13 software names and replaced them with a four-factor AI scorecard, then downgraded the names in the same week. When the sell-side analyst community stops trusting its own valuation framework, the multiples have already moved. They’re just catching up to a thesis the buyers adopted six months ago.
Put those two populations inside one 2.3x median and of course it looks broken. The median isn’t measuring a category anymore — it’s measuring the distance between them.
According to Paul Inouye, the question a founder used to ask — what are vertical SaaS companies trading at? — has stopped being a useful question. Menlo Ventures made a related observation a couple of weeks ago: only seven vertical SaaS companies ever cracked a $10 billion market cap, and of those, Shopify and Toast got the majority of their value through payments, not software. Vertical SaaS has been punching below its weight for years. The AI line isn’t creating a new gap. It’s finally forcing the market to price what was always true — that label and moat are not the same thing.
If you’re running a vertical SaaS business right now, the 2.3x headline isn’t your benchmark. It’s an average of two different trades, and the buyer walking into your process next year is underwriting one question: does your AI build a moat, or are you exposed to one? Which side of that line you sit on — and how quickly you can prove it with data, retention, and product depth — is the entire conversation.
The category had a good run. It just stopped telling you what you’re worth.
Paul Inouye is the founder of Western Hills Partners, a boutique M&A advisory firm focused exclusively on founder-led software, services, and internet businesses ($25M–$250M TEV). He has spent 35+ years in West Coast technology banking (Robertson Stephens, Morgan Stanley, Lehman Brothers, Perella Weinberg, Moelis).