# SaaS Multiples Fell From 16x to 4.7x in Four Years — 8 Shifts Moving AI's Money

SaaS multiples fell from 16x to 4.7x in four years. I've been tracking why — here are the 8 shifts moving AI's money, and where the moat actually sits now.

Author: Maxim Tolmachev
Published: 2026-06-03
Language: en
Canonical: https://maximtolmachev.com/blog/saas-multiples-fell-from-16x-to-47x-8-shifts-moving-ai-money/
Original source: https://www.linkedin.com/in/mtolmachev/recent-activity/all/

SaaS multiples fell from 16x to 4.7x in four years. I've been tracking why — here are the 8 shifts moving AI's money, and where the moat actually sits now.

**1. Money and market valuations**
The SaaS multiple dropped to 4.7x revenue (down from ~16x in 2021). The market overpaid for "software for software's sake."

**2. Who captures the value + lab monetization**
The earners are the resource makers: electricity, datacenters, GPUs and the whole chain. Big labs are mostly in the red and hunting for monetization models (e.g. contextual ads). AI products are still tiny in money terms.

**3. The bottleneck and new distribution**
Cost and time of development stopped being the constraint — a result of vibecoding / AI-assisted coding. The bottleneck shifted to distribution. Channels are changing — GEO: traffic from ChatGPT converts at ~15.9% vs ~1.8% from Google SEO. Discovery is moving into generative engines. Any software now needs API + CLI + MCP so AI agents can use it — otherwise you drop out of the market.

**4. Speed and AI-native ops**
AI-native moves multiples faster: a complaint in support → fix shipped to production in 6 hours. AI tooling gets simpler every quarter.

**5. New business model and the role of SaaS**
SaaS is shifting toward guarantee/support (uptime, fast incident triage — the customer is never left unsupported). From subscription to pay-per-token/action — vendors might get richer, charging even for what used to sit in the base plan. AI reprices whole categories overnight: when Anthropic shipped a legal tool for Claude (Feb 2026), Thomson Reuters, RELX and Wolters Kluwer lost 13–16% in a day.

**6. What AI gives the product**
Hyperpersonalization.

**7. Org, culture, the employee's role**
AI adoption inside teams is still very low. Culture is changing under wild speed; the employee's role shifts to reviewer, to avoid becoming the bottleneck. Value moves into areas that are hard to verify cleanly — be the expert reviewer there once automation is everywhere.

**8. Where to build a moat**
Narrow, expensive niches with a big market the labs won't enter (e.g. US credit cards for dentists, with high margins on medical goods). Verticals with expensive labor (US: lawyers, dentists) — there the cost of AI pays off.

What trends are you seeing? Which do you agree or disagree with?
