by Vittorio Sambuy
For over a decade, SaaS was the undisputed king of venture capital. Predictable revenues, scalable distribution, and clean metrics made subscription software the safest growth story in tech. Then came AI — and with it, a structural break that many founders and investors are still underestimating.
AI is not just another feature upgrade. It is changing how companies are built, funded, priced, and defended. And in the process, it is quietly killing large parts of the traditional SaaS playbook.
From SaaS Multiples to Intelligence Premiums
Historically, VC underwriting in SaaS was driven by clear heuristics: ARR growth, retention, CAC/LTV ratios, and operating leverage. AI disrupts this logic.
AI-native companies don’t scale like SaaS. They scale like infrastructure + intelligence hybrids:
- Early costs are higher (compute, data, talent).
- Marginal costs don’t always trend to zero.
- Value is concentrated in proprietary models, workflows, and data — not dashboards.
As a result, we’re seeing a shift in VC behavior:
- Less obsession with short-term ARR.
- More focus on data moats, model performance, and distribution leverage.
- A renewed appetite for technically deep teams, even at the expense of near-term revenues.
Capital is flowing not to “better SaaS,” but to companies that own intelligence layers.
The SaaS Trap: When Features Become Commodities
AI’s most immediate impact is brutal: it turns SaaS features into commodities.
What once justified a standalone company — scheduling, CRM enrichment, analytics, copywriting, forecasting — can now be:
- Generated on the fly by foundation models
- Embedded into existing platforms
- Rebuilt by competitors in weeks, not years
This creates a dangerous dynamic for traditional SaaS companies:
- Differentiation erodes faster than sales cycles.
- Switching costs collapse.
- Pricing power disappears.
In many cases, AI doesn’t disrupt SaaS by replacing the product — it absorbs it.
Why Many SaaS Companies Will Die (Quietly)
Most SaaS companies won’t fail spectacularly. They’ll stagnate.
The warning signs are already visible:
- Flat growth masked by upselling.
- Increasing sales effort for the same ARR.
- Customers asking “why isn’t this just built into my stack?”
AI-native challengers don’t need to beat incumbents feature by feature. They simply need to:
- Automate the core job-to-be-done
- Reduce human input
- Compress the value chain
When that happens, the SaaS layer becomes optional — and optional software doesn’t survive long.
The New VC Question: “Where Does Intelligence Live?”
For VCs, the core question has changed.
It’s no longer: “Is this a great SaaS company?”
It’s now: Where does intelligence live in this system – and who controls it?
Winning companies tend to sit in one of three positions:
- Infrastructure enablers (compute, tooling, orchestration)
- Vertical AI operators (deeply embedded in specific workflows)
- Data owners with compounding learning effects
Horizontal SaaS without proprietary data or distribution is increasingly fragile.
What This Means for Founders
For founders, the implication is uncomfortable but clear:
- SaaS is no longer defensible by default.
- Speed matters more than polish.
- Owning workflows beats owning features.
The strongest teams are rethinking:
- Pricing models (usage, outcome-based, hybrid)
- Go-to-market (AI-first distribution, not sales-heavy)
- Product scope (end-to-end automation, not tools)
In an AI world, software that assists is losing to software that decides.
At Ventures: What We’re Watching
At Ventures.eu, we see this shift daily across EIC-backed startups, VCs, and corporates:
- More capital flowing earlier into AI-native teams
- Corporates reassessing buy vs build decisions
- Investors demanding technical depth alongside commercial traction
AI is compressing timelines, blurring categories, and forcing harder choices — for founders and investors alike.
The takeaway is simple:
AI is not a SaaS upgrade. It is a redefinition of what a software company is.
Those who adapt will build the next generation of category leaders.
Those who don’t will discover that recurring revenue is not the same as recurring relevance.
