Every PE portfolio review now includes the same anxious question: "What happens to our exit multiples when AI commoditizes everything?" It's the wrong question. Here's the right one.
I've been in a lot of board rooms lately where the conversation goes something like this:
"We built this SaaS company over 15 years. It has real customers, real revenue, real workflows embedded in how industries operate. And now ChatGPT does 80% of what we do for $20 a month. Are we the next Kodak?"
The fear is real. The conclusion is wrong.
The Kodak comparison doesn't hold
Kodak's actual failure wasn't that digital photography emerged. They invented digital photography. Their failure was threefold:
- Denial: "Film is still better quality"
- Cannibalization fear: "We can't kill our own cash cow"
- Identity lock-in: "We're a film company"
Here's what's different about your 15-year SaaS company: Kodak's product was consumed. Film is gone when it's developed. A SaaS tool is embedded in how people work. That's actually defensible, if you lean into it.
The question isn't "how do we protect our portfolio from AI." The question is "how do we use AI to make our portfolio 2-5x more valuable in 3 years?"
AI is the biggest value creation lever PE has seen since cloud computing. And most firms are playing defense when they should be playing offense.
What PE actually cares about, and how AI affects it
Let's talk about what matters: EBITDA expansion, revenue acceleration, and exit multiple expansion. AI affects all three.
| Value Lever | AI Opportunity | PE Math |
|---|---|---|
| EBITDA Expansion | Automate CS, AI support, internal efficiency | 15% COGS reduction at 10x = meaningful bump |
| Revenue Acceleration | AI features as upsell, expand TAM, faster sales | 20% ACV lift + 30% faster close = compounding |
| Multiple Expansion | "AI-native" positioning, strategic buyer appeal | "AI-enhanced vertical SaaS" trades higher |
This isn't theoretical. I've seen portfolio companies add 25% to their ACV by packaging AI features into a premium tier. I've seen support costs drop 40% with AI-powered triage. I've seen sales cycles shorten because AI-generated proposals close faster.
The companies capturing this value aren't the ones with the best AI technology. They're the ones who understand what they actually have.
What you have that OpenAI doesn't
Here's what a 15-year vertical SaaS company owns that ChatGPT literally cannot access:
- Domain-specific data Years of customer workflows, edge cases, industry patterns. This is the training data that makes AI actually useful in a specific domain.
- Embedded relationships You're integrated into customer systems. Their data lives in your product. Switching costs are real.
- Tacit knowledge What actually works in the industry, not what works in demos. Fifteen years of learning what customers really need, encoded in product decisions.
- Distribution An existing customer base you can deploy AI features to immediately. OpenAI has to convince people to use a new tool. You can add AI to something they already use every day.
- Enterprise requirements Compliance, audit trails, permissions, SOC2, GDPR. ChatGPT is a black box. Enterprises need accountability.
A vertical SaaS with 1,000 customers and 15 years of workflow data can build AI features that ChatGPT literally cannot replicate. ChatGPT has generic knowledge. Your portfolio companies have specific knowledge about how industries actually operate.
"Your institutional knowledge is the most valuable AI training data in your industry. The question is whether you're using it, or letting generic AI eat your lunch."
What ChatGPT can and can't replace
Let's be honest about this. Some things are genuinely vulnerable:
What ChatGPT CAN replace
- Generic content generation
- Simple analysis on pasted data
- First-draft anything
- Q&A on public knowledge
- Features that exist in isolation, unconnected to customer data
If 80% of your product is "things ChatGPT can do," you have a problem. But most mature SaaS products have something else: data lock-in, workflow integration, and accumulated edge-case handling that took years to build.
That's the moat. Not the code. The accumulated intelligence about what works.
The counter-Kodak playbook
| Kodak Did | You Should Do |
|---|---|
| Denied the threat | Name it publicly: "AI changes everything, here's how we're adapting" |
| Protected the cash cow | Cannibalize yourself: launch AI features that eat your old margins |
| Stayed "a film company" | Redefine identity: "We're the [domain] experts. Software is just delivery" |
| Let others innovate on their tech | Put your institutional knowledge into AI before competitors do |
| Moved slowly | Ship monthly. Velocity is the new moat. |
The reframe: from tool company to knowledge company
The companies that survive this transition aren't the ones with the best features. They're the ones who understand that their real asset isn't software. It's knowledge.
Here's the positioning shift:
Old positioning: "We make software that does X." This is vulnerable. AI tools can do X.
New positioning: "We know how [industry] should work. Our software encodes that knowledge. Now we're adding AI to make that knowledge 10x more powerful." This is defensible. AI becomes your weapon, not your competitor.
The exit story changes too. You're not selling a SaaS. You're selling the AI that knows [industry]. That commands a premium.
The portfolio-level opportunity
For PE firms, this isn't a company-by-company problem. It's a portfolio-level value creation opportunity.
Imagine deploying a systematic AI value creation playbook across 10 portfolio companies:
- Audit each company for AI opportunity and AI risk
- Prioritize based on data moats and workflow lock-in
- Run 6-8 week AI acceleration sprints to ship features that use each company's unique data
- Deploy shared AI ops capabilities across the portfolio
- Craft AI-enabled exit narratives for each company
The firms that do this systematically will outperform the ones who let each portfolio company figure it out independently.
What to do this quarter
If you're a PE operator reading this, here are three things to do in the next 90 days:
- Map your portfolio by AI exposure Which companies have genuine data/workflow moats? Which are feature-dependent? This determines where to invest and where to worry.
- Identify the 20% ChatGPT can't do For each company, what's the defensible core? What would require your data, your integrations, your institutional knowledge to replicate? That's your new center of gravity.
- Ship one AI feature that uses your moat Not generic AI bolted on. An AI feature that competitors literally cannot build because they don't have what you have. Prove the thesis in 8 weeks.
The message to your portfolio companies
Stop asking "how do we protect ourselves from AI?" Start asking "how do we use AI to become irreplaceable?"
Your institutional knowledge, your domain data, your accumulated understanding of how your industry actually works, that's the most valuable AI training set in your space. Are you using it, or letting generic AI eat your lunch?
The companies that move fastest will capture the value. The ones that wait will become case studies in why incumbents lose to technology shifts.
Kodak didn't have to die. They invented digital photography. They just couldn't bring themselves to use it.
Don't make the same mistake.