Why PE Firms Are Turning Back to Investing in SaaS Firms With High Gross Margins Instead of AI Firms

One of our members has been talking on and off with a well-known private equity group for close to a year, and in one of their more recent conversations, someone on their side said something out loud that a lot of investors have clearly been thinking privately for months. They told him plainly that they've had a wake-up call about the AI businesses they've put money into. The businesses looked exciting on the way in. What they've discovered since is that a lot of these companies carry a cost of goods sold that looks nothing like software, and everything like a traditional, capital-intensive business.

• It's the same investors, a very different tone. The group behind this shift is the same one that spent the last two years chasing AI deals.
• The trigger was the cost structure, not the technology. They still believe in AI as a capability, just not as the product itself.
• This is capital moving, not just opinion. A firm deciding where to actually deploy money next is a stronger signal than an analyst note.

Why PE Is Rotating Back Toward High-Margin SaaS That Incorporates AI, Rather Than AI Products With Thin Margins, and What It's Doing to SaaS Multiples

The example they used to explain their own thinking was almost funny in how far afield it reached. Part of this same PE group's portfolio includes investments in concrete plants across Asia, businesses that run on something like a thirty-five percent markup, a completely normal margin for that kind of heavy industrial business. A classic SaaS investment, by contrast, runs gross margins in the eighty-five to ninety-five percent range, the number that's made software such an attractive asset class for two decades. What they told our member is that a lot of their AI investments are landing much closer to the concrete plant than to the SaaS company, once you account for what it actually costs to run the product.

• Concrete plants: roughly 35% markup. A completely normal, healthy margin for a capital-intensive industrial business.
• Classic SaaS: 85-95% gross margin. The number that's driven two decades of premium valuations for software companies.
• Many AI-native products: landing much closer to the plant. Once the real infrastructure and talent costs hit the P&L, the margin profile shifts dramatically.

That cost breakdown is worth spelling out, because it's the whole thesis in miniature. The cost of skilled AI talent has stayed extremely high even as the tools have matured. The cost of the electricity running the compute is real and growing. The cost of the GPU hardware itself, the NVIDIA cards everyone needs and few can get cheaply, is a genuine capital expense rather than a rounding error. And there's the cost of failure baked into the infrastructure, servers going down, retries, redundancy you have to build because the workloads are so much heavier than a normal SaaS application.

• Talent costs stay high. Skilled AI engineering and research talent commands a premium that hasn't come down even as tools have improved.
• Compute is a real line item. Electricity and GPU hardware costs scale directly with usage in a way traditional SaaS infrastructure never did.
• Redundancy adds overhead. Heavier workloads mean more failure handling and more infrastructure spend baked into every unit of output.
• The margin profile shifts. What looked like a software business on the pitch deck starts to resemble a traditional, capital-intensive operating business once real costs land on the P&L.

The line this particular investor used, in essence, was that their firm is now turning its capital strategy back toward SaaS companies that have figured out how to incorporate AI usefully into an existing high-margin product, rather than chasing companies where AI itself is the product.

• A meaningfully different thesis than two years ago. This isn't a tweak, it's a reversal of the logic that drove capital into AI-native startups.
• It isn't just this one group. Our member noted the framing shows up elsewhere too, it just landed differently coming from a firm actively deciding where to deploy money.

It's also worth noticing what this tells you about how sophisticated capital actually behaves under uncertainty. A year and a half ago, a lot of the same investors were telling founders, directly or indirectly, that traditional SaaS had a short shelf life left, that a small team with the right AI tools could rebuild a whole category of enterprise software in weeks. Some of that fear was a useful wake-up call. It pushed a lot of founders, including several in our own mastermind, to genuinely integrate AI into their workflows instead of assuming the status quo would hold forever. But the same fear also drove a wave of capital into AI-native companies without much scrutiny of what those companies would actually cost to operate at scale.

• The first pass was fear-driven. Capital moved fast into AI-native companies on the strength of the story, before the cost structure was fully understood.
• The second pass is cost-driven. The same capital is now working through the real unit economics and landing on a more traditional answer.
• Durable margin wins both times. Businesses with structurally high gross margins remain the better long-term bet, AI feature or not.

The Chatbot Hype Cycle Is Ending

This connects to a broader shift that's been building for a while. Not long ago, the prevailing fear in enterprise software circles was that AI would make traditional SaaS irrelevant almost overnight, that a couple of sharp engineers with the right tools could replace an entire vertical software category in weeks. Investors have clearly moved past that fear. What they've realized instead is that endless chatbot wrappers built on top of a foundation model aren't especially valuable or defensible, while companies with real EBITDA, real retention, and the scale to sell six-figure enterprise contracts remain exactly as valuable as they always were.

• Chatbot wrappers: thin and easy to copy. A clever prompt over someone else's foundation model isn't a moat, and investors have noticed.
• Durable SaaS: still commands a premium. Real EBITDA, real retention, and the ability to close six-figure enterprise contracts haven't gone out of style.
• The market has separated the two. What used to get lumped together as "AI risk to software" is now being priced very differently company by company.

The practical evidence of this rotation is already visible in the data our member has been tracking through a banking relationship, even though his company is smaller than what that particular bank typically covers.

• Strategic acquirer offers: up roughly 20% in a month. A sharp recovery in valuation offers over the prior month alone.
• Financial buyers are lagging but starting to catch up. Several PE firms are beginning to talk the way strategics are already behaving.
• Talk usually precedes checks by a couple of quarters. Not a couple of years, based on how this pattern has played out before.

What This Means If You Run a SaaS Company Today

If you're running a SaaS business with healthy gross margins and you've been nervous that AI makes your category less interesting to acquirers or investors, this shift is genuinely good news, provided you've actually put AI to use inside your product rather than ignoring it. The distinction PE is drawing isn't AI versus no AI. It's the difference between AI as a cost center that erodes your margin structure and AI as a capability that makes an already high-margin product more valuable without dragging your gross margin down toward a services business.

• AI as a cost center. The AI is the product, and its infrastructure and talent costs are eating the margin that made software attractive in the first place.
• AI as a capability. The AI sits inside an already high-margin product, making it better without changing the underlying cost structure.

The companies best positioned in this new framing are the ones using AI to make existing high-margin software do more, faster support resolution, better forecasting, smarter workflows inside a product customers already pay a healthy premium for, rather than companies whose entire value proposition is a thin layer over a foundation model with someone else's compute costs baked in.

• AI making existing software do more. Faster support resolution, better forecasting, smarter workflows inside a product customers already pay a premium for.
• AI as a thin wrapper on someone else's model. The value proposition, and the compute cost, both belong to the underlying foundation model, not to you.
• If you're already at 85-90% margin with AI layered on top. Without meaningfully changing that cost structure, you're sitting in exactly the category this capital rotation is chasing.

None of this means the fear of the last year and a half was pointless. It forced a lot of founders to genuinely rethink their product and their workflows rather than coasting on the assumption that nothing would change.

• The fear pushed real change. Founders who used it to rethink their product and workflows are better off for it.
• The winners look structurally familiar. Real customers, real retention, real margin, now with AI woven into the product rather than standing in for it.
• A PE group saying this out loud is a signal worth heeding. It points to where the next round of multiples is heading.

The practical takeaway is to be deliberate about how you tell your own story to acquirers and investors over the next several quarters. Lead with the gross margin, the retention, and the EBITDA, the numbers that make you look like the durable software business you are, and talk about AI as the thing that's making those numbers better rather than as the headline of the pitch.

• Lead with the durable numbers. Gross margin, retention, and EBITDA first, since those are what make you look like the business PE is now chasing.
• Frame AI as an enhancer, not the headline. Position it as the thing improving your existing economics, not as a separate story competing for attention.
• Expect this to matter within quarters, not years. Financial buyers are already starting to talk like the strategics that moved first.

Founders who spent the last year quietly building AI into their product without changing their fundamental cost structure are about to look very smart to exactly the kind of capital that's rotating back their way. If that's you, this is a good moment to make sure the story you're telling matches the one investors are now actively looking for.