Are You Buying the AI Story — or the Infrastructure Beneath It?

Only 7% of M&A deals in 2021 involved AI.

By 2025 that number had grown to more than seven times that figure — and the acceleration hasn’t stopped.

Most people read that statistic as a technology story. A signal of how fast AI is reshaping every corner of business and how urgently companies need to get ahead of it.

I read it as a capital story. And the two readings lead to very different investment decisions.

The narrative moves before the pricing does. The gap between those two things is where disciplined investors have always built real positions.

When a Sector Moves This Fast, Two Things Happen Simultaneously

The first is obvious: the compelling plays attract enormous attention, capital floods in, and multiples expand to reflect the excitement rather than the fundamentals. Valuations start pricing in a future that nobody can model with confidence — because nobody’s lived through it yet.

The second is less obvious, and more interesting: the spaces adjacent to the narrative — the ones that aren’t being written about, aren’t on the conference circuit, and aren’t in the pitch decks — start to look very different from a pricing standpoint.

The firms chasing AI-native acquisitions right now are competing on multiples that assume a future nobody can actually model. That’s not necessarily wrong — sometimes the narrative is right, and the early movers who paid up still win. But it’s a bet on the story. And the story is already priced in.

Meanwhile, the operators in adjacent infrastructure — the unglamorous businesses that every AI company depends on to function — are still being valued like it’s 2019.

The Infrastructure Beneath the Algorithm

Consider what it actually takes to run an AI company at scale.

Data centers that require massive, sustained electrical infrastructure. Physical real estate in specific markets where power and connectivity converge. Private credit facilities that fund the buildout before the revenue catches up. Supply chain businesses — cooling, cabling, logistics, maintenance — that have nothing glamorous about them and everything necessary about them.

None of these businesses are being written about the way AI software companies are. None of them are generating the investor enthusiasm that AI-native platforms are attracting. And none of them have had their valuations rerated to reflect the demand surge that’s quietly flowing through them.

That gap — between the demand reality and the pricing reality — is exactly the kind of opportunity that disciplined capital allocation has always been built around.

The question isn’t whether AI is changing everything. It’s whether you’re buying the story — or the infrastructure beneath it.

Five Decades of the Same Pattern

This isn’t a new dynamic. It’s the same pattern that has played out across every major technology transition I’ve watched over five decades of capital allocation.

During the dot-com era, the internet companies captured all the attention and the majority of the capital. The businesses that quietly won that decade were often the ones building the physical infrastructure the internet ran on — fiber, data centers, logistics networks. Boring. Necessary. Mispriced.

The mobile revolution produced a similar dynamic. Everyone was watching the app layer. The picks-and-shovels businesses — tower operators, semiconductor manufacturers, the unglamorous distribution businesses that got devices into hands — built durable positions while the narrative was pointed elsewhere.

The pattern is consistent: the narrative moves before the pricing does. The gap between those two things is where disciplined investors have always built real positions.

AI is not an exception to this pattern. It is the latest iteration of it.

What This Means for Capital Allocation Right Now

Private credit markets are currently seeing increased demand from AI infrastructure projects that require patient, flexible capital — the kind that traditional bank financing doesn’t serve well. The lending opportunity is real, the yields reflect genuine risk-adjusted returns, and the space is not yet crowded with capital chasing it.

Physical real estate — particularly in markets adjacent to major AI infrastructure buildouts — is experiencing demand dynamics that haven’t yet been fully priced into valuations. The industrial and specialized real estate serving data center ecosystems is a case study in narrative lag.

The supply chain businesses — the companies that handle power management, cooling systems, specialized logistics, and maintenance for large-scale compute infrastructure — are some of the most defensible businesses in the current environment. They’re also some of the least discussed.

None of this is a recommendation to ignore AI-native investments entirely. The technology is transformative and some of the companies being built at the application layer will generate extraordinary returns. The question is simply whether you’re being adequately compensated for the risk you’re taking on — and whether the pricing reflects the story or the fundamentals.

The Discipline of the Uncomfortable Position

The hardest part of this kind of investing is not identifying the opportunity. It’s holding the position while the narrative points somewhere else.

When the entire conversation in your sector is about AI-native acquisitions, standing in the infrastructure space feels contrarian in a way that’s uncomfortable to explain at an LP meeting. The thesis requires more context. The exit multiples are less exciting. The story doesn’t fit on a slide.

But that discomfort is often the signal. The best positions I’ve built over five decades were almost never the comfortable ones. They were the ones where the narrative hadn’t caught up to the fundamentals yet — and where the patience to hold through that gap was the actual edge.

The AI story is real. The infrastructure beneath it is equally real, considerably less crowded, and considerably more interesting from a pricing standpoint right now.

The question is which one you’re buying.

What asset class do you think the market is still most wrong about right now?

Joe Cook is a five-decade dealmaker, capital allocator, and steward of complex opportunities others pass on.

Performance-based. Stakeholder aligned. I win when you win.