Boring Infrastructure

There’s a type of company that never gets the standing ovation at demo day. No consumer brand. The product isn’t photogenic. The growth chart doesn’t hit right on a slide deck optimized for applause lines.

It builds infrastructure. And I keep coming back to it.

The Quiet Part

Infrastructure companies operate at a layer most people never think about. The broker-dealer platform that makes secondary markets for private stock actually function. The GPU clusters and data centers that the entire AI economy runs on. The software layer that turns existing medical hardware into a programmable treatment platform. The diagnostic test that finally replaces an invasive biopsy and unlocks an entire downstream treatment market.

None of this is exciting on the surface. But these companies have a structural property that I think is genuinely mispriced at the early stage: they become load-bearing. Once a system depends on them, switching costs go vertical. Network effects don’t compound through virality… they compound through integration depth. A secondary markets platform gets smarter deal flow as more transactions run through it. A neuromodulation protocol layer gets more precise as more patient data trains the system. A diagnostic standard becomes the gatekeeper for an entire category of therapeutics.

That’s a very different animal than most of what gets funded.

Why This Matters

At seed and Series A, the question is rarely “will this market be big?” Most markets that attract smart founders are big enough. The real question is whether this team understands the problem at a level that gives them a durable edge… the kind that doesn’t evaporate when a bigger player decides to care.

Infrastructure founders tend to come from the industries they’re building for. They’ve felt the pain. They’ve watched clinicians work with blunt instruments when precision was possible. They’ve seen capital markets operate on phone calls and spreadsheets when the transaction volume demanded real tooling. They’re not constructing a solution in search of a problem… they’re codifying operational knowledge into software.

That’s a fundamentally different risk profile than a company betting on consumer behavior change. And it’s one I think the market consistently undervalues early on.

What I’m Actually Looking For

Four things, roughly in order of importance:

Is the problem structural? Not a trend, not a cycle… a gap in how an industry actually operates day to day. If the gold standard for detecting a widespread disease is still a biopsy, that’s structural. If private market secondaries still run through opaque, fragmented channels, that’s structural. Trends fade. Structural gaps get filled by whoever gets there first and stays.

Does the team have earned credibility? Domain expertise that can’t be replicated by a well-funded generalist team spinning up in six months. The kind of knowledge you only get by doing the job badly for years before you build the tool. The best infrastructure founders I’ve met don’t pitch a vision… they describe a system they already understand cold.

Is the go-to-market pull-based? Customers who already know they need this, not customers who need to be educated about why they should care. Education-heavy GTM at the seed stage is a yellow flag for me. It usually means the pain isn’t acute enough. The strongest infrastructure plays I’ve seen have customers pulling the product into their workflows before the sales deck is even finished.

Do unit economics improve with density? Revenue that compounds as the customer base deepens, not just widens. This is where infrastructure gets interesting… a compute provider’s margins improve with utilization density, a clinical software platform’s value compounds with every patient protocol it runs, a transaction platform’s data moat widens with every deal. The marginal cost of serving the next customer drops while the value of the network goes up.

The Unsexy Part

I’m comfortable being boring. The best infrastructure investments often look exactly like that for the first few years… steady, methodical, deeply woven into customer workflows.

Then one day the TAM re-rates because the market realizes the infrastructure layer was the bottleneck all along. The diagnostic that replaced the biopsy didn’t just detect disease… it unlocked a treatment market. The secondary markets platform didn’t just facilitate trades… it became the liquidity layer for an entire asset class. The company that looked like a modest vertical play turns out to be a platform.

That’s the pattern I’m underwriting. It’s not a guarantee… sometimes boring is just boring. But when the structural dynamics are right, the asymmetry is hard to beat. And most of the market is still chasing the flashier bet.