Infrastructure doesn't pitch well, yet it wins anyway
Something strange is happening in venture capital. The pitch format that built the industry — ten slides, clean traction and a simple but moving story — has been quietly selecting for the wrong companies with a skew that gets worse every cycle. The mechanism behind this is surprisingly easy to understand, and once you see it, you can't unsee it.
Two eras, one map
For twenty years, venture capital worked like an age of exploration.
Mobile, cloud, social, crypto: each was an unmapped continent. VCs funded the ships, and every pitch was a map to somewhere nobody had been with a simple winning move: get there first.
That era had a defining feature most people never noticed: When territory is unclaimed, while finding it is the hard part, explaining is pretty easy once it's found. "It's Uber, for X." Ten slides was enough, because the idea itself was the asset.
That era is ending. The foundational models went open, the tools once reserved for elite teams are globally available, and the big players aren't searching anymore. Instead, they're securing compute, absorbing talent, and fortifying. The game is shifting from discovering territory to holding it.
Problem is… the pitch format didn't shift with it.
The filter mechanism
This is the machinery of the whole argument, so let's slow down.
VCs, like all investors under time pressure, fund what they can evaluate. Call it legibility: clean revenue, simple models, visible traction. Legibility is still the most rational filter when you're seeing a thousand pitches a year.
During the exploration era, that filter worked, because legibility and opportunity pointed the same direction. A simple story ("Uber for X") often was a real opening, because the territory was genuinely empty.
Problem is, when the map fills in, the simple, legible ideas are exactly the ones that get found first, copied fastest, and defended worst. Meanwhile, the durable opportunities migrate to places the filter can't see: infrastructure, distribution control, deep technical moats. These take longer to explain since their power compounds slowly, and their stories don't fit on ten slides.
So the correlation flips.
The world it was calibrated to has changed, so the filter that used to predict opportunity now predicts a crowd.
That single flip explains most of what follows.
Follow one crypto startup
To feel the weight of this, trace one company.
An African crypto startup raises millions. The metrics are spectacular — high transaction volume, viral growth, exactly what the filter rewards. Investors see the numbers and see gold.
Now look underneath. The volume runs on rails the startup doesn't own. The users came for novelty and yield, not for anything the company controls. There's no infrastructure, no distribution lock, no switching cost. The startup is a toll booth on a road someone else built and is already planning to reroute.
Then the cycle turns. Users drift, volume drops, advertising ROI falls off a cliff. Nothing dramatic happens really, the company just gets quietly displaced, because there was never anything holding its ground.
Multiply that across a funding cycle and you get the pattern we've watched play out: millions poured into legible surges, evaporating on contact with consolidation. The metrics were real. They just measured the wrong thing… temporary motion that was bound to be replaced in a few months, rather than actual position.
What holding ground actually looks like
Now compare two companies that the filter would have scored poorly.
Moniepoint didn't launch another fintech app. It built payment rails — the boring, physical, agent-network layer underneath transactions. Today it processes around $17 billion monthly across 800 million transactions. Interpret that number: that's not an app people like. That's plumbing an economy runs through, which is why it's worth roughly a billion dollars and why nobody can casually replicate it.
TymeBank in South Africa fused cloud-native banking with physical kiosks in grocery stores, reaching customers no app alone could touch. A hybrid distribution machine with a valuation of around $1.5 billion.
Neither had a sexy pitch. Both built something structural where the system was weak, then controlled what everyone else overlooked. Instead of a bet to find gold, they were building the port every ship has to dock at.
The CPU test
Here's a thought experiment that makes the blind spot vivid.
Imagine pitching this: "We're going to take a rock, purify it into silicon, etch billions of microscopic switches into it, and sell slivers of it that perform billions of operations per second."
No traction. No simplicity. No viral loop. You'd be shown the door.
That pitch is literally how CPUs are made — and the companies that executed it became civilizational infrastructure. TSMC crossed $800 billion. NVIDIA passed $3 trillion. Intel's chips ran nine of every ten PCs for decades.
The more foundational the play, the worse it pitches
The value of infrastructure is precisely that it's hard to build, slow to compound, and impossible to summarize — the exact properties the ten-slide filter screens out. The next trillion-dollar company won't be the most recognizable idea in the room. It'll be the most structurally inevitable one, and it will probably sound absurd.
The filter has become folk wisdom
I recently watched a widely shared interview in which a distinguished older gentleman defined a startup as, in essence, a company with explosive growth — immense traction not as a goal, but as a fundamental prerequisite.
The comment section split cleanly in two: founders explaining why their startups weren't viral yet, and people complimenting his delivery while almost nobody examined the definition itself.
So let's examine it. Run it through everything above. If traction defines a startup, then every major NFT project was an incredible startup with spectacular volume and perfect legibility yet gone in eighteen months.
A definition that can't tell Moniepoint from a mint-and-dump isn't a definition. It's the exploration-era filter, recited as folk wisdom a decade after the era it was calibrated for ended.
And notice the definition only works backward. You can't know a company has explosive growth until it happens. So as a filter for what to fund, it's not a filter at all, but rather a scoreboard that can only be read after the game.
And that's the real cost of a broken filter. It doesn't just misallocate capital. It teaches a generation of founders to optimize for the thing that gets them funded instead of the thing that lets them survive, and then the comment sections fill with people apologizing for not being viral enough.
None of this is fatal
Not for venture capital, and definitely not for founders. If anything, it's the opposite: a filter this miscalibrated is an inefficiency, and inefficiencies are where the returns live: for the investors who evolve past it, and for the founders building in its blind spot.
And there's a personal bias underneath my read that I should disclose: I don't romanticize effort.
Camus said we must imagine Sisyphus happy, but notice what that happiness is. It's a posture struck inside futility. The boulder still rolls back and the smile is the consolation. Chasing virality is precisely that arrangement: you push the metric up, the cycle turns, it rolls back down, push again, and the folk wisdom asks that you call this passion.
I reject the premise.
Choose work that compounds, enjoy the doing of it, and you're not a "happy" Sisyphus. You're not Sisyphus at all. You're just someone building what they love, whose pushes accumulate, and whose surprises along the way are gains rather than coping. The allegory only binds you if the work resets. So pick work that doesn't.
So when I argue for the slow, illegible, structural play, I'm not preaching harder work. I'm preaching less wasted work. The most efficient path through a filled map isn't pushing against the filter. Efficiency requires building where the filter doesn't look, and letting compounding do what effort can't.
Is the map really full? (Where I was too clean)
When I first wrote about this shift (The End of Exploration, June 2025), I said the map was full and there were no more unknowns. That was too clean, and I want to correct it, because the correction is where the real opportunity lives.
The map is full where the last generation of investors learned their instincts: Western consumer internet, SaaS, the categories with twenty years of pattern-matching behind them. There, yes — exploration is over and the conquest logic applies in full.
But in emerging markets, both games are running at the same time. In Nigeria, whole layers of commerce still have no digital surface at all. We have genuinely unmapped territory, yet the consolidation wave is arriving on the same schedule as everywhere else, because AI has compressed the timeline.
That's a strange and brief condition: territory that is simultaneously unclaimed and about to be contested. The founders who win it won't be pure explorers or pure fortress-builders. They'll be the ones who claim ground and immediately build infrastructure on it because this time, there is no twenty-year grace period between discovery and siege.
For markets like ours, this is the thesis.
What this changes
For founders: the question is no longer "what's my idea?" It's "where will power accumulate, and what do I control when it does?" Distribution is destiny. This is why Meta pays reported nine-figure packages for AI leaders — not to build another model, but to make sure nobody out-builds them, regardless of the cost. Empires recruit generals, not scouts.
For investors: the fix isn't abandoning clarity, but redefining it. Clarity used to mean a simple story. Clarity now means being able to answer three harder questions:
- Where is the system structurally fragile?
- Who controls distribution over time?
- What can't be replicated by a well-funded fast follower?
Any diligence process that can't evaluate those will keep funding temporary motion and missing defensible positions.
Some capital will evolve and fund the next empires while others keep hunting "X for Y" in territories that emptied years ago. And when the map is full, there's only one direction left to move: through someone else's ground.
This piece revises and extends The End of Exploration (LinkedIn, June 2025). It pairs with The two variables that decide whether AI skips Africa. The same consolidation logic, applied to AI and emerging markets.