Author: Naman Bhansali
Compiled by: Deep Tide TechFlow
Deep Tide's Introduction: In the early stages of a new technology's proliferation, people often harbor an illusion of "technological equality": when photography, music creation, or software development become effortless, does competitive advantage simply vanish? Warp founder Naman Bhansali, drawing from his personal journey from a small town in India to MIT and his entrepreneurial experience navigating the AI-driven payroll sector, reveals a counterintuitive truth: the more technology lowers the barrier to entry (the Floor), the higher the industry's ceiling (Ceiling) rises.
In an era where execution becomes cheap, even something that can be AI "vibecoded," the author argues that the real moat is no longer mere distribution, but rather the hard-to-fake "Taste," deep insight into the underlying logic of complex systems, and the patience to compound returns over a decade-long scale. This article is not only a sober reflection on AI entrepreneurship but also a powerful argument for the power law principle that "democratizing technology leads to aristocratic outcomes."
Full Text Below:
Whenever a new technology lowers the barrier to entry, the same predictions inevitably follow: since everyone can do it now, no one has an advantage anymore. Camera phones made everyone a photographer; Spotify made everyone a musician; AI is making everyone a software developer.
These predictions are always half right: the floor *does* rise. More people create, more people publish, more people compete. But these predictions always miss the ceiling. The ceiling rises faster. And the gap between the floor and the ceiling—the median and the exceptional—doesn't shrink. It widens.
This is the nature of power laws: they don't care about your intentions. Democratizing technologies always produce aristocratic outcomes. Every single time.
AI will be no exception. It might be the most extreme case yet.
The Shape of Market Evolution
When Spotify launched, it did something truly radical: it gave any musician on earth access to distribution channels previously reserved for record labels, marketing budgets, and incredible luck. The result was an explosion in the music industry—millions of new artists, billions of new songs. The floor rose, as promised.
But what happened next: the top 1% of artists now capture a *larger* share of streams than they did in the CD era. Not smaller. Larger. More music, more competition, more avenues to find quality content caused listeners, no longer constrained by geography or shelf space, to cluster around the very best. Spotify didn't create musical communism; it just intensified the tournament.
The same story played out in writing, photography, and software. The internet produced more authors than any time in history, but also a more brutal attention economy. More participants, higher stakes at the top, the same basic shape: the very few capturing the vast majority of the value.
We are surprised by this because we think linearly—we expect productivity gains to distribute evenly, like water poured into a flat container. But most complex systems don't work that way. They never have. Power law distributions aren't a market quirk or a technological betrayal; they are nature's default setting. Technology didn't create it; technology just reveals it.
Think of Kleiber's Law. Across all life on earth—from bacteria to blue whales, spanning 27 orders of magnitude in body mass—metabolic rate scales to the 0.75 power of mass. A whale's metabolism isn't proportionally whale-sized. The relationship is a power law, and it holds with stunning accuracy across almost all life forms. Nobody designed this distribution; it's just the shape energy takes as it flows through complex systems following their internal logic.
Markets are complex systems. Attention is a resource. When friction disappears—when geography, shelf space, and distribution costs no longer act as buffers—markets converge to their natural shape. That shape is not the normal distribution's bell curve. It's the power law. The democratizing story and the aristocratic outcome coexist, which is why every new technology catches us off guard. We see the floor rising and assume the ceiling is following at the same pace. It isn't. The ceiling is accelerating away.
AI will drive this process harder and faster than any technology before it. The floor is rising in real-time—anyone can ship a product, design an interface, write production code. But the ceiling is rising too, and faster. The question to ask is: what determines where you end up?
When Execution is Cheap, Taste is the Signal
In 1981, Steve Jobs insisted that the circuit board inside the original Macintosh had to be beautiful. Not the exterior, the *inside*—the part the customer would never see. His engineers thought he was insane. He wasn't. He understood something that's easy to dismiss as perfectionism but is closer to a proof: the way you do anything is the way you do everything. A person who makes the hidden parts beautiful isn't performing quality; they are constitutionally incapable of shipping anything less.
This matters because trust is hard to build and easy to fake in the short term. We run heuristics constantly, trying to figure out who is genuinely excellent and who is just performing excellence. Credentials help but can be gamed; pedigree helps but can be inherited. The thing that's truly hard to fake is Taste—a persistent, observable commitment to a standard of quality that nobody asked for. Jobs didn't *have* to make the circuit board beautiful. That he did it anyway told you something about what he would do in the places you couldn't see.
For most of the last decade, this signal was somewhat obscured. During the heyday of SaaS (roughly 2012 to 2022), execution became so standardized that Distribution became the true scarce resource. If you could acquire customers efficiently, build a sales machine, hit the Rule of 40—the product itself almost didn't matter. As long as your go-to-market was strong enough, you could win with a mediocre product. The signal sent by taste was drowned out by the noise of growth metrics.
AI has flipped the signal-to-noise ratio. When anyone can generate a functional product, a beautiful interface, and a runnable codebase in an afternoon, whether something "works" ceases to be a differentiator. The question becomes: is this thing genuinely excellent?* Does this person know the difference between good and *insanely great*? Do they care enough to close that last gap, even when nobody is forcing them to?
This is especially true for business-critical software—systems that handle payroll, compliance, employee data. These are not products you trial and abandon next quarter. Switching costs are real, failure modes are severe, the people deploying them are on the hook for the consequences. This means they run all the trust heuristics before signing. A tasteful product is one of the loudest signals you can send. It says: the people who built this care. They cared about the parts you can see, which means they are likely to care about the parts you can't.
In a world of cheap execution, taste is proof of work.
What The New Phase Rewards
This logic has always been true, but for the last decade, the market environment made it almost invisible. There was a time when the most important skill in software wasn't even about software.
From roughly 2012 to 2022, the core playbook for SaaS was solved. Cloud infrastructure was cheap and standardized, dev tools were mature. Building a functional product was hard, but it was a *solved hard*—you could hire for it, follow established patterns, hit the bar with enough resources. The truly scarce thing, the thing that separated winners from the also-rans, was distribution. Could you acquire customers efficiently? Could you build repeatable sales motions? Did you understand unit economics well enough to pour fuel on the growth fire at the right moment?
The founders who thrived in that environment often came from sales, consulting, or finance. They lived and breathed metrics that would have sounded like gibberish a decade prior: NDR, ACV, magic number, Rule of 40. They lived in spreadsheets and pipeline reviews, and in that context, they were right. Peak SaaS created peak SaaS founders. It was a rational evolutionary adaptation.
And I was suffocating.
I grew up in a small town in an Indian state of 250 million people. Every year, only about three students from the entire country got into MIT. Without exception, they came from expensive prep schools in Delhi, Mumbai, or Bangalore—institutions built specifically for that goal. I was the first person from my state to get into MIT. I mention this not to boast, but because it's a microcosm of this essay's argument: when access is constrained, pedigree predicts outcomes; when access is open, deep people win. In a room full of pedigreed people, I was a bet on depth. It's the only bet I know how to make.
I studied physics, math, and computer science, fields where the deepest insights come not from process optimization, but from seeing a truth others missed. My master's thesis was on straggler mitigation in distributed ML training: when you run systems at scale, and parts fall behind, how do you optimize for that constraint without compromising overall integrity.
When I looked at the startup world in my early twenties, I saw a landscape where these depths seemed irrelevant. The market's premium was on go-to-market, not the product itself. Building something technically excellent seemed almost naive—a distraction from the "real game" of acquisition, retention, and sales velocity.
Then, in late 2022, the environment changed.
What ChatGPT demonstrated—in a way more visceral and shocking than years of research papers—was that the curve had bent. A new S-curve had begun. Phase transitions don't reward those best adapted to the previous phase; they reward those who can see the new phase's possibilities before others have priced them in.
So I quit my job and started Warp.
The bet was very specific. The US has 800+ tax jurisdictions—federal, state, local—each with its own filing requirements, deadlines, and compliance logic. There are no APIs, no programmatic access. For decades, every payroll provider handled this the same way: throw people at it. Thousands of compliance experts manually navigate systems never designed to run at scale. The legacy giants—ADP, Paylocity, Paychex—built entire business models around this complexity; they didn't solve it, they absorbed it into headcount and passed the cost to customers.
In 2022, I could see that AI agents were brittle. But I could also see the improvement curve. A person steeped in large-scale distributed systems, watching the model trajectory up close, could make a precise bet: the technology, brittle then, would become incredibly powerful within a few years. So we bet: build an AI-native platform from first principles, starting with the hardest workflow in the category—the one legacy giants could never automate due to architectural constraints.
That bet is paying off now. But the larger point is about pattern recognition. Technical founders in the AI era don't just have an engineering advantage; they have an insight advantage. They see different entry points, make different bets. They can look at a system everyone else accepts as "permanently complex" and ask: what would it take to *truly* automate this? Then, crucially, they can build the answer.
The titans of peak SaaS were rational optimizers operating within constraints. AI is removing those constraints and installing new ones. In the new environment, the scarce resource isn't distribution; it's the ability to see what's possible—and the taste and conviction to build it to the standard it demands. But there's a third variable that determines everything, and it's where most AI-era founders are making a catastrophic mistake.
The Long Game at High Speed
There's a meme in startup land right now: you have two years to escape the permanent floor. Build fast, raise fast, exit or die.
I understand where this mindset comes from. The velocity of AI progress feels existential; the window to catch the wave seems incredibly narrow. Young people seeing overnight success stories on Twitter reasonably assume the game is about speed—winners are those who run the fastest in the shortest time.
This is right in a completely wrong dimension.
Speed of execution *is* critical. I believe this deeply—it's in my company's name (Warp). But speed of execution is not the same as shortness of vision. The founders who will build the most valuable companies in the AI era aren't those sprinting for two years and cashing out. They are those sprinting for a *decade*, and compounding along the way.
The myopia is wrong because: the most valuable things in software—proprietary data, deep customer relationships, real switching costs, regulatory expertise—take years to accumulate and cannot be quickly replicated, no matter how much capital or AI capability a competitor brings. When Warp handles payroll for a multi-state company, we are accumulating compliance data across thousands of jurisdictions. Every tax notice resolved, every edge case handled, every state registration completed, trains a system that becomes harder and harder to replicate with time. This isn't a feature; it's a moat, and it exists because we operated at a high enough quality for long enough that it developed density.
This compounding is invisible in year one. It's a glimmer in year two. By year five, it's the entire game.
Frank Slootman, the former CEO of Snowflake, who has built and scaled more software companies than almost anyone alive, put it succinctly: get comfortable being "uncomfortable." Not for a sprint, but as a permanent state. The "fog of war" in a startup's early days—that disorientation, incomplete information, requirement to make move-decisions anyway—doesn't disappear after two years. It just evolves, new uncertainties replacing old. The founders who last aren't those who find certainty; they are those who learn to move clearly through the fog.
Building a company is brutally hard in a way that's difficult to convey to someone who hasn't done it. You live in a state of constant low-grade fear, punctuated by higher-grade terror. You make thousands of decisions with incomplete information, knowing a string of wrong ones can mean the end. The "overnight successes" you see on Twitter are not just outliers on the power law; they are the extremes of the outliers. Optimizing your strategy based on these cases is like training for a marathon by studying the times of people who took a wrong turn and accidentally ran 5k.
So why do it? Not because it's comfortable, not because the odds are good. Because for some people, *not* doing it feels like not really being alive. Because the only thing worse than the fear of building something from nothing is the quiet suffocation of never trying.
And—if you bet right, if you see a truth others haven't priced in, if you execute with taste and conviction over a long enough time horizon—the outcome is more than financial. You build something that genuinely changes how people work. You create a product people love using. You hire and enable people to do their best work in a thing you built with your own hands.
This is a ten-year project. AI doesn't change that. It never did.
What AI changes is the ceiling for the founders who stick around long enough to see it through.
The Unwatched Ceiling
So what does software look like on the other side of all this?
The optimists say AI creates abundance—more products, more builders, more value distributed to more people. They are right. The pessimists say AI destroys software moats—anything can be copied in an afternoon, defensibility is dead. They are also partly right. But both camps are staring at the floor. Nobody is watching the ceiling.
The future will have tens of thousands of point solutions—tiny, functional, AI-generated tools good enough to solve some narrow problem. Many won't even be built by companies, but by individuals or internal teams solving their own pain points. For some low-stakes, easily replaceable software categories, the market will be truly democratized. The floor is high, competition is fierce, margins are razor-thin.
But for business-critical software—systems that handle money movement, compliance, employee data, and legal risk—the story is different. These are workflows with zero tolerance for error. When payroll systems break, employees don't get paid; when tax filings are wrong, the IRS shows up; when benefits enrollments break during open enrollment, real people lose coverage. The people choosing the software are on the hook for the consequences. That accountability cannot be outsourced to an AI vibecoded together in an afternoon.
For these workflows, businesses will continue to trust vendors. And among those vendors, winner-take-most dynamics will be more extreme than in previous software generations. This isn't just because network effects are stronger (though they are), but because the compounding advantage of an AI-native platform running at scale, accumulating proprietary data across millions of transactions and thousands of compliance edge cases, makes catch-up from a standing start nearly impossible. The moat is no longer a feature set; it's the mass of quality accrued from operating at a high standard for a long time in a domain that punishes mistakes.
This means the software market will consolidate beyond the SaaS era. I don't expect 20 companies with single-digit market shares in HR and payroll a decade from now. I expect two or three platforms capturing the vast majority of the value, and a long tail of point solutions picking up the scraps. The same pattern will play out in every software category where compliance complexity, data accumulation, and switching costs compound.
The companies at the top of these distributions will look very similar: founded by technical talent with real product taste; built on an AI-native architecture from day one; operating in markets where incumbents cannot respond structurally without dismantling their own businesses. They placed a unique insight bet early—saw some truth about what AI enabled that wasn't yet priced in—and then held on long enough for the compounding to become visible.
I've been describing this founder. I know exactly who he is because I'm trying to be him.
I started Warp in 2022 because I believed the entire stack of employee operations—payroll, tax compliance, benefits, onboarding, device management, HR processes—was built on manual labor and legacy architecture, and that AI could replace it. Not improve. Replace. Legacy giants built billion-dollar businesses by absorbing complexity into headcount; we would build ours by eliminating it at the source.
Three years in, the bet is holding. Since launch, we've processed over $500M in transactions, are growing fast, and serve companies building the world's most important technologies. Every month, the compliance data we accumulate, the edge cases we handle, the integrations we build, make the platform harder to replicate and more valuable to customers. The moat is early, but it's there, and it's accelerating.
I tell you this not because Warp's success is pre-ordained—in a power law world, nothing is—but because the logic that got us here is the logic I've described throughout this essay: See the truth. Go deeper than anyone else. Build to a standard that holds without external pressure. Hold on long enough to see if you're right.
The exceptional companies of the AI era will be built by people who understand: access was never the scarce resource, insight was; execution was never the moat, taste was; speed was never the advantage, depth was.
Power laws don't care about your intentions. But they reward the right ones.
