There is a recurring illusion every time new technology emerges: when barriers to entry lower, everything becomes level. Camera phones make everyone a photographer, Spotify makes everyone a musician, and now AI makes everyone capable of coding. The logic is simple—if the baseline rises, competition becomes more open.

But what everyone overlooks is that the ceiling also rises. And it rises faster. Much faster.

This is no coincidence. It is the (power law) of scale( that has nothing to do with your intentions. Technologies promising equality actually produce the most aristocratic outcomes. Every time. No exceptions.

Take Spotify as an example. When Spotify launched, they did something radical—they provided access to distribution channels to every musician in the world, something previously only accessible to major record labels with huge marketing budgets. The result? An explosion in the music industry. Millions of new artists, billions of new songs. The baseline standards indeed rose.

But then something interesting happened: the top 1 percent of artists now capture a larger proportion of plays than during the CD era. Not smaller—larger. More music, more choices, but listeners no longer limited by geography are gravitating toward the best works. Spotify didn’t create equality; it intensified competition.

The same pattern occurs in writing, photography, software. The internet has produced the largest number of writers in history, but also created a much harsher attention economy. A tiny few capture most of the value. We’re surprised because we think linearly—assuming productivity will spread evenly like pouring water into a flat container. But complex systems don’t work that way, and never have.

Scale distribution follows a power law—not market anomaly or technological failure. It is the default setting of nature. Think of Kleiber’s Law—in all living creatures on Earth, from bacteria to blue whales, metabolic rate scales with body mass raised to the 0.75 power. This relationship is highly precise across nearly all forms of life. No one designed it; it’s just a form that emerges when energy follows its inherent logic in complex systems.

Markets are complex systems, and attention is a resource. When friction disappears—when geography, shelf space, distribution costs no longer act as buffers—markets converge toward their natural form. That form isn’t a normal distribution bell curve. It’s a power law.

AI will accelerate this process more than any technology before. The baseline is rising in real-time—anyone can release a product, design an interface, write production code in a single afternoon. But the ceiling is also rising, and faster. The key question: what truly determines your final position?

In an era where execution is cheap, aesthetics become signals. Remember Steve Jobs insisting that the circuit boards inside the first Macintosh had to be beautiful? Not the exterior—inside, unseen by customers. The engineers thought he was crazy. But he understood something that’s often mistaken for perfectionism, but is actually closer to a form of proof: the way you do something is the way you do everything.

Someone capable of making the hidden parts beautiful isn’t showing off quality; they’re personally unable to tolerate launching a bad product. This matters because trust is hard to build but easy to fake. We keep running heuristics to try to understand who’s truly excellent and who’s just pretending.

For most of the last decade, this signal was hidden. During the SaaS boom )2012 to 2022(, execution became so standardized that distribution became a truly scarce resource. If you could acquire customers efficiently, build a sales engine, hit the 40% rule—the product itself was almost irrelevant. Aesthetic signals drowned in the noise of growth indicators.

AI is truly shifting the signal-to-noise ratio. When anyone can produce a functional product, a beautiful interface, and a working code repository in a single afternoon, whether something is easy to use is no longer a differentiator. The question becomes: is this truly extraordinary? Does this person know the difference between “good” and “insanely great”? Even without coercion, are they sufficiently committed to closing that last gap?

This is especially true for critical business software—systems handling payroll, compliance, employee data. These aren’t products you can tinker with and abandon next quarter. The real costs of switching are tangible, failure modes serious, and the people implementing the systems are responsible for the consequences. Before signing a contract, they will run all heuristics of trust. An elegant product is one of the strongest signals. It says: the creators are very serious.

In a world where execution is cheap, aesthetics are proof of work.

I grew up in a small town in India with a population of 250 million. Every year, only about three students across India get into MIT. Without exception, they all come from expensive prep schools in Delhi, Mumbai, or Bangalore. I was the first in my state to get into MIT. I mention this not to boast, but because it’s a micro version of this argument: when the entry threshold is limited, background predicts outcome; when the threshold opens, deep talent always wins.

In a room full of people from prestigious backgrounds, I was the sure bet because of depth. That’s the only kind of bet I know. I studied physics, mathematics, computer science. The deepest insights in these fields don’t come from process optimization but from seeing truths others overlook.

My master’s thesis addressed mitigating stragglers in distributed machine training: when running large-scale systems, how to optimize this bottleneck without compromising overall integrity? When I was in my twenties and looking at the startup world, I saw a landscape where deep insights seemed irrelevant. The market rewarded go-to-market, not the product itself. Building something technically superior seemed naive.

Then, late 2022, the environment shifted. ChatGPT showed—in an intuitive and compelling way—that the curve had bent. A new S-curve had begun. The phase transition didn’t reward those best at adapting to the previous phase but those who could see the limitless potential of the new phase before others understood its value.

I stepped back and founded Warp.

The United States has over 800 tax agencies—federal, state, local—each with its own reporting requirements, deadlines, compliance logic. No API, no programmatic access interface. For decades, every payroll provider handled this the same way: by adding personnel. Thousands of compliance experts working manually, spinning in systems not designed for scale.

Traditional giants—companies like Paychex and others in payroll—built entire business models around this complexity. They didn’t solve it; they absorbed it into headcount and passed the costs to customers. It’s a profitable business, but built on a fragile foundation.

In 2022, I saw AI agents still fragile. But I also saw the curve of improvement. Someone deep in large-scale distributed systems and observing model evolution could make accurate bets: the technology that was fragile then would be very powerful in a few years.

So we bet: build a native AI platform from first principles, starting with the hardest workflows in this category—workflows that, due to architectural constraints, could never be automated by traditional giants. We wouldn’t fix the complexity; we would eliminate it at its source.

Three years proved this bet right. Since launch, we’ve handled over $500 million in transactions, grown rapidly, and served companies building the most critical technology in the world. Every month, the compliance data we gather, edge cases we handle, integrations we build make the platform harder to copy and more valuable to customers.

The moat is still in early stages, but it’s forming and accelerating.

But there’s a third variable that matters most, and it’s a critical mistake most AI-era founders make. There’s a popular meme in the startup world now: you have two years to escape the permanent lower class. Build fast, raise fast, or exit or perish.

I understand where this attitude comes from. The rapid pace of AI development creates a sense of existential crisis. The window to catch this wave seems very narrow. Young people seeing overnight fame stories on Twitter naturally believe the core game is speed—the winners are those who run the fastest in the shortest time.

But this is entirely the wrong dimension.

Speed of execution is indeed crucial. I believe in it deeply—it's even embedded in my company’s name )Warp. But speed isn’t the same as narrow vision. The most valuable companies in the AI era aren’t those that ran for two years to cash out but those that ran for ten and enjoyed compound growth.

The most valuable things in software—personal data, deep customer relationships, real switching costs, regulatory expertise—take years to build and can’t be quickly copied, no matter how much capital or AI capability they have.

When Warp manages payroll for companies across states, we gather compliance data across thousands of jurisdictions. Every tax notice resolved, every boundary case handled, every state registration completed—each trains the system to be harder to copy over time. It’s not just a feature. It’s a fortress, built because we’ve mastered it at a very high quality for long enough to create density of quality.

This compound growth isn’t visible in year one. Year two, it’s faintly emerging. Year five, it’s the entire game.

Frank Slootman, former CEO of Snowflake, has built and scaled more software companies than anyone today. He summarizes it simply: you must get comfortable with “being uncomfortable.” Not to run fast, but to make it a permanent state.

The fog of early startup—disorientation, incomplete information, the necessity of making decisions under uncertainty—won’t go away after two years. It just changes. New uncertainty replaces the old. The founders who survive aren’t those who find certainty but those who learn to move clearly through the fog.

Building a company is brutal. You live in a constant mild fear, sometimes accompanied by much greater fears. You make thousands of decisions with incomplete information, knowing that a series of wrong choices can lead to ruin.

The “overnight success” you see on Twitter isn’t just an outlier in the power law; it’s an extreme outlier. Optimizing for these cases is like training for a marathon by studying the results of people who took wrong turns and just happened to finish 5 kilometers.

So why do it? Not because it’s comfortable, not because the odds of winning are high. But because for some, not doing it feels like not truly living. Because the only thing worse than the fear of “building from scratch” is the quiet suffocation of “never trying.”

And—if you guess right, if you see truths others don’t value, if you execute long enough with aesthetics and conviction—the outcome isn’t just financial. You build something that truly changes how people work. You create products people love to use. You hire and unlock the best potential of the people in the business you build yourself.

This is a ten-year project. AI can’t change that fact. What AI does change is the ceiling that the founders capable of enduring until the end can reach in this decade.

So what will software look like in the future? Optimists say AI creates abundance—more products, more builders, more distributed value. They’re right. Pessimists say AI destroys software moats—anything can be copied in a day. They’re also partly right.

But both focus only on the fundamentals. No one is paying attention to the ceiling.

Thousands of point solutions will emerge—small, functional tools powered by AI, capable of solving narrow problems. Many won’t even be built by companies but by individuals or internal teams to solve their own issues. For low-threshold, easily substitutable software categories, the market will democratize truly. Competition will be fierce, margins razor-thin.

But for critical business software—systems managing cash flow, compliance, employee data, legal risk—the situation is very different. These are workflows with extremely low error tolerance. When payroll systems fail, employees don’t get paid. When tax reporting is wrong, the IRS comes knocking. When benefits are interrupted, real people lose protection.

Those choosing software must be responsible for the consequences. This responsibility can’t be handed off to a shoddy AI cobbled together with “feel-based code” in the evening.

For these workflows, companies will continue to trust providers. Among these providers, the “winner takes all” dynamics will be even more extreme than in previous software generations. Not just because network effects are stronger, but especially because large-scale AI platforms that operate across millions of transactions and thousands of compliance edge cases have a compounding advantage that makes it nearly impossible for new entrants to catch up instantly.

The fortress isn’t just a feature anymore; it’s the accumulated quality of high-standard operations in a domain that punishes mistakes.

This means the level of market integration in software will surpass the SaaS era. In the next ten years, in HR and payroll, there won’t be 20 companies with single-digit market shares. I expect two or three platforms will dominate most of the value, while a long tail of point solutions will hardly get any share.

The same pattern will happen in every category where compliance complexity, data accumulation, and switching costs combine.

The leading companies in this distribution will look very similar: founded by technologists with genuine product taste; built from day one on native AI architecture; operating in markets where giants today cannot respond structurally without dismantling existing businesses. They have been betting from the start on unique insights—seeing truths that AI hasn’t priced yet—and surviving long enough for compound effects to become visible.

The logic that brought us here is the logic I’ve described throughout this article: seeing the truth. Going deeper than anyone else. Building high standards that can be maintained without external pressure. enduring long enough to see if you’re right.

The leading companies in the AI era will be those who understand these principles: access isn’t a scarce resource, but insight; execution isn’t fortress, but aesthetics; speed isn’t advantage, but depth.

The power law doesn’t care about your intentions. But it rewards the right ones.