The Only Thing That Survives

I watched a company raise $4M, ship a beautiful product, get written up in TechCrunch, and die in eleven months. Not because the founders were dumb, they were sharp, the product worked, users liked it. The problem was simpler and more brutal than any of that: three teams built the same thing within six weeks of their launch, and one of them was free. That was 2024, and in categories where differentiation is thin and distribution is weak, it’s gotten meaningfully worse since then. The distance between “idea” and “functional product” has collapsed to almost nothing. What used to take a team of eight engineers four months can now be prototyped by one person over a weekend, which is genuinely wonderful if you’re a builder and terrifying if your entire business *is* the build. Most businesses being started right now are, in fact, just the build. Here’s the question nobody wants to sit with: if the thing you’re making can be reproduced by a motivated stranger with a credit card and a Claude subscription, what exactly are you selling? Not the technology, and not the features, those are table stakes now. You’re selling time. And time advantages, in a world where AI keeps compressing the clock, are not advantages at all. They’re a countdown. This doesn’t mean software is trivial. It means the first version isn’t the hard part anymore. The lifecycle is predictable enough that it should embarrass us. A startup launches, there’s heat, early users show up, and growth looks good for three months while the team hires and raises a round. Then the curve bends. About 90% of startups fail eventually, where about 20% of startups fail in their first year, and nearly half are gone by year five – right in the window where the initial excitement has faded and the real economics haven't kicked in yet. Only 3% of VC-backed software companies ever reach $100M in revenue. Think about the first wave of AI writing tools. Jasper raised at a reported $1.5 billion valuation in late 2022, and by 2024 its revenue had fallen from roughly $120M to around $55M while co-founders stepped down. The category blurred beyond recognition as writing features got bundled directly into Google Docs, Notion, and ChatGPT itself. Neither the technology nor the workflow was proprietary enough to matter, and what looked like a company turned out to be a feature waiting for a platform to absorb it. The gap between “first mover” and “fast follower” used to be measured in years, and now it’s measured in weeks. The copycats aren’t always worse, either; sometimes they’re better, because they get to learn from your mistakes without paying for them. So the real question isn’t “how do I build something good,” because good prototypes are easy while production reliability and distribution aren’t. The question is: how do I build something that gets harder to kill the longer it exists? Most founders, when they hear “defensibility,” think about moats: network effects, switching costs, economies of scale. They’ve read the blog posts and can recite the Buffett quotes. But the way moats are typically taught is almost useless for early-stage companies, because you don’t have network effects when you have forty users. The moat framework describes what mature companies look like, not how they got there, which is a bit like telling someone who wants to get strong that they should “have large muscles.” True, unhelpful. Here’s a better frame. Defensibility isn’t a wall you build but a direction you compound in: the accumulation of advantages that are hard to see from the outside and nearly impossible to replicate from a standing start. I count 7 real sources of this kind of durability, and most have nothing to do with your tech stack. 1. The first is proprietary data that improves with use. Not data in the generic sense, but structured feedback loops where every user interaction makes the product meaningfully better in a way competitors can’t shortcut. Spotify’s Discover Weekly wasn’t good because of the algorithm; it’s good because a decade of listening behavior from hundreds of millions of people is feeding that algorithm, and while you can copy the architecture, you cannot copy the training set. In fintech, this compounds differently — a company processing millions of transactions accumulates a loss history, a library of edge-case fraud rules, escalation playbooks, and chargeback learning loops that together form an underwriting intelligence no competitor can shortcut. That kind of knowledge isn’t code you can clone. It’s scar tissue from years of real money moving through real systems, and the key word is compound: if your data doesn’t compound, it doesn’t defend. 2. The second is trust at the infrastructure level. When you become embedded in how people make decisions, move money, or run their operations, you stop being a tool and start being a dependency. Stripe didn’t just process payments. They wove themselves into the financial plumbing of the internet so deeply that ripping them out would require re-architecting entire businesses. The cost of switching isn’t the fee but the migration risk, the integration work, and the institutional knowledge baked into every implementation. There’s a reliability dimension here that gets underestimated. Prototypes are easy, but production-grade trust: SLOs, incident response, auditability, change control, is not, and the startup with 99.99% uptime and a SOC 2 report has a moat that no weekend hackathon can replicate, because enterprise buyers don’t switch infrastructure over a better landing page. 3. The third is the permission moat. Regulatory licenses, capital requirements, vendor security reviews, and multi-year procurement contracts all move slowly, which is exactly the point. Coinbase is the clearest example. While competitors operated in regulatory gray zones, Coinbase spent years accumulating money transmission licenses state by state, registering with FinCEN, and obtaining MiCA authorization in Europe. When the SEC dismissed its enforcement action in early 2025, those years of compliance work transformed from a cost center into a structural advantage, with non-trading revenue reaching 40% of total revenue by Q3 2025. You can clone a trading interface, but you can’t clone 50 state licenses and an institutional custody reputation built over a decade. The mechanism is slow, then sudden. Auditors start recognizing your logs, procurement teams add you to their approved list, and policies get written around your reporting formats until you’ve become the default, not because you lobbied for it, but because you showed up first and kept showing up. That’s a permission moat, and it compounds like trust: invisibly, then irreversibly. 4. The fourth is distribution that can’t be copied with code. Marketing is rented attention, but distribution moats are structural: being the default integration in a platform’s marketplace, being preinstalled, occupying the procurement catalog that enterprise buyers reference before they even Google your category. Hyperliquid shows a different flavor of this. An 11-person team with no venture capital did nearly $3 trillion in trading volume in 2025 and generated over $800M in revenue, not through partnerships or preinstalls but by becoming the coordination point for decentralized derivatives. Their HIP-3 protocol let builders create their own perpetual futures markets on Hyperliquid’s infrastructure, and once 1.4 million users and hundreds of thousands of builders are operating on your layer, the cost of leaving isn’t a fee; it’s an ecosystem. 5. The fifth is community and brand that exist independent of the product. This sounds soft until you watch it work. Notion started as a niche productivity tool and became the default workspace for an entire generation of startups and operators. Not because it was the only option, but because users built their workflows, templates, and entire businesses on top of it. The template gallery is user-created, the ambassador program runs itself, and "Notion templates" is now a cottage industry with creators earning six figures selling workspace designs. You can clone the editor; you can't clone the ecosystem of people who made it theirs. 6. The sixth is accumulated capital and liquidity. This aspect gets ignored until you try to compete without it. Aave has built a safety architecture where its Safety Module holds over $450M in staked assets, its Umbrella system provides automated loss coverage, and the protocol has accumulated only $2.45M in total bad debt across seven years while managing over $26B in TVL. A new lending protocol can fork Aave's code in an afternoon, but it can't fork the liquidity depth, the loss history, or the trust that comes from processing $210M in liquidations during the February 2025 crash with zero new bad debt. Hyperliquid's open interest works the same way. The deeper the pool, the tighter the spreads, the more traders show up, and the harder it becomes for a competitor to offer equivalent execution. Liquidity begets liquidity. 7. The seventh is physical infrastructure. In a world where AI can replicate most software in days, atoms remain stubbornly hard to copy. Warehouses, sensor networks, fulfillment routes, manufacturing lines, licensed spectrum, charging stations. These require permits, construction timelines, local relationships, and capital expenditure that no amount of clever engineering can compress into a weekend prototype. It's getting easier to coordinate atoms with software, but moving them still requires showing up in the physical world, and that remains one of the most reliable barriers to fast followers. None of these companies started with a defensibility strategy. Patrick and John Collison weren’t drawing moat diagrams. They were obsessed with making payments less painful for developers. Brian Armstrong wasn’t calculating regulatory arbitrage in 2012; he wanted to make buying bitcoin not feel like a crime. That specificity was the seed of everything. Airbnb’s early team flew to New York and photographed apartments themselves, redesigned hosts’ listings by hand, and called users on the phone. None of that scales, but all of it built trust and understanding that did scale, because it informed product decisions that competitors, operating from spreadsheets, couldn’t see. This is the pattern: obsession with a specific user leads to compounding product insight, which leads to compounding trust, which leads to compounding switching costs. The moat gets built from the inside out. The temptation is to skip this process and focus on the addressable market, the viral coefficient, the growth hack. But growth without identity is just rented attention. When you acquire users without understanding them, you’ve built a business that looks great on a dashboard and has no immune system. The mistake I see most often is founders optimizing for the wrong thing at the wrong time, chasing growth before they’ve earned loyalty and building for the pitch rather than for the person using the product at 11pm on a Tuesday because they actually need it to work. The hardest version of this mistake is building for the demo, the product that looks incredible in a three-minute video. Demos raise money, but they’re performances, and performances end. The question that matters comes after the applause: does someone’s life or work get worse if this product disappears tomorrow? If the honest answer is no, you don’t have a company yet. There’s a test I keep coming back to, simple enough to fit on an index card, which is probably why it’s useful. If a team with $50 million cloned us tomorrow, what would they still not be able to replicate in three years? Be honest, not “our culture” or “our vision,” because those are non-answers. What specifically have you built, learned, or earned that requires time and cannot be purchased? A compounding dataset, a loss history trained on millions of real transactions, regulatory licenses that took a decade to accumulate, distribution access that’s structural rather than rented, a community that identifies with what you represent and not just what you sell. If the answer is “nothing,” you haven’t failed, but you should know what you’re holding — a feature, not a foundation. And features get absorbed. That’s not a judgment; it’s a physics problem. So here’s the part where I’m supposed to wrap this up with something clean. I won’t. What are you building that gets stronger the longer it exists? Not bigger, but stronger. Is the thing you’re working on accumulating advantages that compound, or are you running on a treadmill that speeds up every quarter? The economy doesn’t care about your roadmap, and AI doesn’t care about your head start. The only things that survive are the ones that were never just products to begin with: they were positions, earned trust, and permissions that took years to acquire. The answer to a question that someone asks every day, not because they were told to, but because they can’t imagine the alternative. If that’s not what you’re building, now is a good time to figure out what is. Keep on building, Sid