The Art of Arbitrage

Notes on bootstrapping, graduation, and the founder's hardest transition

Tether settles more daily transaction volume than most traditional payment networks. It got there by being the thing no bank would touch: a stable dollar that lives onchain, moves 24/7, and asks no questions.

When Bitfinex launched it in 2014, the use case was simply money movement between exchanges without touching a bank. It worked, and USDT became the dominant liquidity pair for spot and perps trading across every major CEX, so deeply embedded in exchange plumbing that when Binance tried to replace it with their own stablecoin on their own exchange, they couldn't.

Then Tron happened. Cheap transfers turned USDT into the default dollar across Latin America, Sub-Saharan Africa, and Southeast Asia. The users became people in high-inflation economies who needed a dollar that worked. Tether had built the rails for one audience and discovered they served a completely different one.

Today it's one of the largest holders of physical gold on earth.

The technology was almost beside the point. What mattered was the gap between what people needed a dollar to do and what the traditional banking system was willing to let it do. The biggest crypto companies found gaps like this and built growth loops on top of them before anyone else could close them.

The cleaner name for this is arbitrage. Not in the narrow financial sense, but finding a gap between markets or regimes that is big enough to matter, persists long enough to build on, and is hard for incumbents to close immediately. Exploiting that gap to bootstrap a growth loop. Then converting the temporary advantage into something durable before the window shuts.

The arbitrage has three steps: find the gap, build the loop, graduate the advantage. In crypto, almost every team that finds a gap can build a loop on it. The hard part is step three.

Steps one and two require fluency in crypto-native capital markets, things like where the liquidity sits or what kinds of products earn its attention. Step three requires an entirely different vocabulary: compliance, institutional trust, consumer product standards, partnerships with banks and fintechs. The founders who pull off all three are bilingual: native speakers of crypto capital markets who taught themselves the language of the mainstream. They're rare, but they’re the ones who build companies that last.

The Bootstrap

The first two steps of the arbitrage—finding the gap and building a growth loop—run on crypto-native fluency. The market demands it. 76.9% of North American crypto transaction volume in 2022-2023 was driven by transfers of $1 million or more. On Polymarket, >$50,000 in lifetime trading volume is enough to put you in the top 5% of all users. Crypto users aren’t a customer base; they’re a capital market: whales, market makers, and sophisticated power users who treat these platforms as core economic infrastructure. As a result, moats form later and decay faster than in traditional software. The ones that do prove durable (trust, infrastructure rails, deep integrations, liquidity depth, brand) take years to build and can't be replicated even when the underlying code can.

Early on, you have none of these moats, just community and an intimate understanding of where capital wants to go next. The fuel for the bootstrap comes from three sources. They often overlap, but each one tests a different dimension of crypto-native fluency.

Speculation is the most common bootstrap mechanism in crypto. There's a tendency in crypto discourse to either celebrate speculation uncritically or moralize about it. Both miss the point, but the moralizing version is probably worse. Speculation has frankly been the most reliable bootstrap mechanism in crypto's history.

Tether's bootstrap was pure exchange plumbing for speculators.

Circle found its moment during DeFi Summer, when yield farmers needed a trusted stablecoin to rotate through governance token farms. While the tokens being farmed often turned out to be worthless (which, to be fair, everyone kinda knew at the time), the demand for a reliable onchain dollar was very real. Circle didn't plan for DeFi Summer. But they'd already built the product (regulated, transparent, boring by crypto standards) that was ready when the frenzy needed a trustworthy dollar.

Ethena captured a different cohort. Yield-seekers were drawn to USDe, a synthetic dollar that generates yield from the basis between spot and perpetual futures markets. It hit $14.5 billion in TVL, became the third-largest stablecoin, and generated over $480 million in fees. The bootstrap was pure crypto-native financial engineering, and it scaled faster than almost anything in DeFi history.

But not every bootstrap starts with speculation.

Necessity drives adoption where existing financial rails are broken or exclusionary. High-inflation environments, expensive remittance corridors, populations locked out of dollar-denominated savings.

When the pain is acute enough, people switch without needing to be convinced of anything. They just need rails that work. RedotPay is a good example. It's a crypto card company that was founded in mid-2023 and grew to over $150 million in annualized revenue by the end of 2025. It lets people save in and spend stablecoin balances through a neobank-style interface.

You can't build that from an ivory tower in New York or London; you need to understand the situation on the ground. The same product pitched to someone in Manhattan is a novelty. Pitched to someone without reliable access to dollar savings or a usable card, it's a lifeline. That's the arbitrage.

RedotPay's edge is that their fluency already spans both sides—crypto rails and emerging-market distribution. And the retention is more durable because users aren't chasing yield; they're solving a problem that doesn't go away.

Subsidy (token incentives, points programs, venture-funded acquisition) can bridge the gap until organic demand sustains itself. Subsidized activity looks exactly like traction until the subsidy stops. When designed honestly and measured carefully, incentive programs can accelerate the discovery of genuine demand rather than merely fabricating it.

Hyperliquid is the clearest recent example of subsidy done right. A massive points program and airdrop brought traders to the platform, but the product underneath was genuinely best-in-class: fast execution, deep liquidity, and a trading experience that held up against centralized exchanges. Subsidies got people in the door. The product gave them a reason to stay. Activity didn't collapse after the incentives ended; it grew.

All three fuels tend to attract crypto-native participants first. Even necessity-driven adoption in emerging markets flows through crypto-native intermediaries (exchanges, wallets, onchain protocols) before reaching end users. Nearly every successful crypto company's early growth requires fluency in how crypto-native capital markets work. Without it, you're bootstrapping in a market you don't understand.

But crypto-native fluency alone isn't enough to graduate.

The Graduation

Getting the bootstrap right is hard. But the harder thing, and where most companies fail, is graduating from crypto-native traction to mainstream viability. Every company in this piece faced the same question at some point: does the product work when the crypto-native traders aren't the core audience anymore?

Plenty of teams have dominated the bootstrap phase and still lost.

The instincts that work in the bootstrap (speed, community intuition, shipping fast) can actively hurt you in the next phase. New user cohorts show up expecting consumer-grade support, intuitive interfaces, and compliance as baseline requirements. Growth channels shift from CT and airdrop campaigns to partnerships with banks, fintechs, and merchants. Revenue quality matters more than volume. The organization needs predictability it never needed before.

Circle is a good example of graduation. They spent years investing in compliance infrastructure before there was a regulatory framework to comply with, learning the institutional language while the rest of crypto was still speaking only to itself. When the GENIUS Act passed in July 2025, creating the first federal stablecoin framework, Circle's early bet looked less like caution and more like foresight. They IPO'd on the NYSE, finished the year with $2.7 billion in revenue and $75 billion in USDC circulation. The speculators are still there, they just aren’t the whole story anymore.

Ethena is what graduation looks like when it’s still uncertain. When Ethena’s yields compressed in late 2025, TVL fell by nearly half. The bootstrap ran on a single mechanism: the basis trade between spot and perpetual futures. The graduation requires diversifying away from it. Ethena is trying everything at once: exporting yield to institutional investors through regulated wrappers, USDtb backed by BlackRock's BUIDL fund, a perpetual futures exchange built on top of Hyperliquid called HyENA, a Stablecoin-as-a-Service model letting other ecosystems issue their own stablecoins on Ethena infrastructure. The backing itself has already shifted: perpetual futures went from 93% of USDe's collateral to >5% by mid 2026. Whether any of these new initiatives can replace the bootstrap's growth engine at scale is still open. But the direction is clear: Ethena is trying to become bilingual in real time, and the clock is running.

The monolingual failures are predictable. Teams mistake incentive-driven activity for product-market fit. Founders can't simplify their product for mainstream users. Compliance gets punted until "later," and later shows up with a subpoena. Organizations don't retool their hiring, KPIs, or roadmap even as the customer base hollows out underneath them.

And there's a subtler version of monolingualism: founders who try to skip the crypto-native bootstrap entirely, shipping crypto-powered products straight to mainstream users and wondering why nobody cares. The bootstrap is the ignition system (you can't skip it), but teams that keep optimizing for ignition after the engine needs to start running are just as stuck. Both are monolingual. They just speak different languages.

Not everything is arbitrageable, either. Tokenized Manhattan commercial real estate keeps getting pitched, and it keeps not working (every cycle, without fail), because the product appeals to neither crypto-native capital markets (low liquidity, no composability, unexciting returns) nor traditional investors (who already have access to real estate). At some point you have to ask whether the people pitching this have ever talked to the people supposedly buying it. That's not an arbitrage. That's a product without a market on either side.

What makes the graduation especially treacherous is that the arbitrage itself changes shape. In the bootstrap phase, you're exploiting a gap between crypto-native demand and an underserved market. In the graduation phase, you're working a different gap: between the infrastructure and trust you've built and the mainstream channels that want access to it. The gap shifted, and the founders who were already bilingual are the ones standing on the other side.

The Bilingual Founder

In the 1790s, Mayer Amschel Rothschild sent his five sons to five cities: London, Paris, Vienna, Naples, Frankfurt. Each one learned the local language and financial customs. They coordinated through a private cipher outsiders couldn't read, and their courier network outran governments. Plenty of European banking families had capital; it was the Rothschilds who could see across every border simultaneously, and move faster than anyone because they'd invested in fluency on every side of every gap that mattered.

The same structure applies in crypto, but the bilingual founders aren't all the same type.

Some founders are already bilingual. They've built in crypto-native capital markets and they understand mainstream or institutional language from prior lives or hard-won experience. Circle was this from the beginning. They're the sought after bets, and they're rare.

More common is the second-language founder: crypto-native with enough founder-market fit that developing institutional and mainstream fluency is a natural extension, not a metamorphosis. RedotPay looks like this: crypto rails meeting emerging-market distribution, with a team that understood both sides of that gap from the start. They certainly had to learn and adapt, but they did not have to become someone else.

Then there are the founders who never adapt.

Sometimes it's the one who won the bootstrap so convincingly that they can't imagine the game changing. The first language worked so well that learning the second feels unnecessary. The result is the same: the window for graduation closes while they're still optimizing for a phase they've already won.

Sometimes it’s the inverse: mainstream fluency without crypto-native instincts. Everything looks right on paper, but nobody shows up for the bootstrap, because the founder never learned how to speak to the market that needed to carry them through phase one.

Recognizing which type you're looking at is less framework than feel. You develop it over cycles the way the Rothschilds developed it over generations; each one sharpens the pattern recognition a little more.

The question for any founder is the same one that defines every arbitrage: what gap are you taking advantage of, how does your growth loop compound, and what does the graduation look like?

The arbitrage opportunities won't disappear as crypto matures. They'll just become subtler, more operationally demanding, and harder to see without fluency in both worlds. The Rothschilds' advantage compounded over generations because fluency compounds—each border crossed made the next one easier to read. The same is true here. It's the most durable advantage in this market, and it's impossible to fake.