Why I’m Still Building a Crypto Company in 2026

--

It’s been a year since one of my closest friends, Eugene, passed away.

Losing someone changes the way you think, and over the past year I’ve found myself writing more than I used to. Some of that is grief. Some of it is conviction: writing is one of the few things I don’t believe can be meaningfully outsourced to AI. Not because a model can’t produce sentences — it can, beautifully — but because writing, at least for me, was never really about communication. It’s how I think. It’s how I show myself what I actually believe. A model can find the locally optimal solution to almost any problem, but it still needs somewhere to begin, and someone still has to decide which questions are worth asking in the first place.

I started my career at Didi in 2017, building models on the pricing team — demand prediction, dynamic pricing for ride-sharing. Back then we didn’t call it AI; we called it machine learning and deep learning. What stays with me now, almost a decade later, isn’t the sophistication of any of it. It’s the data.

Didi knew almost everything. We knew where people lived and where they worked, what time they left the house in the morning, where they’d gone to school, where their families were, and — with unsettling accuracy — roughly how much they earned. More often than not, we could tell someone was about to order a ride before they had even opened the app.

Most people think of Didi as a transportation company. I’ve never quite believed that, and it isn’t really a finance company either. It’s a data company. And almost everyone who made it one handed their data over freely — nobody was paid, nobody was coerced. They simply wanted a product that made their lives a little easier, and the data was the price they never noticed they were paying.

Years later, I had an idea I was certain was good: a decentralized data-labeling company, something like Scale AI, except the people who created the value would actually own a piece of the network. I even had a name for it — Paymo. I never built it.

It took me a while to understand why, and the reason turned out to be the part underneath the part everyone focuses on. The hard problem was never the labeling. The hard problem was persuading people to give you their data in the first place — and people don’t do that for a few dollars. They do it when a product becomes useful enough that the trade turns invisible. Convenience, status, belonging, entertainment, connection: these have always been far stronger currencies than money. It sounds like one of those startup truisms, the kind of thing everyone nods along to, one plus one equals two. You’d be surprised how few people actually build as though they believe it.

The real moat was never the algorithm. The real moat is building something people genuinely want, because that is the only thing that earns you access to what can’t simply be bought.

One of the quieter costs of being different is that there is rarely anyone whose path you can just copy. Early in my career I sought out advice from people who, on paper, were far ahead of me — better titles, more experience, more money, more status — and I assumed, naturally, that they held the answers. It took me an embarrassingly long time to realize that most of what they told me didn’t apply to me at all. Not because they were wrong, but because they weren’t me.

Life is a strange game, multiplayer and single-player at the same time. Other people matter; relationships matter; networks matter. But at the end of the day you still have to live inside your own life, and if you are wired differently from the people around you, following their path faithfully will only lead you somewhere that was right for them. If the point of all this is to be happy, then surely what makes you happy ought to look different from what makes everyone else happy.

Investing, I’ve come to think, works much the same way. Most of it comes down to recognizing something undervalued before the rest of the market does, and then having the conviction to put real money behind that view. Fred Wilson once said that the best investments are usually the ones almost nobody else believes in.

Today, almost everybody believes in AI, and they’re right that it will reshape the world — I have no argument there. But markets don’t move on technology alone; they move in cycles of collective belief. People believed in crypto in 2018, and again in 2021, and again in 2024, and now that same belief has simply rotated into AI. The tools change. Human psychology rarely does.

A few days ago Avichal Garg — my mentor, my friend, someone I’ve looked up to for years — wrote that “this time is not different.” I can only imagine the pressure many crypto fund managers are under right now: difficult markets, impatient LPs, attention that has drained away toward something shinier. But I don’t think the real story is the market at all. It’s belief. When people stop believing in something, the price falls, the panic spreads, founders pivot, investors leave, and the narrative quietly rotates to whatever comes next. The only question that truly matters, the only one you have to answer for yourself, is whether your belief rotates along with everyone else’s.

People keep asking why I’m still building in crypto in 2026, and the honest answer is that I’m not sure I ever cared about crypto for its own sake. For years I thought I was building trading systems. Looking back, what I was really studying was something older and stranger: how human coordination works: how incentives quietly shape behavior, why some systems compound capital and talent while others slowly rot, and why ownership turns out to matter far more than most people ever stop to consider.

The internet gave us information. AI is giving us intelligence, and with it an almost vertiginous amount of productivity. But all of that productivity will eventually need somewhere to settle: a way to coordinate incentives, distribute rewards, allocate capital, and answer the oldest question we have: who owns what. That layer may not end up wearing the name “crypto.” But a world of autonomous agents, machine-generated labor, and near-infinite digital abundance will require a better system of ownership than the one we have inherited. Of that I have little doubt.

So when people ask why I’m still here, they’re asking the wrong question. It was never about crypto. The people in crypto are mostly trying to make money, and so, if we’re honest, are the people in AI. Plenty of them will tell you they’re chasing some generational mission, that they only care about the technology, and a few of them even mean it — but underneath all of it, everyone is making the same bet: that the future they’re building will become valuable because enough other people come to believe in it too.

That is how every technological revolution has worked, and this one is no different. In the end, the only thing that matters is deciding, for yourself, what you actually believe.

I already have. But you know, something to think about.