I’ve been involved with blockchain since around 2018, with a tilt towards TradFi and enterprise use cases. In all that time, the thing that drew me in was never the price. Back in 2021 I wrote a post with the deliberately unhelpful title Don’t use blockchain. The argument was that blockchain is a bad database and a worse computer: slow, expensive, single-threaded by design. It has exactly one job worth paying that cost for: removing third-party risk. You stop having to trust that some operator keeps running the right software and doesn’t abuse their access. That’s it. Everything else you should minimize.

That hasn’t changed. What changed is that, for a few years, almost nobody wanted to talk about it.

The casino opened instead. Leverage, yield farming, tokens as lottery tickets, and an endless parade of “use cases” that turned out to be speculation with extra steps. The speculation isn’t the problem in itself (speculation is what markets do, and they always will), but it captured the narrative so completely that it became the entire public understanding of what this technology is for. Ask a normal person what crypto does and they’ll describe a casino, because for a while, that’s mostly what was on show.

So this is a post about what’s left when you tune the casino out. The honest answer is: less than the industry sells, more than the cynics think, and it’s finally arriving.

What doesn’t survive

Start by being ruthless, because most of it doesn’t make it.

A lot of the “serious, non-speculative” pitches fail on inspection. There’s DePIN: decentralized physical infrastructure networks reselling compute the hyperscalers already do cheaper, faster, and more reliably. A token-financed data center is just a worse-capitalized data center. There are “tokenize everything” schemes that leave the trusted intermediary (the registry, the custodian, the title office) exactly where it was, now with a token stapled on. And there’s the whole genre of private chains and permissioned ledgers that missed the point entirely: closed systems that still need to be synchronized between distrusting parties, the original problem reintroduced at higher cost.

Even stablecoins, the clearest real thing we have, are mostly noise by volume: the large majority of stablecoin transfers are trading and arbitrage, not someone paying someone. The genuine-payment slice is real and growing fast, but it’s a minority of the headline number.

There’s a simple tell for a casino dressed as utility:

If the economics only work because a token has to appreciate, it’s a financing scheme, not a use case.

The one question

Strip all of that away and I keep coming back to a single question: does this actually need a blockchain?

Not could it use one. Need one. And the test that survives is narrow. Do two or more parties who don’t trust each other need to agree on one shared record that none of them, and no operator sitting between them, can quietly rewrite, censor, or revoke? If yes, you have a real use case. If a centralized service with stablecoin payouts would do the job just as well, then you don’t. You have a database with better payment rails.

That’s the same argument I made in 2021, just sharpened into a test. The point then was to minimize your use of a blockchain down to the one thing it’s good for. The core value was never throughput or token price. It’s the decentralized future I keep coming back to: no single actor (a company, a government, the foundation itself) can revoke the properties you depend on. Most things don’t need that. The few that do are worth the entire cost.

What survives clusters in one place

Run the field through that filter and the survivors aren’t scattered evenly. They cluster, at least for now, around settlement: money and its plumbing.

That’s not a coincidence. Finance is simply where every condition the filter looks for shows up at once and at full strength: mutual distrust, a shared record everyone must agree on, rent-taking intermediaries, and real loss when one side pays and the other doesn’t deliver.

Public blockchains are the first infrastructure that settles shared state globally and in real time without a trusted middleman in the path, and finance feels that first because it has the most to gain. But nothing about the test is financial. It’s about removing a trusted intermediary, and that need turns up wherever distrust meets a shared record. Settlement is just where it bites first.

Why this is a 2026 post and not a 2021 one

If the idea is that old, why write it now?

Because the idea was right and early, and the gap has finally closed. Two things had to happen, and both have.

The technology caught up. Zero-knowledge proofs made it cheap to keep data and business logic private while still settling on a public chain. The thing I pointed at with Baseline back in 2021 (where you hash the shared view and coordinate around it without exposing anything) is now production-grade rather than a whitepaper. Verification got cheap enough to do at home, and proving costs collapsed faster than almost anyone expected.

And the rest of the industry finally moved, because the regulation that was always used as the excuse to do nothing arrived. MiCA, the EU’s crypto-asset rulebook, has been in full force since the end of 2024, and the GENIUS Act became federal law in the US in 2025. You can argue with the details, and people do, but the era of “we can’t touch public chains until there are rules” is over. There are rules now, on both sides of the Atlantic. And they point at the same architecture the Ethereum Foundation just wrote down: a neutral base layer that compromises on nothing, with compliance living upstream, in the contracts, the wrappers, the regulated products. Privacy at the base; KYC in the wrapper. Not a private chain. A neutral one, with the regulated parts bolted on top where they belong.

The US is still arguing out the rest. The CLARITY Act (the market-structure bill that decides which assets the SEC oversees and which fall to the CFTC) passed the House in 2025 and cleared the Senate Banking Committee in May 2026, with the White House pushing to get it signed by summer. It isn’t law yet, and the details are still contested. But the part that matters here is the fight over developers, and the direction of travel is clear enough: writing and publishing software does not, by itself, make you a regulated intermediary. As long as you don’t take custody, can’t freeze or move user funds, and don’t hold the admin keys, the code isn’t what gets regulated. Control is. That is the legal system drawing the same line that architecture does: compliance attaches to the controlling party at the edge (the custodian, the issuer, the wrapper) and finds nothing to attach to at a neutral base, because no one there is in control.

The tell that this is real is what’s arriving, and who is building it. The examples maturing now are, almost exactly, the ones I flagged in 2021. Paul Brody, who spent years building precisely this at EY, left in early 2026 to spin it out as an independent company, Nightfall. Decentralized identity is back too, as the answer to an AI world where proving you’re a real, unique human is suddenly hard. And settlement on neutral rails is already live: stablecoins moving real money, and tokenized Treasuries from the biggest asset managers used as onchain collateral.

None of this arrives overnight. With the cryptography largely solved, what’s left in the way is slower and less glamorous: law that’s still being written, the grind of wiring these rails into the systems that already exist, and the coordination problem of getting enough parties to move at once. The hard part stopped being technical.

The need doesn’t shrink. It expands.

Here’s the part I’d actually bet on.

During the casino, trustlessness looked optional, an ideological garnish on what is, in practice, mostly centralized convenience. The dominant stablecoins keep cash in a bank or a link to other TradFi instruments, with a freeze function and an issuer you have to trust. USDC holders were reminded of that when it broke its dollar peg in the Silicon Valley Bank collapse. Most users don’t mind. At today’s scale, they’re mostly right not to.

But that’s a small-stakes luxury, and the stakes are not staying small.

As value moves out to the edges (settled directly between parties rather than routed through a hub), the trustless framework has to follow it, because the thing it replaces, trust in intermediaries, is going down, not up. We are living through a fragmenting decade. Sanctions, capital controls, payment systems splitting along political lines, and institutions you would frankly rather not have to depend on. In that world, immutability and “no single actor can revoke this” stop being cypherpunk garnish and become load-bearing. They are what you reach for when the intermediary you used to rely on has turned into a liability, something that can be leaned on, captured, or switched off.

And this spans far more than money. It’s settlement, yes: the distinguished kind, immutable and decentralized, rather than the merely convenient kind. But it’s also identity, where you want to prove something is true without trusting whoever issued the claim. It’s provenance, credentials, the rising need to verify rather than trust as the cost of faking anything collapses. The list keeps growing, and it grows in the same direction every time: away from trust this intermediary and towards you don’t have to.

So the bet isn’t that decentralized beats centralized on convenience. It usually won’t. The bet is that the exception becomes the destination: that the trustless architecture, the niche thing, the thing I’ve kept coming back to since 2018, is what carrying real weight in a lower-trust world eventually requires.

We’re leaving the casino. Where we’re going, there’s no house at all.