Everything That Scaled Was Already Liquid
Why tokenization scaled with treasuries, why physical infrastructure is the hardest case, and what it actually takes to bring it onchain
Jacopo Buriollo, founder and CEO, Megawatt. Adapted from his keynote at ETH Belgrade, August 2026.
In brief
Tokenized treasuries, money market funds and short-dated credit scaled because they arrived onchain with three things already in place: a price set by a market, a legal wrapper that already worked, and buyers permitted to hold them. Physical infrastructure such as solar farms and battery storage has none of the three, so each has to be built, per asset. The engineering is largely solved. What remains is legal, commercial and slow.
Two numbers from September 2026 do not sit together.
In August, perpetual contracts tracking the prices of stocks, gold and other traditional assets traded $117.3 billion, according to a16z crypto. That is one month of trading, and it is more than three times the value of every tokenized real-world asset in existence, which Binance Research put at $34.18 billion in mid-September.
The same Binance Research report found that only about 12% of that tokenized value is being used in any onchain application at all. The rest is simply held.
Both live on the same chains, often on the same venues. Exposure to the price of a real-world asset turns over at a velocity no stock exchange permits. Ownership of the asset itself mostly sits still.
If putting an asset onchain produced liquidity, those two numbers would look far more alike.
I want to be careful here, because the stillness is not a failure. People holding tokenized treasuries hold them for yield. Not trading is exactly what they should be doing. The wrapper did not produce the activity in one case and did not prevent it in the other. It was never the variable.
Something else was doing the work, and it was doing that work before anyone tokenized anything.
Why tokenization scaled with treasuries first
Start with what has happened rather than what has been announced.
Treasuries moved onchain and stayed there. Money market funds. Short-dated credit. Real volume, real institutions, real balances. Tokenized equities arrived later and smaller, and they are growing fast.
Now look at what was supposed to happen. In 2017 the standard example for tokenization was real estate. It was going to be first, because illiquid assets had the most to gain from liquidity. Art was next. Then physical infrastructure.
There are live products in all of those, some of them good, run by serious people. Nine years on, their volumes are a rounding error next to the first list.
The prediction was backwards. What scaled was the most liquid, most standardized, most heavily traded instruments in finance. The things that already worked fine.
That is an odd result, and the explanation is worth more than the observation.
The three things every tokenized asset inherited
Every asset that scaled onchain arrived with three things already in place.
A price somebody else sets. Continuously observable, produced by a market rather than by the issuer. This matters more than it sounds. If the price comes from the issuer, every buyer has to trust that valuation and do the work independently. A treasury bill has a price made by thousands of people with no stake in your product. You inherit it for free.
A legal wrapper that already worked. Treasuries, funds and notes were legal constructs long before they were tokens. Somebody had already worked out what the holder is entitled to, which court hears the dispute, what happens in an insolvency. The token inherited a structure that had been tested by people going to court and losing. That is not a small inheritance. It is most of the value.
Buyers who wanted it and were permitted to hold it. An existing base, already eligible, already interested. This is the one that gets missed, and it is what those opening numbers demonstrate. Liquidity is a property of the buyer base. It has never been a property of the instrument. You cannot engineer your way to a bid. Tokenize something nobody wants and you get a token nobody wants, faster.
Tokenization made those assets move better. Faster settlement, cheaper transfer, better visibility, easier custody. All of it real, and all of it valuable.
It did not create any of the three things that made them work.
Why physical assets are the hardest case
Now take an asset that has none of the three, and that can also stop working.
Megawatt, the company I run, finances utility-scale solar and battery storage. Let me be specific about what failure means for something like that, because it is not what most people picture.
It is not that the price falls. It is that equipment degrades faster than the datasheet said. That the buyer of the output stops paying. That a grid connection lapses because somebody missed a deadline in a document nobody read carefully. And most commonly, the boring one: revenue arrives and it is well below what the model assumed.
Nothing broke. Nobody defaulted. The asset is fine. It just earns less.
A bond either pays or defaults, and both outcomes are well understood. A physical asset can underperform for a decade without triggering anything at all.
That case has no established answer, because financial instruments never needed one. They were legal constructs before they were tokens, and they inherited a century of case law about what happens when things go wrong. For a physical asset, that has to be built rather than inherited.
The normal day and the bad day
Here is the problem with all of this, and it is a problem of visibility.
On a normal day, a well-built claim on a physical asset and a badly built one look identical. Both settle in seconds. Both are transparent. Both transfer cheaply. Every observable property is the same.
You find out which one you have on the bad day. And on the bad day, none of those properties help.
A word about claims, because the term does a lot of work here. A claim is the legal version of who owes me, and what exactly do they owe me. Not who holds the token. Who is on the other side of the promise, and what could a court make them do.
So the questions that matter on a bad day are: who do I have a claim against, under which law, worth how much, and how long does enforcement take.
A tokenized stock has a bad day too. Companies fail. The difference is that somebody already built the answer, decades ago, and the token inherited it.
Transfer is solved. Legal finality is not.
Moving the token is a settled problem. Genuinely settled, and it is a real achievement. Settlement, transfer, visibility, custody. All better than what came before.
Knowing what the holder can actually claim, against whom, and in which country, is not.
There is a sharper version of this that took me a while to understand properly.
A blockchain entry can be practically impossible to change. In engineering terms, immutable. But legal finality, the point after which no court can unwind a transfer, is a different thing entirely. It comes from statute. It does not come from cryptography.
And the statutes that make settlement untouchable protect a specific list of designated systems. In Europe, the settlement finality framework dates from 1998, and it was not designed for the pilot regimes built on top of it. In the United States, the bankruptcy protections for financial transactions largely presuppose intermediaries sitting in the middle.
I want to be careful about how I put this, because it sounds like a criticism and it is not. The old machinery works. It is slow, unglamorous, and it has been tested in court. The new machinery is technically superior and legally less certain. A chain can settle faster, cost less, and be impossible to alter, and still carry less legal certainty than the thing it replaced, because it is not on the list.
Nobody finds that out until somebody tries to unwind a transaction.
Immutability is an engineering property. Finality is a legal one. Having the first does not give you the second.
Why tokenize physical infrastructure at all
Because when you finish building those three things, you get something the easy assets cannot give you.
Uncorrelated. A payment that arrives because electricity was sold does not care what funding rates do. Most onchain yield today pays you for taking crypto risk in a different shape: collateral risk, leverage, liquidity constraints, funding rates. They all deteriorate under the same conditions. An electricity payment does not.
Insurable. There is a physical thing. It can be inspected before you buy it, insured while you hold it, repaired when it breaks, and recovered if everything goes wrong. You cannot repossess a governance token.
Durable. The revenue comes from people and businesses needing electricity, which does not stop, rather than from a market that has to keep believing.
Harder to build, and what you end up holding is a genuinely different risk from everything else in the portfolio. That is the whole point of doing it.
What it takes to bring physical assets onchain
Megawatt finances utility-scale solar and battery storage across Central and Southeast Europe: ready-to-build projects, funded through construction, each farm held in its own company. And we are building all three things from scratch.
The price. There is no market quoting what a solar farm is worth today. It has to be constructed from metered output and the energy revenue that follows, then published continuously rather than quarterly. In a market with little operating history, the data is the only thing that builds trust, and it only exists once the asset is running.
The wrapper. Built per asset, so the claim is against something specific and enforceable, rather than a share of an average.
The buyers. Assembled one at a time, because nobody inherits a buyer base for this.
None of it inherited. All of it built. Almost none of it technical.
Inevitable and slow
There is a version of this argument that sounds pessimistic, and I do not mean it that way. I think all of this happens. I think physical assets end up onchain and I think it matters enormously when they do.
But inevitable and slow are entirely compatible, and people keep treating them as though they contradict each other.
Containerization was inevitable from about 1956, when the first container ship sailed. It took roughly thirty years to reorganize global trade. Not because steel boxes are difficult, but because ports had to be rebuilt, unions negotiated with, insurers had to work out how to underwrite a sealed box nobody had inspected, and customs authorities had to agree what a shipment even was.
Every one of those was a legal, commercial or institutional problem. None was an engineering problem.
We are at the same point. None of this takes anything away from what has been built. The infrastructure is good. Settlement, custody, transfer, transparency, all genuinely better than what came before. There are just other things that have to be built before it can carry this kind of asset, and most of them are not technical.
The engineering was never the constraint. It was just the part everyone could see.
Everything left is legal, commercial and slow. And worth doing.
Key questions
Why did tokenized treasuries scale before other real-world assets? Because they arrived with a market price, a tested legal structure and an existing base of eligible buyers. Tokenization made them move faster, but it did not create any of those three things.
Does tokenization create liquidity? No. Liquidity comes from buyers who want an asset and are permitted to hold it. Tokenization makes transfer faster and cheaper, but it cannot create demand.
What makes physical assets harder to tokenize than financial ones? They have no market price, no ready-made legal wrapper and no existing buyer base. They can also underperform for years without defaulting, a case financial instruments never had to handle.
Is a blockchain record legally final? Not automatically. Immutability is an engineering property. Legal finality, the point after which a court cannot unwind a transfer, comes from statute and covers specific designated systems.
What does it take to bring physical infrastructure onchain? A legal claim that holds per asset and is enforceable, output that can be verified without relying on the issuer, and a base of investors permitted to hold the position.
Sources: a16z crypto, analysis of perpetual contracts on real-world assets (September 2026). Binance Research, “The RWA Activation Era” (September 2026).