We didn't ask who holds the shares.
On a Thursday evening in Hangzhou, in a co-working space near the Grand Canal where I still run free workshops for the developers who arrived in this industry after the last cycle broke them, a 26-year-old engineer named Wen asked me a question I could not answer cleanly. "If I buy a tokenized share of Apple at three in the morning on a Saturday," she said, "who is legally holding my Apple?"
The room went quiet. Not because the question was hostile, but because it was obvious, and because a decade of building financial infrastructure has trained all of us to step over obvious questions on the way to something more interesting.
A few days later, Illia Polosukhin, co-founder of NEAR and one of the authors of the paper that gave the world the Transformer architecture, offered his own diagnosis of the same territory. Tokenized stocks, he said, are held back by fragmented liquidity. NEAR Intents, the product his team has shipped, wants to connect on-chain markets with Nasdaq liquidity.
I want to treat this seriously, because it deserves it. Polosukhin is not a marketer; he is a genuine technologist, and the pain point he names is real. I also want to treat it carefully, because naming a problem and solving a problem are different sentences, and this industry has spent a decade mistaking the first for the second.
Setting the table
Let me be precise about what is actually being said, because the news cycle will blur it within a week.
NEAR is a layer-1 blockchain whose mainnet went live in 2020. It has survived multiple market cycles, which in this industry is a rarer credential than any technical benchmark. Its co-founder, Polosukhin, was a co-author of 'Attention Is All You Need,' the 2017 paper that introduced the Transformer. That is not a marketing detail; it is a genuine signal about the caliber of the person making the claim. He has been building long enough to know what he does not know.
NEAR Intents is the product in question. It is an intent-centric system: instead of a user specifying a transaction, the user declares an outcome, and a network of competing solvers figures out how to achieve it. Solvers bid for the right to fill the intent, and the user gets the best execution among the bids.
Tokenized stocks are the asset class in question. A tokenized stock is a blockchain token designed to track a real, regulated equity. The most common design is a custodial wrapper: a licensed entity holds the underlying share and issues a 1:1 token that the holder can redeem. Backed Finance's bTokens, Dinari's dShares, and Ondo's tokenized equities all sit somewhere on this spectrum, with different jurisdictions and different levels of regulatory comfort. Robinhood's European offering runs on Arbitrum. The list is growing.
The narrative wrapping all of this is RWA, real-world assets, which has become the most institutionally respectable story in crypto over the past eighteen months. Tokenized treasuries matured first, because a government bond is a well-understood, well-custodied instrument. Tokenized equities are the harder sequel.
And we are reading all of this in a bear market. That matters for tone and for stakes. In a bull market, a story like this is a lottery ticket you buy for entertainment. In a bear market, it is a claim about whether the infrastructure you are relying on will still be standing when you need it. Survival matters more than upside now. So the question is not how big this could get. The question is what is actually load-bearing.
What an intent actually is, and why it is not the innovation being sold
I need to start with a correction that will annoy some people: intent-centric architecture is not new, and NEAR did not invent it.
By 2024, the pattern was already mainstream. CoW Protocol had been running batch auctions with competing solvers for years. 1inch Fusion turned the same idea into a Dutch-auction mechanism. Across built an intent-based bridge. UniswapX introduced a similar auction layer for swaps. The intellectual move, letting users declare desired outcomes and letting a competitive market of solvers find the path, is well established. It is good design. It is also table stakes.
So when NEAR Intents is described as a liquidity router for tokenized stocks, the correct mental model is not a breakthrough. It is a solver network pointed at a new asset class. The novelty, if there is any, is entirely in the asset class and the connection to traditional market liquidity, not in the mechanism.
This distinction matters enormously for anyone trying to figure out where value will land. If the mechanism is commoditized, then the router cannot charge much for routing. The margin has to come from somewhere else: exclusive access to liquidity, or exclusive access to the assets, or both.
The Nasdaq sentence is doing a lot of work
Here is where I want to slow down, because this is the load-bearing sentence of the entire pitch and it is the least examined.
Wanting to connect on-chain markets with Nasdaq liquidity sounds like a technical integration. It is not. Nasdaq is a centralized, regulated exchange with a formal market-maker system, a matching engine that operates under SEC rules, and a legal structure that does not have a socket into which a permissionless blockchain can plug.
So what could connecting actually mean? Based on my reading of how these systems have historically been built, there are three plausible mechanisms, and the announcement does not tell us which one is real.
The first is indirect mapping. The protocol does not touch Nasdaq at all. It trades a tokenized wrapper whose issuer holds the real share and whose price tracks the underlying through an oracle or through the issuer's redemption mechanism. In this world, Nasdaq liquidity is a metaphor. You are trading a derivative of a derivative.
The second is price reference. An oracle pulls Nasdaq quotes, and the on-chain market uses those quotes as a reference for pricing or for liquidation thresholds. This gives you price discovery without any actual access to Nasdaq's order book. It is useful. It is also not connecting to Nasdaq liquidity in any sense a traditional finance professional would recognize.
The third is a compliant intermediary. A licensed broker-dealer or an alternative trading system sits between the on-chain protocol and the regulated market, executing orders on the user's behalf. This is the only mechanism that could honestly be called a connection, and it is also the most legally and operationally expensive. It requires licenses, custody, clearing, and ongoing regulatory engagement.
I have spent enough time reading whitepapers to know that when a project describes an integration without specifying the mechanism, the mechanism is usually the version that is cheapest to describe and most expensive to build. That is not a smear. It is a pattern.
The critical point is this: the hard part of tokenized equities is not routing. It is the legal and custodial infrastructure that makes the underlying asset real. Everything above that layer, intents, solvers, routing, UX, is comparatively easy, which is exactly why it is crowded.
Fragmentation is real, and it is a symptom
Now let me give the claim its due, because the fragmentation diagnosis is correct.
Tokenized equities today are scattered. Backed issues on multiple chains. Dinari operates in its own regulatory lane. Ondo has its own distribution. Robinhood's European product lives on Arbitrum. Swarm Markets does its own thing under German regulation. Each issuer has its own liquidity, its own chain, its own set of venues, and its own set of restrictions on who can hold the token.
The result is a market where the same underlying exposure, Apple, exists in half a dozen forms, none of which are fungible with each other, each of which has its own depth, its own spread, and its own redemption terms. A buyer who wants the best price has to check every venue, compare prices across incompatible wrappers, and account for the fact that the cheapest quote might come from a wrapper they are legally prohibited from holding.
That is genuine fragmentation, and it does produce real costs. Spreads are wide because market makers cannot net exposure across venues. Slippage is high because depth is thin. Arbitrage that would normally compress these inefficiencies is blocked by the very legal boundaries that create the fragmentation.
But here is the contrarian turn, and I will return to it: fragmentation in tokenized equities is not primarily a technology problem. It is a legal problem wearing a technical costume. The same share exists in six forms because six different issuers found six different ways to be compliant in six different jurisdictions. No router can merge those into one fungible asset, because the reason they are separate is not that the pipes do not connect. It is that the law requires them to stay separate.
A solver network can find you the best price across the fragments. It cannot make the fragments the same thing.
Where the value actually lands
This is the analysis I keep coming back to whenever a middleware layer pitches itself as the solution to a market structure problem.
In any asset market, value concentrates at two points: whoever controls the asset, meaning issuance and custody, and whoever controls the customer, meaning distribution. Everything between those two points is a candidate for commoditization.
In tokenized equities, the asset side is held by issuers with licenses and custodians with real shares in real vaults. The distribution side is held by wallets, exchanges, and brokerages with retail reach. The router, the intent layer, the solver network, the aggregator, sits in the middle.
Middle positions are structurally precarious. If you do not control the asset, the issuer can route around you. If you do not control the customer, the wallet or exchange can route around you. The only durable defense is a liquidity advantage so large that routing around you is more expensive than going through you. That is a scale business, and scale businesses are brutal in their early years.
For NEAR specifically, there is a further question about value capture that the announcement does not answer. NEAR Intents is a product on top of the NEAR chain. Does it charge a protocol fee on tokenized equity trades? Does any of that fee accrue to the NEAR token, or to stakers, or to the treasury? The original discussion offers no fee design at all.
Without a fee design, the connection between processing tokenized stock volume and the NEAR token being worth more is indirect at best. More on-chain activity means more gas and more demand for blockspace, which is real but modest. It is not the same as a direct claim on the flow.
The cost structure nobody is modeling
Here is a piece of technical analysis that I think is missing from every conversation about tokenized equities, and it connects directly to my long-standing view on rollup economics.
Most tokenized equity trading, if it scales, will happen on rollups, not on base layers, because the whole value proposition is cheap, frequent, small-denomination trading. That makes rollup cost structure the silent determinant of whether this business works.
And rollup cost structure is about to change in a way that almost nobody has priced.
Since the Dencun upgrade introduced blob space for data availability, rollups have enjoyed artificially cheap data costs. That is the reason fees collapsed across L2s in 2024 and 2025. But blob space is a finite resource with a fixed supply per block, and demand for it is growing as more rollups, more data-availability layers, and more applications compete for the same blockspace. The economics are not complicated: fixed supply, rising demand, rising price.
My working estimate, and I have held this view since the early post-Dencun period, is that blob space will be saturated within roughly two years of Dencun, after which rollup gas fees will double again from their lows. I do not mean double from 2021 highs. I mean double from the current floor, which is where all of these tokenized-equity unit economics are being quietly modeled.
If you are building a product whose pitch is trading fractional shares of Apple for near-zero cost, and your cost floor is about to rise, you need to model that. The original discussion of NEAR Intents and tokenized stocks contains no cost analysis at all, no throughput figures, no latency figures, no fee estimates. That absence is itself information.
The yield illusion, again
There is a second cost-structure trap, and I have watched this movie before.
Intent networks need solvers. Solvers need incentives. If the natural flow of tokenized equity trades is not enough to pay solvers a competitive return, the network will subsidize them. And subsidized solvers produce subsidized liquidity, which looks like deep markets and real demand until the subsidy stops.
I spent 2017 auditing ICO token distributions and watching how community allocation turned out to mean insider allocation with better branding. I spent 2020 running workshops on Compound and Uniswap, explaining to retail users why an APY of 400 percent was not a yield but a transfer of the project's treasury into the hands of whoever arrived first. The pattern repeats: liquidity that exists only because it is paid to exist is not liquidity. It is a rental.
If tokenized equity depth on a solver network is driven by token incentives rather than organic flow, the depth will evaporate the moment the incentives taper. The announcement offers no evidence either way, because it offers no evidence at all. But it is the first thing I would check before believing any claim about solving the fragmentation problem, because the fragmentation problem can be masked by subsidy without being solved.
The solver trust assumption
One more technical point that deserves more attention than it gets.
The security model of an intent network depends on solvers behaving honestly and competitively. In the good case, solvers compete on price, users get the best execution, and the network is a genuine improvement over doing it yourself. In the bad case, solvers collude, censor, or extract.
This is a softer trust assumption than atomic on-chain settlement, where the transaction either executes correctly or does not execute at all. Intent systems reintroduce a middleman, a competitive one, but a middleman nonetheless, and middlemen have failure modes that atomic systems do not. A solver can refuse to fill an order. A set of solvers can coordinate to widen spreads. A dominant solver can become, functionally, the exchange it was supposed to replace.
For tokenized equities, this matters more than for tokens, because the underlying assets are regulated securities and the legal exposure of a solver that mishandles them is not theoretical. The trust assumptions in an intent network have legal dimensions that the pure-tech framing tends to hide.
The regulatory wall is the load-bearing wall
If I could make one point to everyone reading this, it would be this: the binding constraint on tokenized equities is not liquidity, and it is not technology. It is securities law.
A tokenized share of Apple is a security. It meets the Howey test almost by construction: money is invested, in a common enterprise, with an expectation of profit, derived from the efforts of others, namely the issuer and the custodian managing the underlying. There is no decentralization argument available, because there is nothing decentralized about a custodian holding real shares in a real vault.
That means every party in the chain who touches US retail users is potentially in scope for broker-dealer registration, alternative trading system rules, or an applicable exemption. The router that connects users to tokenized securities is not obviously outside that scope. Regulators increasingly apply a look-through principle: the fact that your architecture is neutral does not make your activity neutral if the activity is matching securities trades.
Nasdaq itself is a registered exchange. Its liquidity is not a public utility that any protocol can tap. Connecting to it, in any meaningful sense, requires the protocol, or a licensed partner, to operate within the SEC's rules. The original discussion does not mention regulation, licensing, custody, KYC, or jurisdiction once. For a product whose entire premise touches US securities markets, that is the single largest omission.
The realistic path, as I see it, runs through jurisdictions with clearer and more permissive frameworks first, the EU under MiCA, Switzerland, Singapore, and treats the US as a later, harder frontier. Any claim of a direct Nasdaq connection that skips this step should be read as aspirational until proven otherwise.
The contrarian turn
So here is where I land, and it is not where the headline wants me to land.
The consensus reading of this story is that NEAR identifies a real problem, fragmented liquidity in tokenized stocks, and positions its intent layer as the solution. If the solution works, NEAR wins.
The contrarian reading is that the problem being identified is real but misdiagnosed, and the proposed solution addresses the symptom rather than the cause.
Tokenized equities are fragmented because they are legally required to be fragmented. Each wrapper exists because a specific issuer found a specific compliance path in a specific jurisdiction, and those paths do not merge. A router can navigate the fragments; it cannot dissolve them. The day a single, fungible, globally recognized tokenized equity exists, it will exist because regulators converged, not because a solver network got clever. And on that day, the router that navigated the fragments becomes a feature, not a company.
We didn't build settlement for the assets we promised to trade. We built routing for them, and routing is the part that gets commoditized.
Which is why the middle of this stack is the most dangerous place to stand. The issuer holds the asset. The distributor holds the customer. The router holds a spreadsheet of other people's liquidity and a hope that nobody builds a direct connection around it. Sometimes that hope is rewarded. Often it is not.
What I would actually watch
I am not saying NEAR Intents is doomed. I am saying the evidence required to believe the thesis is not in the announcement, and I would rather name the missing pieces than pretend they are present.
Watch for the mechanism. If a real Nasdaq connection appears, it will come with a named licensed partner, a jurisdiction, and a custody arrangement. Until those exist, connecting to Nasdaq liquidity is a direction of travel, not a fact.
Watch for the fee design. If NEAR Intents charges for routing tokenized equity flow and routes a meaningful share of that fee to the NEAR token or its stakers, the value-capture question gets an answer. If there is no fee design, there is no claim on the flow.
Watch for organic depth. If tokenized equity liquidity on the network grows while incentives are flat or falling, that is real. If it grows only while incentives are rising, it is rented.
And watch the rollup cost curve, because every unit of tokenized equity economics being modeled today assumes a data-cost floor that I expect to double within two years.
The forward question
Wen's question, who is legally holding my Apple, is the question the whole industry keeps deferring. We have built fast, elegant, permissionless machinery for moving exposure to assets we do not custody, governed by rules we did not write, in jurisdictions we do not fully understand.
The tokenization of equities is going to happen. The only open question is whether the people who build it will treat custody, licensing, and settlement as the core of the work, or as details to be solved later by someone else. The projects that answer Wen honestly will be the ones still standing when the bear market ends. The ones that answer her with a roadmap will be the ones we are writing post-mortems about.
We didn't ask who holds the shares. It is past time we did.