For a decade, the serious parts of finance treated crypto as a sideshow: an industry that had produced spectacular losses, a few useful payment rails and very little that a bank, an insurer or a pension fund actually needed. Citrini Research, in a note published on 8 October 2026, argues that this period is ending — not because crypto has changed, but because the rest of finance has.
Two forces are converging. On one side, AI agents are becoming the users of financial software. On the other, the assets institutions actually own — equities, treasuries, credit, commodities — are being tokenised and brought on-chain. Citrini’s image for this is a wall between decentralised finance and Wall Street that is starting to come down.
The wall between crypto and traditional finance was never only regulatory. It was the absence of a user who needed both sides at once.
The Agentic Bank Run
Citrini’s starting point is a thought experiment. Imagine a personal AI with access to your finances: it puts idle balances to work, makes investment decisions and refinances your mortgage on instruction. Consumer agents can already navigate websites and applications; the firm’s earlier note, “Agentic Reality”, argued that consumer software is the first thing they disrupt.
Taken far enough, the implications are uncomfortable. If millions of agents continuously move deposits toward the highest bidder, banks could lose the cheap, sticky funding that their balance sheets depend on. Citrini frames the question directly: is an agentic bank run coming?
The financial system was not designed for this. It is slow, fragmented and gate-kept by intermediaries, legal authorities and KYC processes — built, as Citrini puts it, by humans for humans. Those walls have so far protected the system from agentic interference. But information and resources are becoming machine-readable, programmable and transferable, and that logic does not stop at consumer data. It extends to money, bandwidth, compute and, eventually, financial assets themselves.
There are, broadly, two ways that can happen. Either existing institutions rebuild themselves around open, programmable standards, or they are eventually disrupted by a system that already has them.
Either incumbents rebuild around programmable standards, or they are displaced by a system that already has them.
A Solution That Found Its Problem
This is where crypto re-enters the story. The blockchain industry has spent fifteen years building always-on, programmable infrastructure where assets, ownership and settlement co-exist. Until recently it had limited purpose: there were no agents to use it, and the assets people wanted to own stayed inside the traditional system. Humans were left wrestling with seed phrases, wallet permissions and gas fees for the privilege of buying memecoins.
Both constraints are now loosening. Financial assets are being tokenised, and agents provide a class of user that does not care about the poor experience of a block explorer. Citrini’s comparison is the early internet: nobody learned networking protocols to go online, because browsers abstracted the infrastructure away. The same abstraction, the note argues, is now happening to blockchains — with agents as the browser.
The scale of the gap being bridged is large. Global equities stand at roughly $158 trillion; the entire cryptocurrency market is around $2.7 trillion. Tokenised real-world assets — gold, a listed stock, a treasury bill, a loan — are the mechanism by which assets from the first pool become available on the rails of the second, globally and around the clock.
The Inflection
Citrini identifies the moment the shift became visible. When the Iran conflict escalated over a weekend in March 2026, Hyperliquid, a decentralised derivatives venue, became one of the only liquid places where traders could price crude oil. Traditional markets were closed; an on-chain market was not.
Since then, adoption has moved into a higher gear. By Citrini’s count, on-chain assets under management in tokenised stocks and equities reached about $4.8 billion by early October 2026, up from a few hundred million a year earlier. In September, Solana’s decentralised exchanges temporarily processed more trades in a week than the New York Stock Exchange — much of it bots and meme pairs, but a signal of traction nonetheless.
Two structural changes sit behind the numbers.
- Tokenised equities have become composable — they exist outside the issuing venue and can move between applications and chains, the way stablecoins already do. Robinhood Chain’s “Stock Tokens” opened the floodgates, and other chains followed.
- Regulators are moving tokenised assets from an offshore experiment toward regulated US market structure. Two days after the CLARITY Act fell short in the Senate, the SEC issued a five-year innovation exemption for tokenised stock trading; Coinbase’s proposed US stock perpetuals are listed by the CFTC as approval pending; and Bloomberg now carries Hyperliquid perpetuals on the terminal.
The question for investors is shifting from whether tokenisation will be legal to who captures the economics.
Citrini’s reasons for “why now” are simple: the assets exist on-chain, the rails are mature, distribution exists through Robinhood, Coinbase and the banks, and regulation is beginning to accommodate rather than reject the technology. The firm’s conclusion — and the trade that follows from it — sits behind its paywall, and we do not reproduce it here.
What This Means for Risk
Citrini writes for investors. Solvendo’s interest is in what the same developments mean for the people who underwrite, lend against and insure the assets involved. Three implications stand out.
First, liability risk becomes continuous. A bank that models deposit stickiness on human behaviour — inertia, habit, the friction of switching — is modelling a customer that may soon be an agent with none of those traits. Funding risk that used to materialise over quarters may materialise over hours. That is not a hypothetical for the first institution whose depositors delegate cash management to software.
Second, collateral becomes programmable. When a treasury bill or a listed share exists as a token that can move between venues around the clock, the questions an underwriter asks change: where is the asset at 3 a.m. on a Sunday, what is it worth when the primary market is closed, and who holds the legal claim when the chain and the registrar disagree? The March crude oil episode shows that on-chain prices can lead; they can also diverge. Risk frameworks built on end-of-day marks will need to account for both.
Third, the speed of repricing rises. This is the thread that connects Citrini’s note to Solvendo’s own thesis on computational forecasting. If markets run continuously and participants are machines, then risk assessment that runs periodically is structurally behind. Continuous risk repricing — probabilities updated as evidence arrives, exposures recalculated as collateral moves — stops being an ambition and becomes a requirement.
When the users are agents and the markets never close, periodic risk assessment is structurally late.
None of this requires believing that the wall has already fallen. Much of the on-chain activity Citrini describes is speculative, the regulatory accommodation is an exemption rather than a settled framework, and the legal status of a tokenised claim in default has barely been tested. But the direction is clear enough that institutions should be modelling it now — not as a crypto question, but as a question about what happens to funding, collateral and risk when money learns to move on its own.
Source note: This article draws on Citrini Research, “Breaking The Wall — Has Blockchain Finally Reached Its Moment?”, published 8 October 2026, and the data and reporting cited within it (SIFMA, CoinMarketCap, DefiLlama, Blockworks, Reuters, Bloomberg). Figures are as reported by Citrini. The interpretation, the section on risk and the conclusions are Solvendo’s.