Jordi Visser's Bitcoin Thesis, Tested: Is Bitcoin Still a Liquidity Asset?

• The man, his incentives, and a thesis that runs against ours
• His system, stated at full strength
• The liquidity cycle is broken, and the recession that will not come
• Tokenisation: the channel he says will replace the central bank
• The users are not human, and the red zone
• Two unlisted companies, and the ghost rails already built
• Where he converges with us, and where he refined us twice
• The sharpest disagreement, and the tension inside his own position
• The claims that cannot fail, the corrections, and the claim he handled best
• The markers, and the verdict

Jordi Visser argues that bitcoin has stopped being a liquidity asset, that the crypto community’s debasement framing is a mistake, and that higher rates are good for it — so this episode states his system at full strength, reports where it agrees with ours, concedes the places it corrects us, and names the claims in it that cannot fail.

DeepDives · 2026 · An AI–human collaboration that weighs every source, mainstream or alternative, under strict rules of truth and logic.

§The man, his incentives, and a thesis that runs against ours

On the nineteenth of September, a macro investor with thirty years behind him told the host of a bitcoin channel that bitcoin is not a liquidity asset, that the entire crypto community has the mechanism wrong, and that higher interest rates are good for it. His name is Jordi Visser. He spent most of his career running money at institutional scale, he writes a well-read macro letter, and he appeared on that show most weekends through the summer.

That is not a small disagreement with us. We have said on air that real interest rates are the arbiter rather than the fiscal arithmetic, that bitcoin is a high-beta, long-duration risk asset which follows the technology index until the day it does not, and that the debasement trade is confirmed only when real yields fall back below one per cent. He is arguing that the sensitivity we describe was a phase rather than a property.

So this episode does what this series does with a serious guest. It states his system at full strength in his own terms, including the parts we would not use. It reports where his account agrees with ours, and the two places where he improved our framework, one of them uncomfortable. It names the sharpest disagreement, which is the whole point. It corrects three of his numbers, and it names the claims of his that cannot fail.

Then the incentives, because they shape the argument. Thirty years in markets, most of it institutional, and a research business built on explaining macro to people who were not taught it — genuine expertise. But this is not a one-off interview. He appears almost every weekend, and this episode is partly the launch platform for a crypto research product he is starting: a weekly video, ten verticals, a paper called Ghost Rails landing this week, and a forty-six-name index of companies he says represent the theme. He is selling the thesis while explaining it, which is not a criticism of his honesty — he is candid about being a long-term holder — but a person launching a research product has an interest in the theme mattering.

The advertising is unusually well matched to the content: an automated trading platform that rotates between stocks, commodities, exchange-traded funds and crypto using what it calls agentic strategies; a bitcoin credit card; and a large crypto conference in Singapore. Two of those three only make sense if his argument is correct.

Our own stake stays on the table: we hold hard assets and bitcoin, and our framework concludes in the assets we own. Nothing here is advice; it is a question about whose description of the machinery holds up.

§His system, stated at full strength

His central claim is not that bitcoin will rise, but why it rises. His words are that the enthusiasm in crypto came from an expectation of debasement, and that this is a mistake he disagrees with the entire crypto community about: bitcoin is not just liquidity. It behaved like a liquidity asset, he says, because nothing was built on it that ordinary people needed. Once there is real use, following the pattern of the smartphone, the growth comes from the technology rather than the credit cycle. And then the conclusion: if rates go up and crypto goes up anyway, the community’s model has been falsified, and what replaced it is demand from actual users.

Set beside that his reading of the past eighteen months. The broad money supply has been growing while bitcoin fell. The community calls this the alligator’s jaws opening and expects them to snap shut in a violent move upward; his reading is that the jaws will not close, because the relationship has already broken.

There is a strong version of that claim and a weak one. The weak version is that bitcoin’s price now responds to more than liquidity. The strong version is that it does not respond to liquidity at all. He argues the strong version — the weak version would be uncontroversial, and the strong one is the claim that can be lost.

§The liquidity cycle is broken, and the recession that will not come

The second pillar is the most quotable thing he says, and the most dangerous. He says the liquidity cycle is broken and that recessions are over, and his evidence is a list that can be checked. In 2020 the world shut its economy and assets collapsed, and the year finished higher. In 2022 policy rates went from zero to five in the fastest sequence on record with inflation in double digits, and the market fell and recovered. In 2023 a large bank went from apparently healthy to gone in a weekend, and the year finished higher. In 2025 tariffs arrived on everything. In 2026 the Strait of Hormuz was closed, one of the gravest interruptions to global energy supply on record. No recession followed any of it, and the line he draws is that he does not expect to see an eighteen-month recession again in his lifetime.

Two mechanisms, in his account. The old chain from money to activity to hiring has been replaced by a chain from money to margins. And there is a monetary bazooka, and the people holding it are willing to fire it at the market when needed. He notes that on two occasions there were two quarters of negative output and the official arbiters declined to call it a recession.

That is where a careful listener should slow down. The mechanisms are plausible and the list is real. The conclusion is a claim with no expiry.

§Tokenisation: the channel he says will replace the central bank

The third pillar is the mechanism he thinks replaces the one that broke, and it is the genuinely novel material in the interview. Liquidity is going to increase, he says, but not through the channels people watch. It will come from tokenisation, which he describes as unlocking dormant assets. His figure is nine hundred trillion dollars of assets, two-thirds of which sit and do nothing, and the mechanism is that these become programmable collateral: you post a claim on something you own for a period, borrow against it, and settle instantly, so the binding constraint on business stops being the time it takes to move money and becomes the velocity of the money itself.

Set that against the calendar and it stops sounding speculative. Two days before this conversation was recorded, the Securities and Exchange Commission issued what it called an innovation exemption: two conditional exemptions running five years, allowing venues to trade tokenised versions of listed stocks through automated market makers and liquidity pools without registering as exchanges. That is a real door opening, and he is right that it gets ignored because it is technical. His framing is that the stablecoin argument is the smaller one and tokenisation the larger.

Where we part from him is on quantity, not direction. The checkable adjacent figure is the size of the global equity market, around a hundred and twenty-six trillion dollars, and the work on tokenised collateral being published by the settlement infrastructure. The nine hundred trillion is a far broader aggregate, and the two-thirds fraction is not something anybody has measured.

§The users are not human, and the red zone

The fourth pillar answers a question we have not been able to answer: who are the users, and his answer is that they are not human. He uses a result that landed in September, when a laboratory published a solution to a long-standing problem in fluid dynamics that had been open since before Einstein, and how it was produced matters more than the result. Roughly ten thousand coordinating agents ran the search over about four days, exchanging millions of messages, with the output formally checked by machine. Human mathematicians could have solved it, and he says so; the difference is that they cannot share everything, because the credit belongs to whoever gets there, whereas agents have no ego to protect. He compares it to the Manhattan Project, which put its physicists together rather than isolating them, and notes the scale: a few hundred people then, ten thousand now, a million next.

Then his most quotable image, which he calls the red zone. Every human attempt at a great problem gets to the twenty-yard line and stops, because the last stretch needs more hands than exist. Agents, he says, finish those drives.

Two things belong beside that. He is right about the result and omits one fact: mathematicians have publicly disputed the credit for it. And it is the strongest part of his case, because it does not depend on any asset price. Either large numbers of agents can be pointed at a hard problem and produce something useful, or they cannot.

§Two unlisted companies, and the ghost rails already built

The fifth pillar is his diagnosis of the bubble, and here he is more precise than most people talking about artificial intelligence. He says the concern about an artificial intelligence bubble is really a concern about two companies. His claim, which he attributes to a well-known venture investor, is that two large laboratories will control somewhere between a third and half of all compute a year from now, that neither is public, and that neither releases its numbers. The market’s most important exposure is to two balance sheets nobody outside can see.

That is a version of the crowding-out argument we published with Doomberg in an earlier examination of this series, arriving from the opposite direction. He is not saying the debt is a threat. He is saying the concentration is the reality and the bubble language is misdirected, because what people call a bubble is a duopoly that has not been priced by the public market at all.

Then the frame he uses for crypto itself, which is his best metaphor. He calls crypto’s infrastructure ghost rails, by analogy with the Chinese developments built for populations that did not arrive: the capacity exists, it is unused, and it will light up when the tenants come. His timeline is that the internet’s white paper moment was 1994 and its smartphone moment was 2007, thirteen years, and that the interval between the bitcoin white paper and now is the same.

§Where he converges with us, and where he refined us twice

Now the convergences, because they are real and substantial. On the largest question, which is where this all ends, he is with us. His closing summary is that the fiat system is getting old and dying, that bondholders have received nothing for seven or eight years, and that people holding private credit are going to have pain getting their money back. That is our debasement thesis, our bond market episode and our risk dashboard’s next transition, in one sentence, from position rather than documents.

His description of the monetary response is the same mechanism as our fiscal dominance work. He calls it a monetary bazooka that the authorities will fire at the market when needed. We call it the fiscal authority overwhelming the central bank, and we trace it to an argument published in 1981 about why persistent deficits destroy central bank independence whether anyone intends it or not. He is describing the behaviour rather than the theory, and the behaviour is the same.

He also supplies something our stressor ranking could not: which part of the economy the pressure actually reaches. Higher oil and higher rates do hit earnings, he says, but they hit the earnings of the economy that existed before — home builders, consumer staples, the sectors that need cheap money to function. They do not touch the earnings of the companies whose products are being bought regardless. That is a sharper version of our claim that stress accumulates unevenly, and it explains why a stressor we ranked first at a hundred out of a hundred and twenty-five has not yet produced the event our ranking implied.

And then the refinements, and there are two, and the first is the one that stings. Our framework was built on the claim that real interest rates are the arbiter. Our own newest episode then had to record a problem: the central bank raised rates in the middle of September, the ten-year yield went back above five per cent, bitcoin rose, stocks held, and none of what our framework says should happen happened. We reported that as evidence against our own stage-three marker, and left it there.

He supplies the missing explanation, and it is a good one. This is not a liquidity-driven market at all, on his account; it is an earnings-driven one, and the mechanism connecting money to activity has been rerouted. His figures are checkable: earnings for the largest companies are growing at close to thirty per cent, and the three-month average of job creation excluding health care has turned negative for the first time outside a recession in more than twenty-five years. Money is reaching margins rather than payrolls, so the old link between tightening and falling asset prices is broken at the payroll end.

That is a real addition to our framework, and not a small one. Our thesis has been that the pressure from high real rates shows up as asset prices falling. His is that it shows up as a division of the economy into a part that is rate-sensitive and a part that is not, with the second part large enough to hold the index up. His version explains the last three months better than ours does.

The second refinement is a channel we have not published on, and it has the shortest distance between claim and consequence. Our published exits from the fiscal problem are growth, austerity, inflation and default, and our working assumption has been that the visible pressure shows up in the bond market and the answer arrives through the central bank. His proposal is a third path that requires neither: that the collateral sitting inert across the world’s balance sheets gets mobilised through tokenisation, releasing spending power without any central bank action at all.

That matters to us for a specific reason. Our falsification markers for the debasement thesis include the central bank buying long bonds in size, and a real yield falling back below one per cent, and both require the monetary authority to act. If his channel is real, there is a path to more liquidity that would not register on either marker, which means our instrumentation could be pointed at the wrong dial while the thing we predicted happens anyway. That is worth conceding, and it costs us nothing, because the tokenisation channel is measurable: on-chain settlement volume either rises against the new exemption or it does not.

He is also right about something smaller: that the memory layer, rather than the model, is the underrated part of the agent stack, a claim about where value accrues that has nothing to do with macro and is testable.

§The sharpest disagreement, and the tension inside his own position

Now the sharpest disagreement, and it is worth stating precisely because it is the best thing in the interview. Our published position, unchanged since the debasement re-test episode, is that bitcoin is a high-beta, long-duration risk asset, that it trades with the technology index until the day it does not, that seventy per cent drawdowns are within its normal behaviour, and that in a credit event it falls with everything else, because it is the first thing sold in a leveraged portfolio. The arbiter is the real interest rate.

His position is the opposite, and he states it as a disagreement with the entire crypto community rather than with us. Bitcoin, on his account, is a technology, and the defining property of a real technology is that it stops being sensitive to the cost of money once it is being used. Rates can rise and the price can rise, because the demand is no longer monetary. He is asserting that the correlation we have been publishing for a year has been retired by adoption.

This is a head-on, testable disagreement. And here is the honest thing to say about the evidence so far: it does not settle it. Bitcoin did rise into and after the rate increase, which is his evidence. But the market coverage attributes that rise to a return of risk appetite and to falling oil prices, which is a macro explanation of exactly the sort he says no longer applies. Both readings survive the same tape.

And there is a tension inside his own position that can be checked without appealing to any of our data. He says he is a bitcoin maximalist for the long term, and that the only asset he can name with confidence for the next thirty years, as distinct from the next three or five, is bitcoin. But the money he says is moving is going elsewhere. He described in an earlier appearance rotating fresh capital into Ethereum and Solana; here he says bitcoin is not the right place to look for the trade he is most excited about, and that his forty-six-name basket is where the asymmetry sits for the next five to ten years, most of it driven by the agent theme. He also argues for Ethereum and bitcoin together as the settlement layers institutions will be comfortable with.

None of that is dishonest. It is a portfolio position and a public position that are not identical, and the gap is informative. The strongest form of his argument — that adoption by agents makes the whole asset class repricing inevitable — does not require bitcoin to be the winner. His rhetoric does.

§The claims that cannot fail, the corrections, and the claim he handled best

Which brings the claims that cannot fail, and they should be named plainly, because they are the difference between a forecast and a world view. The first is that there are no more recessions: a claim that a thing will not happen for the rest of a lifetime cannot be scored by any observation short of that lifetime, and it has a structurally convenient property — if activity weakens and the authorities intervene, the intervention confirms the claim rather than testing it. The second is that bitcoin is the right asset for thirty years, a horizon longer than most investment careers. The third is the venture rule he quotes approvingly, that the steeper a valuation climbs the more likely it is to be undervalued, which cannot be wrong in either direction.

None of that makes the underlying argument false. It means the parts of his case that can be scored are the specific, dated, measurable ones: the five-year exemption that already exists, the earnings-versus-payrolls division already visible in the data, the agent capability already on the record, and his own price level for bitcoin by the end of October.

Now the corrections, and there are three, and two of them cut in his favour. Start with the one that does not. He says two laboratories will control between a third and half of all compute next year. As a claim about compute in aggregate it does not hold, and the measured finding runs the other way: frontier laboratories do not use most artificial intelligence compute. What is true, and nearly as striking: they are the marginal driver of the growth. Capital spending across the large cloud providers is running near seven hundred billion dollars this year, one laboratory alone has told investors it will spend about fifty billion on compute, triple the year before, and the multi-year supply contracts signed by those two companies run into the hundreds of billions. The correction narrows his claim: they dominate incremental demand and incremental spending, not the installed base, and two unlisted companies remain the concentrated exposure of the market.

The second correction is the figure that carried his tokenisation argument, and it should be treated as an aspiration rather than a measurement. The aggregate is his, and no one can measure the fraction, because there is no ledger of idle collateral. The solid adjacent facts are smaller and checkable: the global equity market is around a hundred and twenty-six trillion dollars and is being migrated to blockchain rails with the regulator’s encouragement, and the settlement infrastructure that clears American securities is publishing work on tokenised collateral.

The third is a rounding he does in his own favour and it barely matters. He says oil was a hundred and fifty dollars in 2008 and is a hundred today. The 2008 peak was nearer a hundred and forty-seven, and the current price is nearer ninety-five. His point survives: oil at today’s price against today’s household balance sheet is historically cheap, and that comparison checks out, because household net worth is around a hundred and ninety-six trillion dollars, up roughly three and a half times since those oil prices. That figure is accurate.

One more thing deserves careful handling, because it is the part of the interview most likely to be repeated out of context, and the part where he behaves best. He recounts a claim made publicly by a former presidential candidate, attributed to a person in a leadership position at one of the laboratories: that self-replicating code was released into the internet, and that the laboratories now believe they must build a synthetic internet to train on because the real one has been contaminated. His handling of it is the model. He says it sounds insane, assigns it a probability of around twenty per cent rather than repeating it as fact, notes that the claimant has a political reason to say it, and that a laboratory may have fed it to him as an influence operation. Then the correct posture: artificial intelligence is dangerous, and so are bioterrorism and nuclear weapons, and one should not assume that either the alarming or the reassuring version is the whole story.

§The markers, and the verdict

The markers, so this can be scored rather than remembered. Start with his own, volunteered unprompted and to his credit. He says that if bitcoin is below eighty thousand dollars by the end of October there is a problem with his thesis, and that the level above eighty-two thousand is where a large amount of traded volume sits, the point where holders gave up after last October’s fall. A prediction with a date and a level is the most useful thing a guest can leave behind.

Then the test that adjudicates between his account and ours, which is the one we already publish: whether real yields stay above two per cent while inflation falls toward target. That is the arithmetic that ended the last era in which hard assets had a decade-long lead, and his claim that bitcoin no longer responds to the cost of money has to survive it. If real yields rise and bitcoin rises, and that holds for a couple of quarters, he has retired our framework’s central claim and we should say so in a future episode.

Then two more. Does tokenised settlement volume actually rise against the five-year exemption, or does the door open onto an empty room like the ghost cities he borrows the image from. And does the agent-transaction claim show up where it can be observed, in on-chain payment volumes, rather than only in the valuations of agent-adjacent companies.

The verdict, in three parts. On the diagnosis of what is happening to the fiat system, he is with us, and he adds a physical description of the monetary response sharper than the one we published. On the mechanism connecting money to prices, he has improved us twice, and the first improvement is a correction to a claim we made too confidently: our framework predicted that high real rates would press on asset prices, and they did not. On bitcoin’s nature we hold our position, with a stated counter-argument rather than a dismissal.

The single sharpest thing he said is not about bitcoin at all. It is that the reason the developers of the technology are asking for a pause is not necessarily benevolence, and that the people who control the world’s money are the most credentialed and therefore the most invested in the framework being replaced. One can accept the observation without accepting the conclusion. But it applies to him as much as to anyone: the more training a person has in the old model, the more expensive it is for them to abandon it. That includes the rest of us.

What survives is this. The mechanism connecting money to employment has genuinely been rerouted toward margins, and that explains more of this year’s data than our own framework did. The regulator opened a five-year door for tokenised securities two days before he spoke, and almost nobody noticed. Large numbers of agents can now be pointed at a hard problem and produce something a committee of humans might not, and that is on the record rather than in a forecast. And his central claim, that bitcoin has stopped being a liquidity asset, is a real claim with a real test, which is the highest compliment this series pays to anything. It is also the claim that a year of our own published work argues against.

This has been a deep dive into whether bitcoin is still a liquidity asset, and what a macro investor with thirty years behind him says has replaced it.

Sources: this examination names its sources inline — the interview between Jordi Visser and the host of a bitcoin channel, with his commercial interests noted; the Securities and Exchange Commission’s innovation exemption issued two days before that conversation; the published figures for the global equity market, the tokenised collateral work of the settlement infrastructure that clears American securities, and the household net worth series; the result in fluid dynamics that landed in September, and the credit dispute around it; artificial-intelligence capital spending and the compute commitments of the two unlisted laboratories; earnings growth for the largest companies and the employment series excluding health care; and the market tape for bitcoin, the ten-year yield and oil. Every source is weighed with the same skepticism regardless of politics or business model. The audio version of this article, read by DeepDives, is available on Wavlake.