There is a particular rhetorical structure that has come to define the end-of-year macro letter in crypto, and Arthur Hayes has mastered it. The structure does not require data. It requires only memory and a calendar. You take the anxiety that preceded each previous expansion โ the block size war of 2017, the pandemic crash of 2020, the twin shocks of FTX and the rate-hike cycle in 2023 โ and you lay them out as rungs on a ladder. Then you point to the rung that has not yet been climbed and name it. For 2026, the rung is AI.
His argument, stripped of its cadence, is that every super-cycle begins with the market having to overcome a defining fear, and that the fear of 2026 will be the froth and fragility of the AI-plus-crypto complex. Once that fear is digested, he contends, the monetary floodgates โ long delayed, never cancelled โ will open, and asset prices will re-rate upward in the familiar, indiscriminate way. It is a bullish thesis with a narrative spine and almost no verifiable joints.
I spent part of the last quarter handling this framework against the data I trust, and the more I turned it over, the more it reminded me of something I learned auditing reserves during the 2022 collapse: the most dangerous claims are not the ones that are wrong, but the ones structured so that they can never be tested. A forecast that cannot fail is not a forecast. It is a posture, and postures are contagious precisely because they never have to be right.
To read Hayes correctly you have to understand what he is, which is not a forecaster in the conventional sense but a producer of narrative frameworks. He co-founded BitMEX in 2014 and helped build one of the first venues to turn perpetual swaps into a mass-market instrument. That pedigree matters, because his instincts were forged in the mechanics of leverage โ funding rates, liquidation cascades, the reflexivity of open interest. When he writes, he is rarely telling you what to buy. He is telling you how to feel about the water level.
Since leaving the BitMEX executive suite, he has run a blog โ the Crypto Trader Digest, now published under the banner of his family office, The Periphery โ that functions as a kind of liturgical calendar for a certain slice of the market. Each cycle gets its sermon, its label, its villain. The villain is always a fear. The fear is always temporary. The resolution is always liquidity.
This is the machinery I want to take apart, because the machinery is more instructive than any single call. A framework that reliably predicts "eventually" is not a forecast; it is a disposition. And a disposition that assigns a new mascot to every cycle is doing something subtler than analysis: it is manufacturing the narrative anchor that the market will later use to interpret its own mood.
The historical rungs are real events, which is what makes the ladder persuasive. The block size debate of 2015 to 2017 was a genuine existential schism that split Bitcoin into BTC and BCH. The March 2020 crash was a genuine liquidity event that saw BTC lose more than half its value in 48 hours. The 2022 to 2023 period delivered FTX's fraud and the fastest rate-hike cycle in four decades. Each was, in retrospect, a low. Each was followed by expansion. The pattern is true. The question is whether the pattern is causal or coincidental โ whether overcoming a worry produces a bull market, or whether bull markets simply occur after worries, because worries are always occurring. That distinction is where the entire thesis lives or dies. And it is where I want to bring in the only variable Hayes himself treats as load-bearing: money.
The Engine Is Liquidity, and Liquidity Is a Ghost
Hayes' framework is often read as a crypto-native thesis, but it is not. Strip away the crypto vocabulary and what remains is a pure macro-liquidity bet dressed in digital-asset clothing. His wager is that central banks โ above all the Federal Reserve โ will be forced back to accommodation, and that when they are, the tide will lift every risk asset, with crypto, as the highest-beta expression of that tide, rising the most. Everything else in the essay โ the AI fear, the cycle labels โ is ornamentation on a single load-bearing beam.
I understand the appeal of that beam intimately, because I spent a stretch of 2025 trying to quantify exactly how tightly crypto's price is bolted to it. Using eighteen months of daily data, I built a regression linking BlackRock's spot Bitcoin ETF inflows to changes in global M2 money supply, and I found a relationship that was both real and uncomfortably lagged: roughly a fourteen-day delay between a liquidity injection and the price appreciation it eventually produced. The signal was there. The timing was not instantaneous. Liquidity is a ghost; solvency is the body โ and ghosts move through walls on their own schedule.
That fourteen-day lag matters more than it first appears, because it is the difference between a tradeable thesis and a philosophical one. If liquidity reliably drove price within a fortnight, a disciplined reader could position around central bank balance-sheet announcements with a genuine edge. But the lag is not stable. It compresses during periods of high conviction and stretches toward infinity during periods of policy ambiguity. The relationship I measured was strongest precisely when the direction of policy was already obvious, and it dissolved when policy was contested. The liquidity beta is real, but it is conditional โ it pays off when the market already believes it will pay off.
This is the first crack in the Hayes framework. His thesis requires that liquidity arrive and then lift prices. What my model showed is subtler and less comforting: liquidity arrives, and prices rise only if the market has already decided that liquidity is coming. The causality runs through belief before it runs through balance sheets. A central bank can print for a year and accomplish nothing if the market does not trust the printing to continue; a central bank can do nothing at all and watch prices climb if the market merely expects it to act. The ghost moves first.
Hayes, to his credit, seems to sense this. His phrasing is never "the Fed will cut in March." It is always "the money printing will come." That is a deliberate choice of tense. It is a bet on a state of the world, not a date on a calendar โ and the difference between those two things is the difference between a forecast and a superstition.
Here is the part that the liquidity-beta framing tends to obscure. Crypto's sensitivity to liquidity is not a virtue; it is a symptom. An asset class whose price is set primarily by the availability of money, rather than by the cash flows it produces, is an asset class that will rise the most when money is cheap and fall the most when it is not. That is the definition of high beta, and high beta cuts both ways. Hayes' framework treats the upside of that beta as the thesis and the downside as the fear to be overcome. But the same mechanism that produces the spectacular expansion produces the spectacular contraction โ and the ladder has no rung for the contraction, because contractions do not get labeled. They just end the climb.
The Transmission Chain, and Where It Gets Stuck
If the liquidity thesis is the engine, the transmission chain is the drivetrain โ the sequence of mechanical steps that carries a central bank decision from the policy rate down to the price of a DePIN token. Hayes' essay skips this entirely, and the omission is not accidental, because the chain is where the thesis is most likely to slip.
The textbook version is well known. A central bank eases, the risk-free rate falls, the discount rate applied to every future cash flow drops, and risk assets re-rate upward in order of their duration and their beta. Bitcoin, with no cash flows at all and a valuation that rests entirely on future expectation, is the longest-duration asset in the world. In this framing, crypto is simply the far end of the duration curve, and liquidity is the force that bends it. The logic is clean. The problem is that the chain has several links, and each one can seize.
The policy decision is a political act as much as an economic one. The banking system must actually pass easier conditions through to the entities that trade risk. The crypto on-ramp โ the stablecoin rails, the fiat gateways, the custody arrangements โ determines whether macro liquidity can even reach the asset class. And the market's own belief structure, discussed above, determines whether arriving liquidity is spent on risk or hoarded in safety.
I spent six months in 2024 watching the third link up close, embedded in a research team monitoring the State Bank of Vietnam's pilot for a digital dong. I documented more than two hundred technical inefficiencies in the central bank's distributed ledger implementation โ latency spikes, privacy leaks in the settlement layer, architectural choices that optimized for control over throughput. I refused to publish until I had mapped the entire settlement architecture, which delayed my first major report by a month. The lesson that came out of that deep dive was not really about Vietnam. It was about the friction between sovereign monetary policy and decentralized technical standards โ two systems with incompatible assumptions about who gets to see the ledger, who gets to reverse a transaction, and what finality even means. A central bank that issues digital currency is not extending the liquidity chain to crypto; it is building a parallel chain with a checkpoint at every link.
This matters for the Hayes thesis because the transmission of liquidity into crypto is not a single pipe. It is a set of gates, some of them controlled by the very institutions least interested in watching liquidity leak into unregulated assets. The macro tide can rise and still fail to reach the shore, if the channels are dammed. When I look for evidence that liquidity is actually reaching crypto, I do not look at M2 alone. I look at the net supply of stablecoins, because that is the closest thing we have to a real-time measure of fiat that has already made it through the gates. If stablecoin supply is flat while M2 expands, the chain is seized, and the liquidity beta is theoretical.
And the gates are not only technological. They are also legal. When Hong Kong's virtual asset licensing regime matured, the market read it as the territory embracing innovation. The plumbing read it differently: as a jurisdictional bid to absorb the flows and the firms that Singapore had spent a decade attracting. Licensing is a channel โ a way of determining which liquidity reaches which assets, under whose rules. A bull market that routes through a handful of licensed venues is a different bull market from one that routes through the open rails, and the difference is invisible on a price chart.
The Safest Word in Macro
Let me be precise about what "eventually" buys you. If you predict a recession that never arrives, you are wrong, and the market tells you so. If you predict that money printing will eventually resume, you are almost certainly right โ because over any sufficiently long horizon, the record of fiat monetary systems is one of near-monotonic expansion. Central banks print in crises and print in calm; they print to fight deflation and, with slightly more reluctance, print to accommodate growth. To say that money printing will eventually come is not a prediction. It is a description of the regime we live in.
This is not a cheap shot. It is the central epistemic problem of the entire early-bull thesis. A claim that cannot be falsified within the horizon over which it is being deployed is not a claim you can act on. It is a mood. And moods are contagious in a way forecasts are not, because moods do not require the reader to be right โ they only require the reader to feel less alone.
The asymmetry is stark. If Hayes is wrong about the timing โ if the Fed holds tight for another eighteen months, if inflation re-accelerates, if the fiscal picture forces a longer squeeze โ he is not wrong in any way his framework can register. The printing has not come yet. The thesis is intact. It is always intact. Meanwhile the reader who levered into the thesis on the strength of "eventually" absorbs the loss, because leverage does not have the luxury of waiting for a state of the world to arrive. Liquidity can be patient. Solvency cannot.
I have watched this exact asymmetry destroy portfolios. During the 2022 de-pegging episode, I worked with two independent cryptographers to audit the reserve transparency of three major stablecoins, and I found a fifty-million-dollar discrepancy in the proof-of-reserves reporting of a mid-tier algorithmic issuer. The team behind that coin had a thesis too: the peg would eventually hold, the reserves would eventually be sufficient, the market would eventually regain confidence. Every one of those claims was, in a narrow sense, defensible. None of them was falsifiable before the coin collapsed. The people who held through the collapse were not holding because the thesis was strong. They were holding because the thesis was structured to be unfalsifiable โ and an unfalsifiable thesis is indistinguishable from no thesis at all, right up until the moment it becomes a very expensive one.
That is the shape of the risk in the Hayes framework. It is not that he is bullish. It is that the bullishness is engineered to survive contact with reality, which means it cannot help you survive contact with reality either.
The Named Segment: Why AI, and Why Now
There is a second, quieter claim buried inside the essay, and it deserves more scrutiny than the headline. By singling out AI crypto as the fear of 2026, Hayes is making an implicit structural argument: that the AI-plus-crypto complex has grown large enough, and frothy enough, that its reckoning could drag down the entire market's mood. You do not name a segment as the cycle's defining worry unless you believe the segment has achieved sufficient weight to matter.
I think he is half right, and the half he is wrong about is the more interesting half.
Start with the half that is right. The AI-plus-crypto category has, over several cycles, absorbed an enormous share of narrative oxygen. It spans at least four loosely related clusters: decentralized physical infrastructure networks that sell compute and bandwidth as tokens; AI-agent tokens that promise autonomous economic actors; data and model marketplaces that claim to monetize training corpora; and the broad category of AI-adjacent memecoins that are honest about being neither AI nor infrastructure. The technical paths within this space are wildly divergent, and the shared thread โ that blockchain can coordinate AI resources โ is more of a marketing posture than a settled architecture.
Late in 2026 I designed a theoretical framework for exactly this kind of system: a model in which ten thousand AI agents perform autonomous audits and generate roughly two million dollars in daily transaction volume, settling micro-payments on-chain for data verification. I spent two months refining the game theory before I was willing to publish, because the incentive structure has to be mathematically sound or the whole thing is theater. And here is what that exercise taught me about the category: the demand side of AI-plus-crypto is almost entirely unproven. The supply side is trivial โ anyone can spin up a token and call it an agent economy. The demand side requires that autonomous agents actually need to transact on-chain, at a scale that justifies the infrastructure, for reasons a centralized database could not serve more cheaply. That demand has not arrived. It may arrive. It has not.
So when Hayes names AI as the fear of 2026, he is naming a segment whose valuations run well ahead of its verifiable usage. That is precisely the profile of a category capable of producing a spectacular, trust-shattering correction โ the kind of event that generates genuine FUD. His instinct is sound. If any corner of the market is positioned to deliver a 2026 shock, it is the one where the gap between story and settlement is widest.
But the quieter claim โ that this correction will be a mere rung on the ladder, a fear to be overcome before the real expansion begins โ is where I part ways. A trust crisis in AI-plus-crypto would not be a psychological hurdle to clear. It would be a repricing of the belief that blockchains have a native role in the AI stack. Those are different things. The 2017 block size war resolved into a technical compromise and a fork; the market could move on because the underlying asset, Bitcoin, was untouched by the dispute. A 2026 AI reckoning would not leave the underlying narrative intact. It would damage the narrative itself. You cannot climb past a rung that has been sawed through.
The Ladder as an Instrument
Now consider the ladder not as a prediction but as a tool. Hayes is not merely observing that each cycle has had a defining fear. He is pre-defining the fear of the next cycle, in public, months or years in advance. This is not a neutral act. It is the construction of an interpretive frame, and frames are self-fulfilling in markets in a way they are not in physics.
The mechanism is straightforward once you see it. If a sufficiently influential voice declares that the market must overcome its fear of AI before the bull can begin, then every subsequent wobble in the AI-plus-crypto segment is instantly legible as the predicted fear, and every recovery is legible as the market overcoming it. The frame does not need to be correct to be useful. It needs only to be memorable. And "2017 block size, 2020 pandemic, 2023 FTX, 2026 AI" is memorable the way a rhyme is memorable โ it compresses a complex, chaotic history into four syllables that fit on a slide.
This is where the framework is most vulnerable, and also most revealing. The historical analogies are all cases where the fear was, in fact, overcome โ because we only remember the fears that were overcome. The worries that were not overcome did not become rungs on a ladder; they became the tops of cycles nobody wants to remember. There were years when the market's defining fear was not a temporary hurdle but a terminal condition, and in those years the ladder had no next rung, because the climb was over. The framework is built on a sample pre-filtered for survivorship, and it cannot see the cases that would falsify it, because those cases are, by construction, absent from the record.
I want to be fair here, because there is a lazy version of the survivorship critique โ the version that says "past performance does not guarantee future results" and stops. That is not my objection. My objection is more specific. The ladder's persuasive power comes from the fact that its four rungs are all real events that preceded real expansions. But the base rate of "anxiety preceding expansion" is uninformative, because anxiety precedes everything. The question is not whether there was a fear before each bull market. The question is whether there was a fear before each bull market that the market was forced to overcome โ or whether the fear was simply present, and the bull market happened anyway, for reasons that had nothing to do with overcoming it.
Those two stories feel identical in hindsight and are completely different in mechanism. The first implies that fear is a gate โ that you must clear it to proceed. The second implies that fear is weather โ present or absent, mostly irrelevant to the terrain underneath. If Hayes is right, the AI FUD is a gate. If I am right, it is weather. The entire positioning implication turns on which of us has read the mechanism correctly.
Who Actually Benefits, and the RWA Fiction
A liquidity-driven bull market is not distributed evenly, and the industry-transmission map is worth drawing precisely because it exposes where the narrative and the cash flows diverge.
The clearest beneficiary of a genuine liquidity expansion is the trading layer โ exchanges, market makers, the venues that monetize volume. When risk appetite returns, spot and derivatives volumes rise together, and fee revenue follows mechanically. This is also the layer where Hayes' own history sits, worth remembering when reading his bullishness as analysis rather than as position. The second beneficiary is DeFi, where looser conditions lift total value locked and revive yield activity, though the revival is often in emissions rather than durable revenue โ a distinction I have been tracking since the 2020 backtest that first taught me to distrust headline yields.
Further down the chain sit the categories that require not just liquidity but belief: the RWA complex, the gaming-and-NFT world, and the AI-plus-crypto segment. Here the transmission is weakest and the narrative does the most work. Take RWA โ the tokenization of real-world assets โ which has been a three-year storytelling exercise. The pitch is that traditional institutions will bring their balance sheets on-chain. The reality is that the institutions with the largest balance sheets have the least need for a public chain: they can settle on permissioned ledgers, with known counterparties, under existing legal frameworks, at a fraction of the compliance cost. The public chain is a solution in search of a problem the incumbents do not have. Liquidity does not change that calculus; it only makes the storytelling cheaper to sustain.
This is the pattern I keep returning to. The sectors that benefit most cleanly from liquidity are the ones that already have revenue, and the sectors that benefit most loudly from narrative are the ones that do not. In a genuine expansion, both rise. In the aftermath, only one is still standing. The Hayes framework does not distinguish between the two, because its unit of analysis is the market as a whole โ and the market as a whole is exactly the abstraction that hides the difference between the exchanges that will compound and the tokens that will evaporate.
The Discount You Should Apply, and the Trap You Should Avoid
There is a fact about the speaker that the framework does not advertise, and it belongs in any honest reading of the thesis. Arthur Hayes co-founded BitMEX, and in 2022 he pleaded guilty to violating the U.S. Bank Secrecy Act, an anti-money-laundering statute, and was sentenced to probation and a fine. I raise this not to dismiss his analysis but because the credibility of a macro thesis is inseparable from the incentives of the person delivering it, and incentives are inputs, not footnotes.
A man who built a derivatives venue and still holds crypto assets is structurally long the tide he is describing. His bullishness may be sincere โ I suspect it largely is โ but sincerity and bias are not opposites. A sincere bull is still a bull. The correct move is not to discard the thesis but to apply an independence discount to it: to read the liquidity argument on its merits while recognizing that the speaker has a commercial and reputational stake in the market's direction. Code is law, but humans write the loopholes โ and humans also write the forecasts that flatter their own positions.
Now the trap. The most seductive feature of the worry-ladder is that it immunizes the reader against bad news. Once you accept that every cycle begins with a fear to be overcome, you have a ready-made interpretation for every negative signal you encounter. A protocol depegs? Early-cycle fear. A sector collapses? The rung we were told to expect. A regulatory crackdown? The obstacle that precedes the climb. The framework does not merely predict a bull market; it reclassifies all evidence against the bull market as evidence for it. That is the definition of an unfalsifiable belief system, and it is the single most dangerous cognitive artifact in a bear market, because it converts risk management into heresy.
I have made this mistake in miniature, and it cost me. In 2020, during the DeFi summer, I backtested Ethereum's early liquidity pools against T-bill yields and built a model showing that the eye-catching staking yields were largely an artifact of token emissions rather than genuine protocol revenue. My advisor wanted a standard market overview. I delayed the draft by three weeks to stress-test the yields, because I wanted the model robust โ and what the stress test revealed was that I had, for a while, been treating the yields as early rather than unsustainable. The two labels feel similar. They lead to opposite positions. Tracing the silent hemorrhage of algorithmic trust is only useful if you are willing to call the hemorrhage a hemorrhage, and not a phase.
Where This Leaves the Reader
So where does that leave you, standing between a framework that cannot be wrong and a market that can be brutal?
I would not discard the liquidity thesis. I would strip it of its narrative and reduce it to the observable variables it actually rests on. The Fed's rate path, the trajectory of global M2, the net supply of stablecoins, the funding rates on perpetual swaps, the ratio of open interest to spot volume โ these are the things that will tell you whether the tide is turning, and none of them requires you to believe a story about ladders or fears. Track the ghost, not the sermon.
And when the AI-plus-crypto reckoning arrives โ if it arrives โ do not assume it is a rung. Ask instead whether the segment's demand has caught up to its story. If it has, the fear is weather and you can walk through it. If it has not, the fear is a gate, and a gate that has not been opened is not an obstacle to climb past. The ledger does not sleep, it only waits โ and it is waiting to tell you which one this is. The only question is whether you will still be solvent enough to read it.