Arthur Hayes’ recent take on AI, liquidity, and Bitcoin has been widely simplified into a single directional call: after the AI bubble bursts, central banks will step in to rescue the situation, and Bitcoin will benefit. But if you go back to the original logic of D is for Dilution, the framework isn’t really about being bullish on BTC. Its core is the clear sequence between two judgments: first short AI’s large IPOs as they drain dollar liquidity, then, after a credit shock forces policymakers to pivot, go long the liquidity that the rescue re-prices.
Right now, the market is only pricing the first half of the chain.
What Exactly Is Hayes "Bearish" On?
Hayes’ core call isn’t that "AI will collapse," and it isn’t that "BTC will rise immediately." What he breaks down is the transmission path from a large AI company’s IPO to equity supply, credit conditions, and dollar liquidity. SpaceX, Anthropic, and a potential OpenAI IPO will convert a large amount of previously locked-up equity into tradable supply. At the same time, they will shift the center of gravity for AI financing from private markets to the public bond market.
Supply pressure drags down stock prices. Weak stocks raise bond yields. Higher financing costs then feed back by shrinking the room for AI-related credit expansion. Banks’ deep involvement in data-center and GPU capital expenditures means that when credit slows, dollar-liquidity expansion cools. In this phase, crypto—being a high-beta asset—first absorbs the pressure of outflows.
In other words, what Hayes is doing at the front end of the chain is a "liquidity contraction trade," not a denial of Bitcoin’s long-term value. The rescue scenario sits at the back end. Between the two, there’s a credit event that has to be validated.
How Much Money Will the AI Giants’ IPOs Actually Pull?
Just SpaceX alone: its IPO plan is to raise about $75 billion, with a target valuation near $1.8 trillion (fully diluted basis). Anthropic plans to push ahead with its listing before Thanksgiving, targeting $80 billion to $100 billion in fundraising, with an implied valuation range of $1.25 trillion to $1.75 trillion. OpenAI hasn’t filed for a public listing yet, but its latest round of private financing has already shrunk the first tranche to about $11 billion, implying a $1.25 trillion valuation—below earlier plans.
The bigger pressure comes from the unlock schedule after listing. Hayes estimates that by June 2027, SpaceX could add about 8.6 billion shares available for trading, roughly expanding current share count by about 50%. Anthropic’s early public investors could face up to 83% dilution in the same period. These numbers don’t necessarily mean all supply hits at the same time. Still, they point in one direction: over the next 6 to 12 months, the AI sector will shift from a "capital sink" into an "equity supply source."
For crypto markets, which also rely heavily on risk-on capital, this creates direct competition for funds.
Why Does Crypto Always Face AI Credit Shock First?
In a tightening marginal-liquidity regime, Bitcoin behaves more like a high-beta risk asset than an immediate safe-haven tool. Hayes has repeatedly noted that Bitcoin tends to move in the same direction as dollar liquidity expansion. When newly added dollars primarily flow to AI-related infrastructure and the credit chain, the incremental capital crypto can capture is structurally lower.
This squeeze isn’t just a narrative. From 2022 to 2026, roughly $1.5 trillion in AI-related debt was issued to fund data centers, GPU purchases, and compute leasing. The holders of that debt overlap heavily with marginal risk capital. When AI credit conditions tighten, liquidity contraction doesn’t stay confined to AI stocks—it propagates via shrinking risk budgets across the whole high-beta asset complex.
Right now, BTC has been relatively weak in the Gate market range of $74,000 to $75,050, consistent with the front-end characteristics of that transmission path. It’s not that crypto has introduced a structural negative catalyst on its own. It’s that marginal liquidity is being absorbed by a more certain AI capex narrative.
Is CoreWeave’s 16% Yield High Enough to Meet the "Central Bank Rescue" Threshold?
CoreWeave’s credit metrics provide an early window in step four of the chain, but they aren’t the rescue trigger by themselves. Its 2032 maturity bond with a 9.625% coupon has already pushed through the 16% yield mark. Five-year credit default swaps briefly surged to 1,289 basis points, above the October high. Nvidia’s CDS is up 76% over the past month; Broadcom’s is up 49%; and Oracle’s CDS spread is around 298 basis points.
These signals suggest credit markets are repricing leverage expansion for AI infrastructure. But there’s still a gap between "spreads widening" and a "systemic credit event." What truly forces policymakers to shift from routine liquidity management to a rescue isn’t whether one issuer’s yields are high or low. It’s whether financing disruption begins threatening bank balance sheets and triggers a cross-market collateral haircut spiral.
Recently, usage of the Fed’s standing repo facility has risen to around $26 billion, the second-highest level since 2020. Meanwhile, the reverse repo balance has fallen from $2.4 trillion in 2022 to very low levels. This is evidence that the liquidity buffer has thinned. But the policy implication is still "keep money markets functioning," not "rescue AI credit."
How Did AI Debt Issuance Dodge Bitcoin?
Path dependence in liquidity transmission matters more than total number. The funding raised by AI-related debt first flows into data centers, GPU clusters, power contracts, and compute leasing agreements. In accounting and credit terms, that creates financable assets—not tradable risk positions. Morgan Stanley’s forecast for global AI-related debt issuance in 2026 is about $570 billion. UBS’s estimate for the same period is close to $990 billion.
No matter which estimate you use, the first destination of incremental capital isn’t Bitcoin. The "bypass" Hayes describes is fundamentally a use-of-funds question. When new credit gets locked into the capex cycle for compute infrastructure, crypto can only wait for this capex cycle to slow and credit conditions to loosen again—then potentially re-enter the marginal allocation view.
That’s why "AI bond issuance is huge" and "BTC isn’t going up" can both be true. Liquidity expansion in the total aggregate doesn’t automatically translate into liquidity expansion for crypto.
If a Rescue Ultimately Comes, What Determines the Transmission Delay?
Hayes describes the rescue scenario as "glorious for Bitcoin," but he also adds an easily overlooked qualifier: "Be patient." The post-rescue liquidity release could push BTC toward a million dollars or higher in the long-term scenario—if the size of the credit shock is large enough to change policymakers’ reaction function, not just trigger a one-off spread fluctuation.
The current dollar-liquidity structure offers clues about timing. Bank reserves are around $2.96 trillion. The Treasury General Account (TGA) balance sits at elevated levels—roughly $900 billion to over $1 trillion. The overnight reverse repo buffer is basically depleted. SOFR has been set about 5 basis points above IORB. That implies the system’s cushion against shocks is thinner than in 2020. Still, the policy pivot requires a clearly identifiable credit event to trigger the change, not just a natural drift in liquidity indicators.
From step three to step four, the decisive variable isn’t time itself. It’s whether AI credit contraction escalates from "higher financing costs" to "credit channels shutting."
Which Verifiable Signals Should You Track Now?
Instead of trying to guess when a rescue will arrive, break Hayes’ framework into observable nodes.
The first set of signals comes from equity supply. SpaceX’s lock-up arrangements, the early unlock schedule after Anthropic’s listing, and whether OpenAI enters the public markets in 2027 all determine the duration and intensity of the first step.
The second set comes from AI credit pricing. Whether CoreWeave’s CDS continues moving from 1,289 basis points toward a systemic threshold. Whether Nvidia’s and Oracle’s CDS spreads accelerate wider from their lows. And whether AI-related bond issuance starts getting delayed or failing—these are direct reads on how close step four is.
The third set comes from the remaining room in dollar-liquidity buffers. Bank reserves, TGA balances, and the premium of repo rates versus IORB together determine how many ordinary tools policymakers still have in the face of a credit shock, rather than being forced into a rescue mode.
Summary
Arthur Hayes’ six-step chain provides a timing structure, not a single directional call. Equity dilution and tighter credit conditions from major AI IPOs are putting front-end pressure on crypto through the dollar-liquidity channel. The liquidity release driven by central bank rescue sits at the back end, where it needs to be validated by a credit event. With BTC trading in the $74,000 to $75,050 range in the Gate market, the picture reflects front-end pricing, not confirmation of the end scenario. Monitoring the unlock schedule, AI credit spreads, and dollar-liquidity buffers fits the logic of the framework better than trying to front-run a "rescue trade."
FAQ
Does Hayes’ framework equal "bullish on Bitcoin"?
No. The first three steps point to liquidity contraction and crypto taking pressure. The last three steps are where rescue and BTC’s liquidity benefit come into play. Both directions share the same underlying logic chain, but they map to different time windows and different trigger conditions.
If CoreWeave’s CDS breaks above 1,200 basis points, does it mean a central bank rescue is coming soon?
Widening credit spreads are an early signal, but the rescue threshold is far higher than a single issuer’s credit deterioration. You need to watch whether financing disruptions begin to propagate into bank balance sheets and broader collateral markets.
Why is AI bond issuance so big, but BTC doesn’t benefit in sync?
The use of incremental capital determines the transmission path. AI-related debt first funds data centers, GPUs, and power infrastructure. That creates compute assets rather than tradable risk positions. In this phase, crypto is on the weaker end of the capital competition.
What does "Be patient" specifically mean in Hayes’ context?
It means not pricing the rescue scenario ahead of time until the credit shock is validated and policymakers’ reaction function actually shifts. It’s a description of timing, not doubt about direction.
