Arthur Hayes Says AI IPOs Could Drain Bitcoin's Liquidity. The Bank Data Says Something Else

Arthur Hayes is right that bank balance sheets, not the Fed, decide whether Bitcoin gets a bid. He's pointing at the wrong villain though.

The Liquidity Chain

Hayes's thesis runs like this. The AI names everyone wants a piece of, SpaceX, Anthropic, OpenAI, come public. Then the lockups expire. Early employees and venture holders take the exit at the first good bid. The dilution soaks up capital that would've gone to work somewhere else, AI credit tightens, bank balance sheets shrink, and the dollar liquidity that funded crypto's August rebound drains out from under it.

Here's what matters: Hayes isn't arguing about the Fed. He's arguing about credit creation. Commercial banks make loans and the loans become deposits. That's the money. A Bank of England paper put a number on it years ago and it's still the cleanest explanation out there. Banks create the overwhelming majority of money in circulation, not central banks.

So if the AI capex machine stalls, the credit machine stalls with it. And Bitcoin, which trades like a liquidity barometer with extra caffeine, feels it first.

He's also honest about the weak link. The latest weekly dip in bank balance sheets, he concedes, is one data point. One week doesn't make a trend, and we've all watched a single Friday print get dressed up as a regime change.

The Counterpoint

Here's where I push back. An IPO doesn't destroy a dollar.

When a founder sells into a lockup expiry, cash moves from buyer to seller. Both of those people bank somewhere. The deposit stays inside the system. Same story for primary proceeds that land on the company's balance sheet, which then get parked in T-bills or a money market fund. It's a change of ownership, not a shredder.

So what actually chokes? The marginal appetite for new credit. That's a function of returns on AI capex, not a lockup calendar. If hyperscaler guidance comes in hot and credit spreads stay tight, IPO supply gets absorbed and nobody notices. If returns disappoint, the selloff starts long before the first share unlock.

What the street is missing: the bigger liquidity drains aren't sexy. Treasury issuance pulls cash out of the system every Tuesday. The reverse repo balance, stablecoin growth hoovering up T-bills, and whatever the Fed does with runoff all matter more to the dollar float than a handful of venture distributions.

And what if the AI listings are actually bullish liquidity? A hot IPO pulls sidelined cash off the sidelines and into bank deposits and brokerage balances. That's not a drain. That's redeployment.

The Verdict

I'll commit. Hayes has the mechanism right and the trigger wrong. Bank balance sheets really do drive Bitcoin's beta, and credit creation really is the variable that matters. But share unlocks aren't the lever. They're the headline.

The thing to watch is the funding market. If bank reserves keep sliding week after week while SOFR spreads widen and AI capex guidance softens, then the AI credit cycle is topping out and Hayes gets his drawdown. One weekly blip in balance sheet data doesn't get you there.

From a risk perspective, position for the credit signal, not the lockup calendar. The numbers tell the story. Lockup expiries trade around a stock. Bank reserves trade around everything.

Watch next: the first mega AI listing's pricing date, the first lockup expiry after it, and the weekly bank reserves print. Two of those three turning the wrong way together is when this thesis earns real conviction.