cryptopath on Nostr: The Credit Factory For a thousand dollars, you can now become a bank — manufacture ...

The Credit Factory

For a thousand dollars, you can now become a bank — manufacture credit for a data center from your phone. It feels brand new. A Victorian economist already explained exactly what you're doing, and why crypto was always going to end up here.

A writer I read just deposited a thousand dollars into an onchain protocol and, in his words, became a data-center financier. The money is lent out against GPUs — the chips that run AI — and he earns a yield for supplying the credit. From his couch. For a thousand dollars. He's only half joking when he says he's now a bank.

He's not joking at all, actually, and a nineteenth-century economist named Henry Dunning Macleod explained why more than a century ago. A bank, Macleod wrote in 1891, is not an office for borrowing and lending money — it is a Manufactory of Credit. A factory. Credit isn't something a bank has and passes along. It's something a bank manufactures and sells.

Hold that image — the factory — because crypto has now rebuilt it. Not once. Three times, in order. And tracing the three rebuilds tells you something uncomfortable about where this whole thing is going.

The three ways to run the factory

There are, roughly, three ways a bank makes money. Crypto has reconstructed every one.

One: clip coupons — the narrow bank. Take deposits, don't lend them, just park them in risk-free government debt and keep the spread. Traditional banking barely does this anymore. Crypto does it at enormous scale: the stablecoin issuers. Circle holds essentially all of its tens of billions in cash and Treasuries. Tether holds most of its ~$189 billion the same way — enough that, as an earlier essay noted, it's become one of the largest holders of U.S. government debt on earth. That's a narrow bank, reborn as a token. Deposits in, T-bills behind them, spread collected.

Two: originate and hold — the lending bank. Take deposits, make loans, keep them

on your books, earn the difference — and swallow the duration mismatch, because your deposits can leave tomorrow while your loans mature in years. That mismatch is what makes banks vulnerable to runs. It's the scene from It's a Wonderful Life: George Bailey begging his panicking depositors to understand that the money isn't sitting in the vault — it's in Joe's house, and the one down the street. Crypto's version was the CeFi lender — Celsius, dangling 18.5% yields, its CEO fond of saying that either the banks were lying or Celsius was. The banks weren't lying. Borrowing short and lending long is a genuinely hard business, Celsius blew up exactly as George Bailey would have predicted, and its founder is now serving twelve years. The lending bank, rebuilt — and detonated — on schedule.

Three: originate and distribute — the securitization machine. Make the loan, then

sell it to investors. You pocket fees, and — more importantly — you free your balance sheet to make the next loan. This is the model behind your mortgage: your bank almost certainly sold it, where it got pooled with thousands of others and sold on as a security. It is also, not incidentally, the machine at the center of 2008. And crypto now has it too: a new class of onchain protocols take deposits, originate loans against real-world collateral like GPUs, and sell those loans on an exchange. Credit, manufactured expressly for resale — and assembled faster onchain, in weeks, than the offchain version manages in quarters.

Narrow bank. Lending bank. Securitization machine. Macleod's factory, rebuilt three times, in the exact order history built it the first time.

The thing built to escape the bank became a faster bank

Here's the uncomfortable through-line, and it's the real essay.

Bitcoin was built to be money without banks. A bearer asset, no counterparty, no manufacturer, no one to run from — not your keys, not your coins. The whole point was to make the factory unnecessary.

And within about fifteen years, the ecosystem that grew up around it rebuilt the factory anyway — the narrow bank, then the lending bank, then the 2008-flavored securitization machine that packages loans and sells them to strangers. Crypto didn't escape banking. It speedran it — compressing two centuries of banking evolution into a decade and a half, arriving right back at originate-and-distribute, the most fragile model of all.

That's Macleod having the last laugh. Credit is a manufactured product. And wherever there is demand for it, someone will build the factory — even on rails specifically designed to make factories obsolete. The demand for leverage is apparently stronger than the architecture built to resist it.

The honest fork — genuinely new, or just faster to break?

Now the fair question, because this isn't pure cynicism. Two things about the onchain factory are genuinely better than the old one. It's transparent. The collateral, the loans, the repayment waterfalls are visible onchain — not buried in a bank's books and sliced into tranches nobody could see into, which is precisely what made 2008 detonate in the dark. And it's permissionless at both ends. The thousand-dollar lender can now manufacture credit; the small borrower that institutional finance abandons below some deal size can now access it. That's Jevons for credit — a whole category of lending that the old factory simply forbade, suddenly unlocked. Real democratization: the credit factory, opened to everyone instead of the chartered few.

But two things are the same, or worse. The cascade risk is identical — originate-anddistribute is the exact model that blew up the world in 2008, and putting it on rails that settle in seconds doesn't repeal duration mismatch or contagion. It may just let the fire jump the wire faster. And the denomination is dollars. The deposits are dollar stablecoins; the loans are dollars against real-world machines. So this isn't Bitcoin banking at all. It's dollar banking, rebuilt on crypto rails — which mostly extends the dollar's reach, the same "Bitcoin wins the settlement, the dollar wins the denomination" trap from an earlier essay.

Where Bitcoin actually stands

Which brings the cleanest line in all of this. Bitcoin's entire proposition was to be the one thing in finance that is not a manufactured credit product — not a liability, not a loan, not a tranche, not anyone's promise. Macleod's factory manufactures claims. Bitcoin manufactures finality. They are opposite products.

So the honest way to see the space is that "crypto" has split into two projects wearing the same word. One is the credit factory — stablecoins, CeFi, onchain securitization which is faster banking, dollar-denominated, 2008's model with better plumbing and better lighting. The other is Bitcoin — the refusal of the factory, the bearer asset that exists precisely so you never have to trust a manufacturer again. These are not the same movement. On the deepest axis, they are opposites.

The one line to keep

Macleod said a bank isn't an office for borrowing and lending but a Manufactory of Credit — and crypto has now rebuilt all three versions of that factory: the narrow bank, the lending bank that blew up on cue, and the originate-and-distribute machine quietly assembling loans to sell. The thing built to escape the bank became a faster bank.

The genuinely new part is transparency and permissionless access — credit manufacturing opened to everyone. The uncomfortable part is that it's 2008's machine on faster rails, denominated in dollars. And the one thing in the entire space that refused to become a factory is the only asset that was never a claim on anyone in the first place.

Crypto didn't escape the bank. It rebuilt the factory — shinier, faster, open to all — and in rebuilding it, rebuilt its risks. The question for the next decade isn't whether crypto can bank. It obviously can, better than ever. It's whether, having lovingly reconstructed the entire factory, anyone still remembers why someone once wanted to build the one thing that isn't one.