Money Stuff: Stablecoin Narrow Banking

If you have dollars, you can put them in a checking account at a bank. This is convenient (the bank will let you use those dollars to make payments) and fairly safe (the bank probably won’t fail, and if it does your deposits are insured, up to some limit, by the US government), but not especially lucrative. The bank will use your money to do stuff — make loans, buy securities, etc. — and will make some money from those activities, but you won’t see much of that money. The bank will lose some money (from bad loans or bad trades), and it will use a lot of the money to pay for branches and salaries and stock buybacks and stuff. It will pay you something close to 0% interest on your checking account. 

Of course, there are all sorts of ways to earn more money on your money, at the cost of convenience (your money might get locked up) and safety (you might lose some of it). You can put the money in private credit funds or Nvidia stock or Bitcoin or whatever. None of these things is really a substitute for a checking account. 

Another thing you can do is invest in Treasury bills. Short-term US Treasury bills pay about 3.6% interest. They are fairly convenient (they mature in a month or two, and you can sell them any time in a hugely liquid market) and quite safe (they mature at par in a month or two, and the US government will probably pay). They are not a perfect substitute for a checking account, though. You can’t write checks against them; you can’t instantly use your Treasury bills to pay your bills. You have to convert them to cash, put the cash in a checking account, and write checks against that.

Here is an arbitrage:

1. Start a bank.
2. Offer checking accounts.
3. Take deposits from customers.
4. Put their money directly into short-term Treasury bills paying 3.6%.
5. Don’t do anything else: No loans, no credit cards, no long-term investments, no branches, no customer service line, no free toasters, almost no salaries. (Obviously you get a salary!) Costs are low (a website and your salary), say 0.1% of the money you take in.
6. Pass the other 3.5% on to the customers.

This product is much better for customers than a regular checking account in two respects:

• You are paying them 3.5% instead of roughly 0%.
• You are safer than the other banks. The other banks probably won’t fail, and if they do their deposits are insured up to some limit, but banks do fail sometimes, and some deposits are over the limit. Your bank, though, kind of can’t fail: All your money is in short-term Treasury bills with essentially no credit or interest-rate risk. All the stuff that banks do that could cause them to fail — lending, trading, interest-rate bets — you don’t do. (Obviously the other thing that banks do that cause them to fail is “steal all the money,” which you could do, but you should try not to.) You are a “narrow bank,” taking deposits and parking them somewhere perfectly safe.

It is worse than a regular checking account in some other respects — minimal customer service, no loans, no branches, no toasters — but this is 2026 and a lot of your customers are probably fine with that. They do everything on their phones anyway; they don’t need branches.

Can you do this? Well. You’re kind of not supposed to, not exactly, not literally. For one thing, US bank leverage requirements are a problem: You need to have some capital against your holdings of Treasury bills, so you can’t really pass all of the interest back to customers; a lot of it has to go to shareholders. More broadly, US bank regulators take a dim view of this sort of thing. We have talked a few times about TNB USA Inc., “The Narrow Bank,” which aimed to do roughly this trade. (Instead of Treasury bills, it would put the money in interest-bearing deposits at the Federal Reserve, which have similar properties of being extremely safe and paying interest.) The Fed didn’t like it. “Fed Rejects Bank for Being Too Safe” was my headline.

The problem is: What about all the other banks? They are taking deposits and using them to make loans; they use depositors’ money to provide the credit that lubricates the economy. There is some cost to that lubrication, and the cost shows up in forms like “the bank pays you 0% on your deposits” and “occasionally a bank fails and is bailed out by the government.” You can get rid of the costs through narrow banking, but that might also get rid of the lubrication.

The more specific problem is: Look, in good times, people will keep their money in their local bank because they like the fancy branch and the customer service and the toasters. But in a panic, people will take their money out of their local bank and put it into a narrow bank, because it’s safer. Narrow banking arguably undermines the stability of the broader banking system; it makes the regular banks less safe.

So you mostly can’t do exactly this sort of narrow banking in a bank. Not never, though. We talked last month about a thing called N3XT, a “Wyoming Special Purpose Depository Institution” where “every dollar of deposits is backed one-to-one by cash or short-term U.S. treasuries, making N3XT the first ‘narrow bank’ in the United States.”

And outside of “banking,” you can get pretty close to this trade. The classic example is a government money-market fund. That’s just a pot of money invested in short-term US Treasury bills. There are lots of those, and they tend to pay about 3.6% interest these days: They buy Treasury bills, collect the interest and pass almost all of it on to customers. Unlike banks, they do not really have capital requirements. And your broker will probably let you write checks against them.

There is a little tension here: US bank regulators are nervous about narrow banking because they worry that it will undermine the stability of the banking system, but they allow government money market funds. And in fact there’s an argument that those funds do undermine the stability of the banking system: Back in 2023, we discussed worries that deposits were fleeing troubled US regional banks for money market funds, and that this would put the banks at risk. Still, the banks have mostly survived competition from money market funds.

Then there are stablecoins. As we have discussed, stablecoins are a form of narrow banking: They take deposits and (mostly) park the money in cash and Treasury bills. You occasionally hear worries that this will undermine the banking system: People will prefer to put their money in stablecoins (which are convenient for crypto-y sorts of payments, and invested in Treasury bills) rather than in banks (which are less crypto-native and take more credit risk).

In the US, the regulatory solution to this worry is prohibiting stablecoins from paying interest. That’s in the GENIUS Act regulating stablecoins: They can’t pay interest. They are safer [1]  and more crypto-y than bank accounts, but they are no more lucrative than checking accounts, and less lucrative than many savings accounts.

As we have discussed a few times, though, there are some pretty straightforward ways around that restriction:

• The stablecoin issuer can’t pay interest, but it can pay fees to crypto exchanges (for encouraging their customers to hold the stablecoin), and the exchanges can pass those fees along to the customers. To the customers, this looks pretty much exactly like “interest on a stablecoin.”
• If you’re a customer who holds a stablecoin, you can lend it out to crypto traders for their crypto trades, and get a second layer of interest.

And so in practice stablecoins kind of do pay interest. Does this undermine the safety and profitability of the banking system? I mean, the banks think so. The Wall Street Journal reports:

The cryptocurrency and banking industries are locked in a lobbying battle over digital tokens that yield annual payouts, a fight that threatens to derail legislation intended to bring crypto into mainstream finance.

The two sides are clashing about what crypto firms call rewards, or annual payments to investors based on a percentage of their total holdings. They are commonly used for stablecoins, popular tokens typically pegged to the U.S. dollar and used for trading, overseas payments and money transfers.

To banks, rewards on stablecoins from companies such as Coinbase Global that pay out 3.5% resemble high-yielding deposits—but without the regulations they face for holding customers’ cash. Bank-industry groups have flooded lawmakers with letters and phone calls arguing the rewards would decimate Main Street lenders. The national average interest rate for a standard interest-bearing checking account is below 0.1%.

“We are hearing every day from community bankers who are worried about the impact stablecoins offering yield will have on their deposit bases and their ability to lend and support their local communities,” said Brooke Ybarra, senior vice president for innovation and strategy at the American Bankers Association, an industry group.

In  theory, one solution here is to allow stablecoins to pay interest (like banks) but also impose capital requirements (like banks). I would not bet on that happening though.

Here is a rough stylized story of covenants. In the olden days, companies had two ways to borrow money: They could issue bonds to public investors, or they could get loans from their banks. In a bond offering, the company would send out a disclosure document (a prospectus or offering memorandum) to dozens of potential buyers — insurance companies, mutual funds, hedge funds, individuals [2]  — and would take money from whoever wanted to invest at some market-clearing terms. In a bank loan, the company would sit down with its relationship banker, who had known the company for decades, and would negotiate a loan based on the banker’s deep knowledge of the company and trust in the character of its executives.

Because bonds were impersonal capital markets transactions, while bank loans were products of long-term personal relationships, they had different terms. Among the differences: Loans traditionally had “maintenance covenants.” A maintenance covenant says something like: “The company’s debt can’t exceed six times its earnings,” or “the company’s interest expense can’t exceed half its earnings.” [3]  The company promises to have at least a certain amount of earnings, relative to its debt. What if it doesn’t? What if it has a bad year and doesn’t make as much money as it promised? Well, then it is in default, and the lender can demand its money back immediately. Even if the company is current on its loan payments, if it breaches a covenant, then it is in default and has to pay back its loans.

This seems harsh, and you could imagine the company objecting. “Look, if we make enough money to pay your interest, and we do pay your interest, what do you care if we make twice as much?” The bank could reasonably reply: “If you are only barely making enough to cover our interest this quarter, that makes your loan much riskier than we thought, and we want the ability to get our money back. We are not in the business of taking wild risks; we want a safe credit where you make plenty of money, and if things change we want out.” 

But that was mostly not exactly what the banks said. Instead, the standard answer went more like this: “We understand that this seems harsh. Obviously, if you have a bad year and don’t make a lot of money, it would be terrible for you if we also demanded that you repay our loan early. It would be monstrous of us to do that. But we are not monsters. We have a long personal relationship with you; you know and trust us. No, the point of the maintenance covenant is not that we’ll take our money back at the first sign of trouble. The point of the maintenance covenant is that we want to work with you if there’s trouble. We want you to call us so we can work something out. Maybe we’ll extend the term of the loan. Maybe we’ll ask for more collateral. Maybe we’ll help you find an outside investor. We’re a bank; we can do lots of things. But in that scenario, where your business isn’t going as well as we both expect, we want an early warning, and we want leverage. We want the ability to call in our loan, so that we have a seat at the table while you work things out. But we won’t use that ability, or not immediately and arbitrarily. The covenant is just a part of our relationship.”

That answer broadly made sense, and maintenance covenants in bank loans were pretty common. But this whole line of reasoning doesn’t work for bonds. A bond is not like a bank loan, where the lender is a single bank with a long relationship with the company. In a bond deal, there are lots of lenders, they change every day as people buy and sell the bonds, and they don’t necessarily have any special relationship with the company’s managers. So bonds do not normally have maintenance covenants; bonds do not normally say “if the company’s income falls below X it’s a default and the bondholders get their money back immediately,” because that would be a disaster. [4]  It would be too hard for the company to negotiate with all the bondholders if that happened, so they probably would demand their money back, and the company might go bankrupt.

That is the old-timey story. One thing that happened over the last few decades is that “bank loans,” and particularly “leveraged loans” to non-investment-grade companies, became much more like bonds. In fact the name “bank loan” is now somewhat antiquated, and people are more likely to call them “broadly syndicated loans.” As the name implies, they are broadly syndicated: A company’s lead bank will arrange the loan, but the money will come from lots of lenders. The lenders will include some banks, but also hedge funds and credit funds and collateralized loan obligations. The loans will trade in a secondary market; the company’s lender base can change each day.

And so the old arguments for maintenance covenants — that they were part of a relationship, and that the lender would work with the company in case a covenant was breached — got much weaker. “If our income goes down, we don’t want to be in default,” companies could reasonably say, “because our collection of random hedge funds and CLO managers won’t work with us and we’ll just go into bankruptcy.”

And lenders were largely like “yeah I see your point,” and there was a long rise in “covenant-lite” loans, that is, loans without maintenance covenants. The old-timey relationship theory of covenants no longer made sense, so they dropped out of a lot of loans.

The big story in credit in recent years is the rise of private credit. Private credit is in many ways like the old model of bank lending: A company negotiates one loan with one lender, the lender puts up the money and holds the loan to maturity, and the company has a relationship with the lender. If things go wrong, the lender will work with the company to fix them; their interests are aligned and they have a long-term relationship. [5]  The difference is that the lender is not a bank, but it is not a rapacious hedge fund either; it’s a long-term asset manager that understands credit markets and invests in relationships with its borrowers. Of course the company should be happy to work with its private credit lender if it runs into trouble, and of course the private credit lenders — who hold loans to maturity and take concentrated risks on individual borrowers — should want covenants to protect themselves. And private credit loans usually have maintenance covenants.

And then the inevitable continuation of that story is:

1. Private credit loans get bigger and more distributed.
2. They start trading.
3. Everything starts to feel a bit more impersonal.
4. Covenant-lite private credit.

We talked about that story back in 2023, and it’s continuing. Bloomberg’s Francesca Veronesi and Kat Hidalgo report:

Private credit firms’ efforts to reap leveraged debt business from Wall Street is coming at a steep cost — safeguards that made them less vulnerable to an economic downturn. …

The safeguards, along with access to nonpublic financial information, are part of why industry chiefs like to claim private is safer than bank credit. They also say that they vet deals more carefully. While the rates at which private borrowers default is up for debate, they’ve been reported in a range of 2% to 3% — lower than for leveraged loans made by banks.

Traditional private credit carries maintenance covenants — limits on leverage that are tested regularly. As soon as a borrower breaches its limits, a lender can seek an equity injection from shareholders or private equity owners, push them into asset sales for the good of lenders, or demand more collateral. If a lender favors a debt restructuring, or even wants to take over a company, a covenant breach would be the starting point.

The right to curb unhealthy debt levels is also a way for private credit firms to hedge risks like recession or tariffs that can zero corporate profits.

But now direct lenders are making deals on the same terms as banks — breaking long-standing convention that gave them more rights to manage a borrower’s debt.

In the long run, private credit will probably end up like every other kind of credit.

Back in simpler times, a year ago, BitMine Immersion Technologies Inc., as its name implies, was a technology company that mined Bitcoin using immersion. Something like that. From its annual report in December 2024: 

Since July 2021, our business has been as a blockchain technology company that is building out industrial scale digital asset mining, equipment sales and hosting operations. The Company’s primary business is self-mining bitcoin for its own account, as well as hosting third-party equipment used in mining of digital asset coins and tokens, specifically bitcoin. ...

We plan to operate our data centers using immersion cooling technology. Immersion cooling is the process of submerging computer components (or full servers) in a thermally, but not electrically, conductive liquid (dielectric coolant) allowing higher heat transfer performance than air and many other benefits. 

Intuitive! But 2025 was a pretty weird year, and if you ran a public company that was at all connected to crypto, it was awfully tempting to become a digital asset treasury company. As we discussed a lot around here, there was a period of at least several months where a pot with $100 worth of crypto would trade at $200 on the stock exchange, so everyone rushed to do that. That bubble has now pretty definitively burst: Many digital asset treasury companies are trading below net asset value, and now a pot of $100 worth of crypto is more likely to trade at $90 than $200.

Like everyone else, BitMine did the pivot: “During 2025, the Company refined its business strategy to emphasize digital asset treasury operations, reflecting a transition from primarily mining and hosting activities toward the long-term accumulation and optimization of digital asset holdings.” Never mind about the immersion. Also it transitioned from Bitcoin to Ethereum, “becoming the largest corporate holder of ETH, with over 3,737,140 tokens valued at approximately [$10.5 billion] as of November 30, 2025.” “The Alchemy of 5%” is its slogan, apparently meaning that, if it ends up  holding at least 5% of Ethereum, then, uh, that would be good. In 2025, that was a strategy!

Like everyone else, BitMine might be having second thoughts. By this Monday, its holdings were up to $14 billion of assets, consisting mostly of “4,167,768 ETH at $3,119 per ETH,” or about $13 billion of Ethereum, but also “193 Bitcoin (BTC), $23 million stake in Eightco Holdings … and total cash of $988 million.” BitMine has essentially no debt, so that’s a net asset value of about $14 billion. Bloomberg tells me that BitMine’s market capitalization on Monday was about $13.3 billion. [6] Ah well.

What do you do, if you pivoted to the crypto treasury strategy and it stopped working? We have talked about one approach, which is to run the strategy in reverse, selling crypto to buy back stock. BitMine has another idea:

LAS VEGAS, Jan. 15, 2026 /PRNewswire/ -- (NYSE AMERICAN: BMNR) Bitmine Immersion Technologies, Inc. ("Bitmine" or the "Company") the leading Ethereum treasury company in the world, announced a $200 million equity investment into Beast Industries. Bitmine also implements an innovative digital asset strategy for institutional investors and public market participants.

"MrBeast and Beast Industries, in our view, is the leading content creator of our generation, with a reach and engagement unmatched with GenZ, GenAlpha and Millennials," said Thomas 'Tom' Lee, Chairman of Bitmine. "Beast Industries is the largest and most innovative creator based platform in the world and our corporate and personal values are strongly aligned."

I suppose part of the point here is, like, get a YouTuber to get the kids more interested in Ethereum? Like here are some words:

“It’s our view that ethereum, which is a smart contract platform, is the future of finance, where digitalization of not only dollars but stocks and equities [are] going to take place,” Bitmine Chairman Tom Lee said on CNBC’s “Squawk Box” on Thursday. “Over time, that really blurs what is a service versus what’s digital money, and that’s where a collaboration and investment into Beast Industries makes sense.” ...

“I think it’s part of the sort of evolution of digital platforms and money, and I think it’s really uniting the No. 1 creator in the world with the biggest ethereum platform in the world,” Lee added.

There are other advantages. There is some controversy about whether digital asset treasury companies are “investment funds” or operating businesses. [7]  This controversy strikes me as dumb: They are obviously investment funds; their business is mainly investing in crypto, and their value comes from the value of that crypto. But they are a weird passive sort of investment company, not “our hard-working team of professionals makes brilliant investing decisions” but rather “hey here’s a pot of Ethereum.” 

They don’t have to be that way, though. There was a time when “hey here’s a pot of Ethereum” was a really good pitch to investors, so lots of crypto companies pivoted their strategy to accumulating as much as they could of their favorite token. Now that pitch has stopped working. But all those companies pivoted to becoming investment funds; now they have huge pots of cash that they invest in stuff. Mostly Ethereum, in BitMine’s case, but … you can change that easily enough? You’ve got a $14 billion pot of money and a free hand to invest it; investing it all in Ethereum used to be cool, but now it is less cool. Maybe do some venture capital too.

You know the deal: Every bad thing that a public company does is also securities fraud, and “bad” is in the eye of the beholder. For instance, one thing that has happened in financial markets in the last few years is that a lot of big technology companies have committed to spend a zillion dollars building data centers for the artificial intelligence boom. Is that bad? Well, the companies don’t think so. Their shareholders, by and large, don’t think so — there is a lot of enthusiasm for the AI boom and the profits that it might one day create — though there are dissenters.

On the other hand, there are some investors who might be less happy. If you were a creditor of a big tech company a few years ago, you probably felt great. “This company makes tons of money, doesn’t have to spend that much, and has very little debt,” you thought; “that’s just how Big Tech works.” Now, the way Big Tech works is that companies borrow money and sign leases obligating them to pay tens of billions of dollars a year for many years to build data centers so they can compete in AI. Having zillions of dollars of lease and interest obligations is a worse credit profile than not having that. There are ways to mitigate this risk — companies can do some sleight of hand to make the data-center debt not quite count as debt — but this can only accomplish so much.

So if you are a bondholder of a big tech company, you might think “the AI boom is bad, so it is securities fraud.” (I mean: If you were a bondholder before the AI boom, or, at least, before the company’s latest round of debt financing or data-center commitments. Lots of creditors love the AI boom, in the sense that it allows them to buy new bonds at high interest rates. But if you were a pre-boom bondholder who bought old bonds at low interest rates, the boom is bad for the price of those bonds.)

Bloomberg’s Caleb Mutua reports:

Oracle Corp. was sued by bondholders who claim that the database giant failed to disclose plans to raise more debt when it borrowed $18 billion in one of 2025’s largest corporate bond offerings.

The Ohio Carpenters’ Pension Plan, which was among bondholders that bought bonds issued in September, claims that Oracle didn’t tell investors that it needed to raise a “significant amount of additional debt” to finance its artificial intelligence infrastructure, according to a lawsuit filed in a New York state court Wednesday.

Several weeks after issuing the notes, Bloomberg reported that banks were also providing a $38 billion debt offering to help fund data centers in Texas and Wisconsin tied to Oracle. As a result of the additional debt, Oracle’s bonds began to trade with yields and spreads similar to lower-rated issuers as concerns about Oracle’s credit risk grew. 

“The offering documents were false and misleading and omitted to state that, at the time of the offering, Oracle was organizing to raise that additional debt, which would ultimately bring the creditworthiness of these bonds into question,” according to the lawsuit.

Here is the complaint. These are not exactly pre-AI-boom bondholders — they bought these bonds in September 2025 — but since then “reports emerged that Oracle was looking to raise an additional $38 billion in debt sales to help fund its AI buildout” and these bonds traded down. The bond offering documents did say things like “Oracle Corporation and its subsidiaries may incur additional indebtedness in the future,” but they didn’t say that it would incur the additional debt, so, lawsuit.

I suggested above that all of this makes a certain amount of sense as a matter of disgruntled Big Tech bondholders, but that Big Tech shareholders might have less cause to sue about the AI boom. But that’s only loosely true. Incurring zillions of dollars of debt to build data centers feels (1) fairly straightforwardly bad for credit but (2) ambiguous for equity. But of course “ambiguous for equity” doesn’t mean that people can’t sue. Every bad thing that a public company does is also securities fraud, and “bad” is in the eye of the beholder. If the stock went down, sure, securities fraud. And Oracle’s stock is down since September. Here’s a law firm press release:

LOS ANGELES, Jan. 15, 2026 (GLOBE NEWSWIRE) -- The Portnoy Law Firm advises Oracle Corporation, (“Oracle" or the "Company") (NYSE:ORCL) investors that the firm has initiated an investigation into possible securities fraud, and may file a class action on behalf of investors. ...

On September 10, 2025, Oracle and OpenAI OpCo, LLC (“OpenAI”) announced a $300 billion, five-year cloud computing contract to supply OpenAI with computing power.  On November 13, 2025, reports emerged that Oracle was seeking to raise an additional $38 billion in debt sales to help fund its AI buildout, with loan proceeds to fund two data centers that would support the Oracle-OpenAI agreement.  On this news, Oracle’s stock price fell $9.42 per share, or 4.15%, to close at $217.57 per share on November 13, 2025.  Then, on a December 10, 2025 earnings call, Oracle’s Executive Vice President and Principal Financial Officer disclosed that the Company “now expect[s] fiscal 2026 CapEx will be about $15 billion higher than we forecasted after Q1.”  On this news, Oracle’s stock price fell $24.16 per share, or 10.83%, to close at $198.85 per share on December 11, 2025.

Yep, building AI data centers, securities fraud.

Amazon Blasts Saks Funding Deal, Says Equity Is ‘Worthless.’ Inside the Mad Dash to Save Saks, America’s Last Luxury Retailer. Goldman’s Stock Traders Beat Their Own Wall Street Record. BlackRock Total Assets Hit Record $14 Trillion as ETFs Surge. Citi CEO Warns of More Job Cuts, Calls Time on ‘ Old, Bad Habits.’ Kalshi and Polymarket race to crack ‘parlays’ as stakes rise in sports betting. Hedge Funds Revive Dispute Over Argentina GDP-Linked Securities. Blackstone Real-Estate Fund Stages Comeback With Best Return in Three Years. Donald Trump’s ‘unpredictable’ policies to fuel multiyear shift from US, Pimco says. Blackstone to Offer Private Investments in Empower 401(k)s. BlackRock, Microsoft AI Partnership Raises $12.5 Billion So Far. OpenAI Forges Multibillion-Dollar Computing Partnership With Cerebras. Hedge Funds Hire Talent Scouts in Japan to Tap Comeback. Key Wall Street Regulator Grapples With Crypto, Sports Gambling. Amazon Becomes First Buyer of Arizona Copper Made by Rio’s Nuton. Mastercard, Visa, Revolut Lose London Suit Over Card Fee Cap. Soho House Shares Rise After New Funding Secured For Take-Private Deal. People Are Paying $99 a Month to Talk to a Tony Robbins Chatbot. India tells delivery companies to stop promising 10-minute service. Elon Musk bows to pressure over Grok creating sexualised AI images. Meta Lays Off 1,500 People in Metaverse Division.

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!