$135 Billion in T-bills, No Banking License

I read a list yesterday.

The largest holders of US Treasury bonds. Japan at the top. The UK at number two. China at number three, after years of selling.

I scanned down. Around number seventeen, something stopped me.

Tether. The crypto stablecoin company. $135 billion in US T-bills as of last fall.

That puts Tether above South Korea on the list. It holds more T-bills than Saudi Arabia, and roughly the same amount as the central bank of Germany.

If you've never held a stablecoin, here's the basic idea. It's a digital token on a blockchain that's supposed to trade at one dollar. You send it to anyone with an internet connection. Tether's token is called USDT. There are 189 billion of them in circulation.

The peg holds because Tether keeps a dollar in reserves for every token. About 80% of those reserves sit in short-dated US Treasury bills. So when you "deposit" a dollar with Tether, your dollar buys T-bills. Tether keeps the interest.

In 2024 alone, Tether reported $13.7 billion in profits. That's more than Pfizer made that year, and about the same as Coca-Cola.

The money does not come from American bank accounts. About 80% of stablecoin transactions settle outside the United States. The largest user bases sit in emerging markets — anywhere with capital controls or unstable local currencies. Argentines hold USDT instead of pesos. Vietnamese workers send wages home with it. And in Nigeria, importers use USDT to settle invoices in dollars the central bank won't release.

So here's the picture. A company with no banking charter, no FDIC insurance, and no central bank backstop has become one of the twenty largest holders of US government debt. Its "depositors" are millions of unconnected retail users in emerging markets. Its reserves include $23 billion in gold and $8 billion in Bitcoin. And every token can be redeemed instantly, around the clock, on any blockchain that supports it.

I have no idea how this resolves. The numbers may keep working for years.

But the United States now has a structurally important Treasury buyer that has never been tested. The next time a major emerging-market currency cracks, or a major crypto exchange goes down, the redemptions will run through Tether. And Tether will have to sell its T-bills into the open market to honor them.

We can't know when that day comes. We can only count the buyers it would take down with it.

Three more threads worth keeping track of this week, all running independent of the lead story.

Japan's 40-year auction. Japan's 40-year government bond yield touched 4% in January for the first time since the tenor was launched in 2007, and has hovered near record highs since. The 30-year sits around 3.8%. That part of the curve depends on domestic life insurers, who have largely met their regulation-driven buying targets. Prime Minister Sanae Takaichi's government has just pledged tax cuts and higher defence spending for the lower-house election. Supply is rising. The marginal buyer has left the room. The August 2025 yen-carry unwind dropped the S&P 500 about 8% in three days. This time the fuse looks shorter.

Equal weight ahead of cap weight. The Invesco S&P 500 Equal Weight ETF is ahead of SPY by roughly 5 percentage points year-to-date in 2026. That sounds modest. It is the first meaningful reversal of the Magnificent Seven concentration trade in three years — the cap-weighted index outperformed equal-weight by about 32 percentage points from early 2023 through the end of 2025, slightly more than the peak gap of the late-1990s dot-com bubble. The top ten stocks still make up roughly 38% of the S&P 500. Mean reversion in mega-cap concentration tends to start slowly and accelerate.

Argentine reversal risk. Argentina has two recent dollar bonds maturing a year apart. The 2027 bond yields about 5.1%. The 2028 bond yields about 8.9%. The 380 basis point gap is what the market thinks happens to Javier Milei's reforms after his term ends in December 2027. Net international reserves are barely positive. Country risk sits around 580 basis points. The IMF programme requires a further $4 billion of reserve accumulation this year. Argentina has had 23 IMF programmes. The previous 20 ended early. We'll see.

In 1907, a New York banker named Charles Barney ran the second-largest trust company in America. It was called Knickerbocker Trust. The building still stands at Fifth Avenue and 34th Street.

Trust companies in 1907 were the shadow banks of their time. They took deposits, made loans, and held reserves like banks. But they sat under much lighter regulation. They didn't need large cash buffers. They could lend to stock speculators without collateral. And they were not members of the New York Clearing House — which was what passed for a Federal Reserve in those years.

The trust sector had grown fast. New York's trust companies held about $4 billion in deposits by 1907, against $5 billion at the city's national banks. Nearly half the system. Knickerbocker alone held $62 million.

On October 16, 1907, two financiers named Heinze and Morse tried to corner the United Copper Company's stock. The attempt failed. Barney, an associate of Morse, had also bought in. Two days later, a New York newspaper reported the connection. A run on Knickerbocker began that afternoon.

The clearing bank that handled Knickerbocker's checks asked the New York Clearing House for help on October 21. The Clearing House said no. Trust companies were not members.

J.P. Morgan was called in that evening. He sent a young analyst named Benjamin Strong — later the first president of the New York Fed — to look at Knickerbocker's books. Strong came back and told Morgan he couldn't tell whether the institution was solvent in the time available. Morgan refused to commit funds.

On October 22, Knickerbocker opened its doors, paid out $8 million in three hours, and suspended. The run spread to the other trust companies. Within a week, the New York Stock Exchange was almost paralyzed for lack of call money. The Panic of 1907 was on.

The country needed a lender of last resort because the next time the trust companies broke, the Morgan library wouldn't be enough.

Morgan spent the next two weeks locking bank presidents in his library on Madison Avenue until they coughed up rescue funds. The panic broke. But the country had seen what an unregulated parallel banking system could do when the safe-asset chain ran in reverse.

The reaction took six years. The Aldrich-Vreeland Act passed in 1908. The Pujo Committee investigated the money trust in 1912. The Federal Reserve Act passed in December 1913. The motivation in the legislative record was explicit: build a lender of last resort before the next panic.

Every modern crisis is a rhyme of this one. Lehman in 2008 was a Knickerbocker moment — an institution outside the bank safety net that nobody had quite priced. The New York Fed has said as much in its own histories.

So whenever a piece of finance grows large outside the bank perimeter, holds short-term obligations against long-dated or volatile assets, and has no clear backstop on the day it's needed — write down the name of it. Know who runs it. Know who lends to it when the headlines turn.

We'll see.