The Lords Are Minting Coins Again

Why Amazon, Meta, and Tether are quietly issuing currency — and what the rise of tech neofeudalism actually looks like up close

Someone asked me the other day whether there’s ever going to be an Amazon stablecoin. The honest answer is: there kind of already is, and that’s the part of the story everyone keeps missing.

In early May, AWS quietly turned on USDC payments for AI agents through Coinbase and Stripe. There was no press tour, no logo reveal, no “AmazonCoin” branding. Just a payment rail, switched on under the hood, so that machines running on Amazon’s infrastructure could pay each other in dollar-pegged tokens that Amazon doesn’t issue but does intermediate. It’s the most boring possible version of a private currency, which is exactly why it’s the one that’s actually shipping. The Wall Street Journal later reported that Amazon and Walmart are both exploring branded stablecoins of their own — which is the part that finally got people typing the phrase into search bars.

And Amazon isn’t alone. Meta launched a stablecoin pilot on Facebook the same month, drawing an immediate complaint letter from Senator Elizabeth Warren — Meta’s second swing at this after Libra died in 2022. Tether is launching an official stablecoin with the government of Georgia, which is a sovereign state outsourcing its mint to a private platform. SoFi is rolling out a bank-issued stablecoin to fifteen million users. Thirty-seven European banks are backing a euro stablecoin called Qivalis ahead of a 2026 launch. The U.S. Treasury Secretary has named stablecoin growth a national priority, because every new private dollar in circulation creates new demand for U.S. debt — the sovereign isn’t fighting this, it’s recruiting. It’s the kind of pattern that’s earned a name — tech neofeudalism — and the name is mostly right and partly misleading at once, which is exactly why it’s worth taking seriously.

Before the gloom gets too thick, the part worth saying out loud: for now, this is mostly working, and in a world where the ground shifts under everyone every six months, the case for letting a few very large hands carry the load isn’t crazy. SpaceX is putting up rockets that the rest of the planet can’t match. Amazon is the one running the experiments at the scale where you actually learn something. OpenAI and Anthropic are doing real work on data processing that wasn’t possible three years ago. If you’re a country trying to stay in the front pack while AI, energy, and space all shake out at once, the concentration is partly how you stay in the front pack — five firms moving fast beats five hundred moving carefully when the window is this narrow. That’s a real argument and it deserves a hearing before the feudal frame eats the room.

The worry isn’t now. The worry is the second act. Concentration is fine while the lords are still building cathedrals. It curdles when they start blocking anyone who tries to build a different one. Standard Oil was great for getting kerosene cheap into American homes — for about fifteen years. Then it spent the next fifteen making sure nobody else could refine. That’s the turn worth watching for, and the early signals are already in the corpus.

The instinct to call this neofeudalism is everywhere right now, and the instinct mostly looks at the wrong stuff. The colorful story is Peter Thiel enrolling his kids in Buenos Aires schools and buying a plot in Uruguay that might be a bunker, while Milei’s cabinet chief pitches Argentina as “the new land of freedom” for fleeing billionaires. The dramatic story is the U.S. government taking equity stakes in Intel, in nine quantum firms at once, in chip and mineral firms across the board — Lutnick at Commerce pressuring CHIPS recipients into ownership concessions. Those stories are real. They are also probably overcalled. Every tech wave does this. Railroads in the 1860s had robber barons and captured legislatures. Standard Oil ate the 1880s. AT&T ate most of the 20th century. Each one got concentrated, then unwound — Sherman 1890, the consent decree of 1984. The tools exist, and historically they showed up after the build-out delivered most of its gains, not during.

What people are missing is that the concentration the press is panicking about — billionaires, equity stakes, oligarchic politics — isn’t the structural part. Those are symptoms of a normal tech buildout. The structural part is quieter: a small business today operates on four or five pieces of land it doesn’t own. Amazon’s marketplace for selling. Google’s index for being found. Apple’s App Store for distribution. Meta’s audience for reach. Stripe or now AWS for clearing. None of those are owned by the seller. All of them charge rent. All of them can change the terms unilaterally, and have. Epic Games is in front of the Supreme Court right now asking for review on Apple’s App Store fees — a vassal asking whether he can deal with customers directly without paying the lord his cut. France tried to regulate Amazon’s marketplace fee floor for books and lost at its top administrative court. Meta has begun the long-delayed monetization of WhatsApp — two billion users who built their daily communication on someone else’s land are now learning what the rent is.

That arrangement — working land you don’t own, paying the lord his cut — is the actual definition of the feudal relationship. Not concentration of wealth. Wealth concentrates in every era; republics have handled oligarchy before. The defining medieval move was that peasants worked the manor but had no claim to it, and the lord’s claim wasn’t earned by competition but by control of the ground. That structure, in commercial form, is already complete. The currency layer is just the last sovereign function snapping into place.

Run through the four classical jobs of a sovereign and look at who actually performs each one now:

• Coinage — Meta’s stablecoin pilot. Tether issuing for Georgia. AWS settling AI-agent payments in USDC. The European bank consortium behind Qivalis.
• Justice — App Store appeals. Amazon’s seller-dispute machinery. Meta’s Oversight Board reading like a small constitutional court. Content moderation has quietly become case law.
• Force — Palantir threading itself into Western police infrastructure. OpenAI’s own staff publicly split over when the company should call the police on its users. The data-broker layer that quietly decides who gets stopped and who doesn’t.
• Identity — Apple ID. Google account. “Sign in with…” as the new passport.

No single platform has all four. Each has two or three. That’s actually more feudal than one lord owning everything, because real feudalism was always a patchwork — the peasant owed allegiance to the manor for one thing, the guild for another, the church for a third, and the king, in theory, for justice. A modern small business answers to roughly the same four overlapping authorities. The uniforms are different. The shape is not.

The honest question isn’t whether this arrangement is bad — it’s been remarkably productive so far. The question is what happens when the next interesting thing tries to start. A better search engine gets acquired before it scales. A cheaper payment rail gets routed around by a default deal. A new social network can’t reach anyone because reach itself is rented. That’s the point at which concentration stops being the engine and starts being the brake, and you can already see the early version of it in the smaller stories the corpus is throwing off — the seller who can’t leave Amazon because there’s nowhere else, the developer who can’t leave the App Store, the founder who quietly relocates because the local market is sewn shut. Talent doesn’t fight that arrangement for very long. It moves.

So the headline probably isn’t Thiel buying a bunker in Uruguay. The headline is that the median seller in 2026 is a tenant on four manors at once, and the manors are starting to issue coin. The state, for now, looks like an early-modern monarchy — present, formally sovereign, but increasingly dependent on the chartered companies it’s also trying to license. For the moment the deal mostly works: the rockets fly, the models train, the warehouses ship. Whether it stays that way depends on whether the lords keep building cathedrals or start blocking the next builder. The currency rails being laid this year will probably decide which one it is.