A newsletter that can call the investor who sold at the bottom a “fool” and then, in the next paragraph, tell you that the same instinct that helped your ancestors survive a lion attack is why you should buy SpaceX at the IPO—that newsletter is not doing analysis. It is doing liturgy. And the liturgy’s function, whether the editor knows it or not, is to naturalize the extraction so the locked-in public does not ask what it is paying for.
The Markets A.M. edition asks for your “shoeshine-boy story”—the old Joe Kennedy anecdote about getting out in 1929—and then inverts it, the way a con man inverts a mark’s own suspicion to make the sale. Kennedy heard shoeshine boys giving stock tips and concluded the end was near. The Journal’s editor heard a few ride-share drivers mention SpaceX and concluded the runway is long: mass participation is proof the prosperity is broad and real, and anyone who walks away too soon “is a fool.” That is a clever reframing. It is also, as Galbraith might observe, the precise logic of the bezzle—the interval between the extraction and the moment the locked-in public realizes it has already paid. When the cab driver and the barista are reciting your bull case back to you, the extraction has already reached the retail layer; everybody in the room is now somebody’s customer.
It is true, in the narrow sense in which a business-page newsletter is usually right, that market psychology is real—that the instinct to join a rally, to follow the herd, to trade on a story, is a force and that ancient hominids did better spotting predators in groups than alone. The trouble is that the infrastructure consuming the megawatt-hours right now is not a predator on the Pleistocene savannah. It is a set of deliberate vertical-integration bets, each contracted out in signed power-purchase agreements that shift the long-tail financial risk of the AI buildout onto ratepayers while keeping the equity upside entirely private. Microsoft’s twenty-year contract to restart Three Mile Island locks in 835 megawatts of nuclear baseload, and Meta’s parallel deal for the full 1.1-gigawatt output of Constellation’s Clinton plant does the same in Illinois. Those contracts are not psychology. They are architecture. And architecture, unlike market sentiment, is designed.
The newsletter would prefer you not to ask who built it, or to what end. It lists, in the same breath, the SpaceX IPO, Marvell Technology’s induction into the S&P 500, Micron rejoining the trillion-dollar club, and a bitcoin-heavy treasury company clawing back from a 24 percent weekly drop, and it treats them all as trivia in a bull market that “gets stronger” when “everyday people talk stocks with the confidence of veterans.” The list is presented as a neutral roundup, but the roundup has a structural argument smuggled inside it: the concentration everyone else is nervous about is exactly what a durable bull market looks like when a handful of firms are building the next infrastructure layer of the global economy. That is a claim. And the claim, if you check it against the documentary record, is the claim a rent-extractor makes just before the rent turns into a loss.
The mechanism that delivers those rents is what Cory Doctorow calls enshittification, and the frame insists that we treat it as a mechanical outcome rather than a moral lapse. The platform is good to users to lock them in, then good to business customers to lock them in, then it extracts value from both for shareholders, and then it collapses. What makes collapse not just possible but predictable is that the four forces that historically constrained the extraction—competition, regulation, adversarial interoperability, and worker leverage—have been systematically dismantled. Competition was ceded when the Justice Department let the default-search deals bake in, allowing a single firm to pay Apple twenty billion dollars a year for a distribution moat no rival can cross. Regulatory bite was blunted when network infrastructure was reclassified under a light-touch Title I framework, a deliberate retreat from the Bell-era common-carrier obligations that once prevented a network owner from throttling a competitor. Interoperability was outlawed by trade agreements that hardwired anti-circumvention provisions like DMCA §1201 into domestic law, making it a federal felony to build, distribute, or use the reverse-engineering tools that a rival needs to compete on the merits. And worker leverage was bled out by the layoff cycle and the non-compete enforcement that followed: the engineer who might have refused to enshittify the product on moral grounds learned, after a quarter of a million colleagues were cut in a single year, that her options were narrower than her equity grants had told her. What remains is a regulatory layer the newsletter treats as irrelevant, because to treat it as relevant would be to admit that the bull market is not an instinct but an artifact.
The newsletter’s logic—that caution “creeps in” after a “healthy purge of fluff,” that the nervous crowd mistakes concentration for fragility, that the “American innovation engine” transcends a little oil shock and a few missile strikes—is logic that naturalizes the extraction by pretending it is a weather event rather than a designed system. Asian stocks are tumbling and Brent crude is spiking five percent because the geopolitical realities the newsletter gestures at are real; what the newsletter does not mention is that the same hyperscalers whose IPOs and S&P 500 inductions it is selling you this morning now control seventy-one percent of international submarine-cable capacity, a share that grew from ten percent in 2014 not because of ancestral survival instincts but because of a deliberate strategy of acquiring the physical layer underneath a policy apparatus that had stopped asking whether vertical integration into the seabed might someday constitute a chokepoint. The TSMC dependency, the CUDA lock-in, the hyperscaler nuclear PPAs, the app-store tax on every downstream startup that reaches scale—none of these appear in the report because the report is not designed to let you see them. It is designed to make them feel like the weather.
The reply that the AI buildout is genuine innovation, not extraction, mistakes the pile of capital expenditure for the outcome. Capital expenditure is not productivity; it is a bet. The question the newsletter refuses is the engineer’s question: who bears the downside if the bet fails? The power-purchase agreements answer it. When the algorithmic payoff disappoints, the megawatts are already drawn, the reactors are already restarted, and the bill is already filed with the public utility commission. Innovation that requires the public to underwrite the downside and surrender the upside is not innovation. It is a rebranding of extraction, and the rebranding is the newsletter’s only product.
There is a structural position on the other side of this trade, though it will not appear in the morning roundup. Interoperability mandates that lower switching costs so the physical layer is not permanently locked into four corporate silos. A federal privacy law with a private right of action so enforcement does not depend on whether the commissioner happens to feel like biting. Right-to-repair legislation that bans parts-pairing at the hardware layer so the data-centre supply chain is not captive to a technician’s unlock code that a farmer outside Des Moines could have written herself if the DMCA did not make it a felony. The consultations that produce these rules are open in the background—at the Competition Bureau, at the CRTC, at the FTC—and none of them will produce a bill in the current Parliament or Congress, and the bill they eventually produce will be worse than the submissions. The CRTC’s wholesale-access proceeding is scheduled to wrap in 2027. The Three Mile Island restart comes online in 2028. The regulatory clock and the physical clock are not synchronized, and the locked-in public pays the difference. By the time the rule is written, the asset is already locked in for two decades.
The subscription button is at the bottom of the newsletter. The deadline, for now, is the one the regulated respect. The data centres are drinking the aquifer in the meantime.