BThe post lighting up Bluesky: Oliver Willis basically says if they just keep selling AI data centers to each other, nothing can go bad.
AAnd another Bluesky account put it sharper: 'Oh, yeah, the AI bubble is definitely popping now.'
BNothing says rigorous market analysis like 'Oh, yeah.'
AExcept the joke lands because some AI infrastructure deals do look circular. One hyperscaler leases from another, the other buys from the same chipmaker, and everybody books revenue.
BRevenue is revenue. If two companies sign real contracts, even within the same ecosystem, that's real demand.
AReal contracts can rest on hopes of future demand. That's the definition of a bubble.
BCome on, Adam. The companies building data centers aren't startups with no cash. They have real customers, real power, real land.
AThey have real debt, too. If AI demand disappoints, these leases unwind at once.
BThat's exactly my point! The buildout is an arms race. You can't wait for demand to prove itself. You build ahead or you lose.
ANo, you don't raise a hundred billion dollars to see who has the biggest warehouse. At some point the bill comes due.
BAnd if you're late, you're Blockbuster. Infrastructure has always been built ahead of demand, from railroads to fiber. Some went bust, sure, but the survivors created the modern economy.
ARailroads were massively overbuilt. Half the mileage went bankrupt and investors got wiped out.
BSo what's the alternative? No data centers, no AI, ride horses?
ANobody said ride horses. But when the same handful of companies are both suppliers and customers, I want more third-party demand before I call it real.
BFair. If most of the revenue is just the same five names trading leases, it's a warning sign. I'll grant you that.
AThat's my point. Not popping today, maybe, but the structure is frothy.
BAnd if it pops, I say it'll be because AI didn't produce cash returns, not because the leases look incestuous.
AHere's what bugs me: if three companies are each other's top customers, that's not a market. That's a daisy chain.
BBut the data center customer list is broader than three names. You have enterprises moving to cloud, startups renting compute, insurers, hospitals.
AReal, but a lot of the biggest contracts are still between the big tech names. That's the circularity.
BCash-rich companies can afford to overbuild for a decade. That's not a bubble, it's a transition.
ACash-rich until they borrow to pay for it. Then the pain hits all at once.
BDifferent kind of paper trade now: a fight in Pennsylvania.
ALet me guess — another lawmaker and a brokerage account?
BScranton mayor Paige Cognetti, quoted by The Bulwark on Bluesky, says Congressman Rob Bresnahan ran two years ago saying, 'We should not have congressional stock trading.'
BThen, she claims, he went to Congress and traded eight million dollars' worth of stocks.
AWe couldn't independently confirm that eight-million-dollar figure, so treat it as a political charge until the disclosures are public.
BBut if he did trade individual names, that's the problem. Lawmakers shouldn't trade with information voters don't have.
AI agree on principle, but a ban on single stocks pushes people into index funds, and they still have a stake in policy.
BAn index fund is miles away from buying defense stocks a week before a committee vote.
AThere are ETFs for everything now. You'd have to ban all equities, and then you end up with only rich people in Congress.
BSo we allow some trading to avoid a rich-people-only Congress? That's a terrible trade.
AIt's not corruption unless there's proof. Right now it's a claim from a local mayor in the other party.
BVoters don't care if it's proven. They see the trades and they punish the whole institution.
AThen ban it. Broad funds only, held in a blind trust or run by an independent board. But don't pretend enforcement is easy.
BI'd support that. Problem is, every time someone proposes it, leadership kills it in committee.
AAnd if you allow ETFs but ban individual stocks, you can still tilt toward sectors you oversee. Energy, defense, banks.
BStill less direct than buying one stock the day a merger is announced.
ALess direct, but the public sees any million-dollar portfolio and assumes the worst.
BThen make disclosure real-time. I don't need perfection, I need fewer obvious conflicts.
AOn the same theme, David Roberts on Bluesky claims President Trump's personal oil and gas holdings grew by as much as thirty-five point eight million dollars this year, bringing them to as much as eighty-one point six million.
BThose are big numbers, if accurate. That's a direct conflict on energy policy.
AWe couldn't confirm the specific figures, and it's one post. But the president has always had family exposure to oil and gas. That part isn't new.
BIt matters more when he's setting drilling permits and tariffs.
APresidents with family businesses have always faced this. The solution is divestiture or a blind trust, not hoping nobody notices.
BWhich he didn't do any more than the last few did.
AExactly. The office itself is a financial conflict machine.