The people who must live with AI capital’s decisions are not the people who get to make them; I used to trade the paper claims before the crops were planted, and this is the same machinery at larger scale. The giant bet is a paper harvest with no local accountability. Erin Woo’s Wall Street Journal technology newsletter, “AI Spenders Get Spooked”, the source for Main Street Independent’s roundup, reports that OpenAI raised less than half of the $30 billion it sought at a $1.25 trillion valuation; Google reported that its AI offerings were bleeding growth from its search-ad cash cow and, separately, saw its market capitalization shed $500 billion in the stock’s worst day ever; and Microsoft offered voluntary buyouts to 60,000 employees, about one-quarter of its staff, as it continued investing in AI. This is not merely a financing scare. It is a map of concentrated decisions and diffuse consequences.

The AI buildout is financing itself with voluntary buyouts, debt, and promises of a future nobody has yet earned. That is not merely a technology story; it is the rentier story wearing a new circuit board.

The strongest honest case for the cooling is plain enough: AI infrastructure is expensive, debt is no longer cheap, data centers are delayed, and monetization remains unfinished. Lenders are not obliged to finance a story at any price. If much of this is a bubble, demanding harder terms before more borrowed money is spent on data centers is not pessimism; it is ordinary discipline. Search advertising is a cash cow, not a perpetual guarantee, and no board should pretend that yesterday’s profits are already tomorrow’s AI earnings.

The strongest case for the money is also honest. Computing infrastructure can make tools, research, and knowledge available at a scale no single laboratory or small firm could afford, and capital sometimes must build before it knows what a new machine is for. The Microsoft buyouts do not establish that every departure was caused by AI, and Anthropic’s founder voting control can be defended as stewardship of long-horizon research. None of this proves the machines will prove useless. It proves that the capital surrounding them has been asked to price them as though certainty had already arrived. Prudence is not the betrayal.

But the cooling exposes the bargain beneath the boom. Ahead of an expected IPO next year, OpenAI had sought $30 billion at a $1.4 trillion valuation and raised less than half of that at $1.25 trillion. Anthropic’s IPO papers do not explicitly list a total addressable market; the filing suggests a $40 trillion-a-year opportunity only if the company captures the entirety of demand, while the papers give seven founders voting power at the expense of incoming investors. Some neo-cloud companies are delaying or downsizing their IPOs as financing conditions tighten. MSI’s earlier reporting on tighter AI-financing terms and Anthropic’s nearly $9 billion in infrastructure spending show that the strain runs through the financing stack, not merely through one company’s bad morning.

Call OpenAI’s undershot round what it is: market discipline, not fraud. Call Google’s $500 billion fall what it is: evidence that the old search-ad cash cow is being consumed by the thing expected to replace it, not proof that AI is worthless. Call Anthropic’s founder voting control not fraud either, but concentrated decision rights. Lenders and prospective investors finance the promise while seven founders keep the steering wheel. That can produce prudent research. It can also produce a private command center whose consequences extend beyond the people who ordered the journey.

Microsoft’s announcement belongs to the same indictment. The voluntary buyouts need not all be AI-caused for the capital choice to matter. While directing enormous resources toward AI, Microsoft offered 60,000 workers a voluntary exit, about one-quarter of its staff. That makes the settlement of working lives adjustable to the buildout model. This is not merely technical change. This is power deciding whose life may become the margin.

I know the seduction of the paper claim because I lived inside it. I used to trade agricultural futures on a Chicago desk. Corn can be hedged without being grown; weather can be priced without a farmer; risk can travel farther from the field than any bus or train. The abstract claim can be useful. The mistake begins when the claim is mistaken for the thing.

A farm contract can outrun its crop for a season, but it names a quantity, a delivery month, a grade, and a place; somebody must answer when contract and harvest fail to agree. The AI buildout has made a grander fiction: the final product is perpetual growth, the near-term subtraction arrives as a voluntary buyout, and the counterparty is a future market no one has yet entered. AI capital knows the chips, the data-center schedules, the debt terms, and the promised utilization. It does not know what an institution owes the place in which it operates. The people who do the work and live with the result often know things a distant calculator cannot. I helped build part of that abstraction. I also know what it leaves behind.

This is where the old fusionist promise broke. The Burkean tradition treats society as “a partnership between those who are living, those who are dead, and those to be born”; on that view, property is a form of stewardship, not merely a claim upon the world. The new buildout keeps the arithmetic and cuts the covenant. Capital keeps the upside of the forecast; labor absorbs the subtraction; founders keep the votes; and everyone else is invited to admire the spreadsheet.

Ownership is being used to secede from consequences. Private property is a serious good, but it answers to the human purposes that justify it. A corporation does not become innocent because it is incorporated. No claim on future earnings outranks the dignity of the person asked to bear its cost. The earth was given for all; that truth does not abolish property, but it prevents property from becoming a sovereign right to dispose of other people’s lives.

One proposed remedy is to place the machinery in public hands: state compute, state data, a national AI authority. The fear behind that remedy is legitimate; a private titan can become a private government, answerable to no one. But an unanswerable public command is not made just because its payroll comes from the public. Concentrated capital and concentrated state power are the same disease in two coats. A conservative cure must disperse power before it merely renames the owner.

The fashionable answer, which usually arrives from my own side when the market shocks it, is to put a more competent state in charge. A public AI authority would not mend answerability; it would centralize the servers, power contracts, data, and failure in the same distance. The statist answer treats the public payroll as a cure for private command; my own side treats the corporation as fate. Both spare us the work of building institutions capable of refusal. The cure for a command center should not be a larger command center with better manners.

Here in Adams County, I remember fair boards, parish councils, hospital boards, and farm cooperatives—not as perfect, but as places where decisions had names behind them. None was pure. Some excluded people. Some failed. But each put decisions through institutions that could be named, argued with, and sometimes called to account at the next meeting. We do not romanticize them; we remember what it cost to lose them. Their value was not that they satisfied a spreadsheet. Their value was that the actual community could remain negotiable rather than disposable in advance.

A company must be able to operate far beyond its birthplace. But when its dependence upon a community becomes vast, the burden of answerability cannot vanish through distance. At some scale, the distant spreadsheet is not enough. The people who live with the consequences need a seat, a voice, and a way to say no.

The countermodel begins close to home. A utility serving rural homes should be owned by its members. Local credit should be owned by the people who save and borrow through it. Farm marketing and processing should answer to producers rather than distant middlemen. Municipal broadband can own the pipe instead of renting it; a local cooperative can own part of the compute serving its members; workers can share in the productivity they create; and credit unions can finance tools for farms, clinics, and shops without turning every loan into a wager on a distant IPO. Libraries, hospitals, universities, and parishes can serve as civic stewards of machines used on behalf of people with little bargaining power.

Where AI’s physical dependencies become ordinary civic life—power, transmission, broadband, data centers, and technical services—cooperatives or mutuals should operate at the lowest capable scale. Open standards and interoperable systems should keep one owner from holding an entire community’s tools hostage. Patient local capital can finance what remote speculation will not. None of this asks the state to become the machine. It asks each institution to remain answerable to the persons whose common good it serves.

“Local” must have enforceable teeth, not decorative consultation. Suppose a proposed AI data center would draw enough power to alter residential rates, taxes, road burdens, or grid reliability. The affected member-owned utility’s board should have to approve the large-load contract after a public hearing and a published cost-and-risk study—not merely receive notice. Where a project depends on public land, water, sewer lines, roads, or tax concessions, the relevant county or municipality should possess a consent right before permits or incentives are issued. Contracts joining land, labor, and energy should reserve binding votes, not advisory courtesy seats, for workers and residents affected by the siting. State law can empower those institutions, and the mechanisms should be written into tariffs, intergovernmental agreements, site leases, and financing covenants. A project that cannot win the consent of the community whose infrastructure it wants to use does not have permission to proceed.

Private capital may fund experiments. It should not own without answerability the bottlenecks on which everyone else must depend. Government should enable, regulate, and step in only where no smaller body can do the job. The point is not to abolish distance; it is to make distance answer to the people whom it uses.

It will be harder here. AI capital is immense, technical, and impatient. A cooperative board cannot announce a hundred-billion-dollar data-center campus next Tuesday; patient capital will not move at the speed of a trading desk; founders do not surrender their votes to a parish meeting; and a local institution may struggle to recruit the people who can build the thing. The people will argue about cost, siting, benefit, and risk, and sometimes the answer will be no. That difficulty is the point. It is where distributed ownership stops being a slogan and begins to mean that the people answerable for the decision are the people who live with it.

An Oakeshottian preference for the familiar over the unknown cannot repeal the machine, but it can tell us not to surrender the whole order to it. The point is not to romanticize smallness. It is to keep capital productive without making it sovereign. Patient finance is part of freedom, not a defect in it.

Here in Friendship, the working example is not a national AI authority. It is the form of the Adams-Columbia Electric Cooperative: infrastructure held in common by the people who depend on it. The AI version will be harder, slower, and less flattering to anyone who wants the next trillion-dollar story told before breakfast. But if the machinery is coming, the town must have a chair at its table. Otherwise we shall have built the future in our own county and discovered, too late, that we were only allowed to furnish the room.

We cannot restore the rail crews, the parish schools, the mill jobs, or the shops on Main Street. We can refuse to make the next institution untouchable. Build the cooperative, the mutual, the member-owned utility, the local credit union, the restored Main Street. Let private capital gamble at its own scale. Leave the town its life.