Kevin Warsh’s First Fed Meeting: A Quiet Signal for AI Infrastructure Owners

The Fed published a forecast it does not believe. The contracts, the megawatts, and the wafers tell a different story. We trust the things you can count.

On Wednesday the Federal Reserve left rates where they have sat since December, at 3.50 to 3.75 percent. The statement was the shortest since the emergency cut of the pandemic. The roll call of who voted for what was gone. Forward guidance, the practice of telling markets where policy is headed, was eliminated. The dot plot, which in March pointed to rate cuts this year, now leans toward a hike. And the official forecast says GDP holds near 2.2 percent and unemployment near 4.3 percent for the next three years, with less disagreement among forecasters than at almost any point on record.

So the Fed published a forecast in which nothing happens. It did this during the largest supply side shock in a generation. Then the chairman stood up and said the opposite.

Two Meetings in One Room

The Summary of Economic Projections is one document. Kevin Warsh’s press conference was another, and they did not agree.

The forecast hugs the trend. It refuses to take a view on artificial intelligence, which is the single most important force acting on supply, productivity, and capital formation today. When a forecaster does not know what is happening, he projects no change. That is what the dots are doing. They cluster around the familiar because the unfamiliar is hard to model.

Warsh took a view. He treated the AI capital cycle as both a demand shock and a supply shock, and he described strong productivity led growth as something to embrace rather than fear. The reference point is Greenspan in the late 1990s, the chairman who recognized a productivity boom and declined to choke it. Warsh has said openly that he wants to govern in that tradition. A Fed that will not fight a supply side boom is, at the margin, a Fed that lets the buildout run. For everyone laying the physical foundation of this thing, that is the headline.

The Sting in the Tail

He did not stop there. He disowned the trade off between inflation and employment that Fed chairs have leaned on for forty years. He said inflation is a choice. That is a strong thing to say while consumer prices are running at 4.2 percent, the hottest in three years.

It raises a question he did not answer. If the policy rate is no longer the lever, what is? The most plausible answer is the balance sheet. And balance sheet tightening does not work by cooling the demand for workers. It works by draining capital out of the system. It raises the price of money and thins the pool of it.

Warren Buffett has a line I keep close. “Interest rates are to asset prices what gravity is to the apple.” When money is cheap, the pull is light and everything floats. When money is dear, gravity returns. The channel barely matters. Whether the Fed raises the price of money through the policy rate or shrinks the quantity of it through the balance sheet, the pull on asset prices strengthens either way.

So the question that decides our portfolio is not where the funds rate sits next quarter. It is whether capital gets dearer or scarcer. And the honest answer is that the Fed did not tell us. It removed its easing bias. It let its own dots drift from a projected cut toward a possible hike. Then it told markets to ignore the dots entirely. It disowned the mechanism that used to explain how policy reaches prices and replaced it with a task force.

The case runs both ways. Warsh is openly friendly to productivity. He was installed by an administration that wants lower rates, and he has argued for cuts himself. If the energy shock fades, the surprise could just as easily be easier money, which would lighten the pull rather than increase it.

That is the point. This was not a signal that gravity is strengthening. It was a signal that no one can tell you where the gravity is set. For the most capital hungry enterprise in history, that uncertainty is the variable to watch.

Scarcity Does Not Punish the Owner

Here is the part the market will get backwards, the way it always does.

Forget the direction for a moment. Gravity, wherever it settles, does not pull on every asset the same way. It pulls hardest on the things highest off the ground. The speculative project with no revenue. The site with no power. The fab that exists only on a slide. It barely touches the asset already sitting on the floor, generating cash, owned outright. And uncertainty alone does some of gravity’s work. When no one can price the cost of capital with confidence, lenders demand more to part with it, and the marginal project is starved first. Scarce or merely uncertain capital does not hurt the person who already owns the finished thing. It entrenches him. A drought does not ruin the farmer who already dug his well. It ruins the man still hoping to find water.

Look at our three positions through that lens.

The construction layer, the engineering and procurement firms that pour the concrete and pull the power, does not weaken when financing dries up. The speculative projects stall first. The funded, contracted backlog keeps moving. The ability to build, the crews and the permits and the interconnection, becomes more valuable, not less, because fewer rivals can finance their way to the same place.

The data center owner is the building in a city where no new building can be financed. When capital is dear, the energized site with a tenant already inside it does not lose value. The waiting list behind it grows. Every project that cannot get off the ground improves the position of the one already standing.

The wafer constraint is a capital constraint wearing a different coat. A leading edge fab is among the most expensive objects humans build. When money tightens, fewer of them get financed and fewer get built. The bottleneck that already favors the incumbents simply lasts longer. Our thesis does not weaken in a capital squeeze. It hardens.

Each of these is the same lesson said three ways. Supply is the binding constraint. A shortage of capital is just another form of supply constraint. It punishes the marginal builder and pays the owner of the ground.

If you are finding value in how I think about the physical layer of AI, consider subscribing. I write for readers who prefer arithmetic to excitement.

The Forecast on the Radio

There is a deeper point in all of this, and it is the one I keep returning to.

The Fed published a forecast it does not believe, alongside a chairman who all but told you so. He spent his time at the podium describing task forces to rebuild how the Fed measures inflation, communicates, and reads an economy its own instruments no longer capture. When the people steering admit their gauges are disconnected from the thing they are steering, you should stop reading the gauges and start counting the assets.

The dot plot is the weather forecast on the radio. It is dramatic and confident and revised every few months. The backlog, the rental rate, the power queue, the wafer allocation, that is the soil. We have always farmed the land and ignored the broadcast. This week the Fed effectively told everyone to do the same.

I do not know what rates will do this year. I never have. I do not know which way the Fed will lean, and after this week, neither does the Fed. But I know the contracts are signed, the sites are full, and the wafers are spoken for. If capital gets dearer, it will not be the people holding those assets who suffer. It will be everyone still trying to build them. And if capital stays cheap, the buildout runs harder and the same owners collect the rent. The asymmetry favors the same side either way.

The official forecast says nothing is changing. The megawatts, the contracts, and the wafers say everything is. We trust the things you can count.

Neel Khokhani

Founder and CEO, Epochal Corporation

@neel_epochal