The Fed raised rates a quarter point. Draw the line from there to a data center.

On 16 September the Federal Open Market Committee raised its target range for the federal funds rate by a quarter of a percentage point, to 3.75–4.00%. The vote was 12–0. The statement describes an economy expanding at a solid pace, with job growth keeping up with the workforce and unemployment stable, and says the move supports a timelier return to the Committee's 2% inflation goal. It was the first increase since 2023.

For most readers that is a headline about mortgages. For this map it is a headline about concrete.

The chain, one link at a time

An AI data center is a capital project. Someone buys land, builds a shell, installs power and cooling, fills it with servers, and only then earns revenue from it. The gap between spending and earning is months to years, and it is bridged with debt: project finance, private credit, leases on the equipment, bonds at the parent company. That is true whether the name on the building is a cloud provider, a specialist operator or a landlord.

Debt has a price, and that price is set off benchmark rates. When the policy rate rises, the cost of bridging that gap rises with it — not on existing fixed-rate debt, but on everything financed from here. A project that penciled at one cost of capital has to clear a higher bar at another. Nothing about the technology changed; the arithmetic did.

This is also why the money behind the build-out shows up on our map at all. BlackRock and SoftBank Group are on it not because they build anything, but because the AI economy currently runs on other people's balance sheets.

What it does not do

It does not, by itself, break anything. A quarter point is a quarter point.

The thesis tracker we follow — Brad Gerstner's published framework, which we treat as his view and not ours — puts his line somewhere else entirely: the 10-year Treasury yield approaching about 5.5%, with oil retreating, is the combination he describes as the real pressure point. The policy rate is not that. Long rates are set by the bond market's view of inflation and growth over a decade, and they can fall while the Fed is raising.

So why does the map treat this as a thesis-changing event rather than a minor one? Because of a rule we wrote before this happened and have to live with: a decision by a central bank is a binary catalyst, not an estimate or a forecast. Either the rate moved or it did not. When a rule is inconvenient in a particular week, that is when it is doing its job — and this week it points the rates row against the thesis, even though the number Gerstner named has not been reached.

The part where we argue with ourselves

There is a genuine counter-argument, and leaving it out would be dishonest. The Fed is raising into what it describes as solid growth, strong productivity and robust capital investment. Capital investment, right now, substantially means this build-out. A central bank raising rates because the economy is strong is a very different environment from one raising because it has lost control of prices. Companies whose revenue compounds at triple digits, which is the part of this economy the earnings prints keep showing, are also less sensitive to a quarter point than companies growing at 4%.

So: a real headwind, in a strong economy, well short of the level the loudest bear case requires. All three of those are true at once, and the useful discipline is to hold them at once rather than picking the one that matches your position.

How to read this on the map

You will not find a red ring around a company because of this event. It attaches to a thesis row, not to a business, because we have no evidence about how any specific company's financing costs changed. Signals sit where their evidence sits. That restraint is the whole product.