Actually, the U.S. restricting data centers (AIDCs) is a good thing.
Just think of an AIDC as a Bitcoin mining farm.
All of a sudden, no one’s allowed to build new mining farms.
Who’s panicking?
The people selling mining rigs?
Or the people who already own mining farms?
Mining farm owners must be quietly thrilled.
The mining farm you own instantly becomes one of a kind.
Others want to get in, but they can’t.
That’s a moat—and the government itself is digging it for you.
It’s exactly the same with AIDCs.
Existing data centers instantly become exclusive suppliers.
Their bargaining power surges, and all the orders pile in their laps.
Why?
Because no one else can build them.
Delete the plans for new construction outright.
If it's already under construction, keep building.
All the damage lands on the “future”: future orders, future capacity, and the premium paid for hardware in the future.
What about what's already there?
Not a hair harmed—in fact, it's gone up in value.
So who takes the hit?
Hardware.
Every blow lands on hardware.
To everyone buying the dip in hardware: you read the moat right, and you read the demand right too.
The only mistake was leaving out one player.
Once policy enters the equation, every damn thing—demand, moats—has to take a back seat.
But that's not all. Think one level deeper, and it gets really interesting.
If the U.S. really gets serious, what happens to those millions of cards?
They can only flow back to Asia.
China's AIDC sector won't be short on cards anymore, and Chinese models will actually become more competitive.
Well, would you look at that.
The U.S. is pushing its own global tech leadership away with its own hands.
Pushing it to whom?
The very people it least wants to hand it to.
Think about it. Really think about it.
And if you keep following the logic, it gets even more brutal.
If the bill passes: compute stays scarce, existing data centers sit back and collect rent, AI prices stay firm, and the sector charges ahead.
If the bill fails: the uncertainty is gone, and AI charges ahead all the same.
Either way, it's a win.
That's what the saying “if it doesn't happen, that's still good news” really means.
How could anyone get that lucky?
They can.
Because the ones who are afraid aren't the players in the game—they're the ones watching from the sidelines.
If new construction is restricted, the first tier gets split in two.
Data centers already built and generating cash flow get crossed off the “first to die” list—they go from constantly borrowing money to sitting back and collecting rent.
The ones still borrowing to expand will die even faster.
They were betting on new projects, and those plans have just been wiped clean.
What about hardware?
It won't die, but the premium will.
The business shifts from making money on growth to divvying up what's already there.
Multiples? They'll collapse on their own.
Like housing under purchase restrictions: old homes become one of a kind, nobody even looks at new developments despite the discounts, and agents selling new homes get their commissions slashed in half first.
From here on, you only need to watch one thing.
The next capex guidance from cloud providers, and the divergence between data center stocks and hardware stocks.
If the gap widens, the thesis holds.
If it narrows, it means the market never believed this ban would actually take effect.
That's easy, then.
Treat it like nothing happened and get back to calculating FCF.