Dear Fellow Investor,
Elon Musk confirmed this week that he is building a fourth data center in the Memphis area.
That is the fourth, in one metro area, in roughly two years.
And it landed in the same week his crews began dragging the old power supply out of the third one.
Sixty-nine industrial gas turbines, bolted to flatbed trailers in Southaven, Mississippi, are now being removed on a timeline he agreed to with state regulators on July 30.
Every one of them is to be gone by next July, replaced by a permanent 1.2 gigawatt power plant he is building himself, on his own land, because the local utility told him plainly that it did not have the capacity — and neither did anyone else.
So here is the position the richest man in the world is actually in.
He is adding demand faster than he is adding power.
I want to show you what that costs him, because it is the most interesting number in the market right now and almost nobody is looking at it.
It is not a dollar figure.
It is a date.
If you want a large power transformer today — the refrigerator-sized machine that steps high-voltage transmission current down into something a building can actually use — the average wait in the United States is now about 128 weeks.
Two and a half years.
For the biggest generator step-up units it runs closer to 144 weeks, and specialty units are being quoted at four years.
Five years ago those same machines shipped in 50 to 80 weeks.
Earlier this month, Siemens Energy reported that its grid equipment backlog had hit a record 51 billion euros, with orders up 28% and running roughly one and a half times faster than the factory can ship.
Their fix — a 50% capacity expansion — does not arrive until 2030.
GE Vernova is carrying a backlog north of $100 billion, the largest ever reported in power equipment.
And here is the part I find genuinely remarkable: the bottleneck is not money and it is not even steel.
It is people.
Roughly fifteen thousand Americans know how to hand-wind the copper coils inside these machines.
You cannot download that, you cannot 3-D print it, and you cannot pay it to happen faster.
Which brings me to the lesson I want you to take away today.
A backlog is a queue, and a queue is a price.
In my career I have watched capital solve almost every problem put in front of it. Capital cannot solve a delivery date.
When money runs into a physical line it does not stand in it — it pays somebody to move it forward.
And that payment does not go to the company with the biggest war chest.
It goes to whoever holds the open slot.
That is the whole game, and it is not confined to one billionaire in Memphis.
On July 22, roughly 3,800 megawatts of data-center load dropped off the grid in northern Virginia after a single 230-kilovolt line faulted — the largest event of its kind in the history of the grid operator there, which announced earlier this week that it is now writing new rules for facilities like these.
In Texas, the grid operator is sitting on about 1,800 interconnection requests, roughly nine out of ten of them data centers.
Everywhere you look, the answer the electrical system is giving the AI boom is the same word: wait.
Which is exactly why I have spent the last nine months studying one small, publicly traded American company.
It is not a chip designer. It is not a hyperscaler. It does not have a single word of artificial intelligence in its filings.
It builds the custom high-voltage equipment everyone in that line is waiting for, and it built its entire reputation on one thing the giant conglomerates cannot offer at any price: speed.
Wall Street still has it filed under boring industrials.
I think it is the most important company almost nobody on this list has heard of, and I have written up the full case — the name, the ticker, what to pay for it, and when to sell it — in a new research report called The Transformer Trade.
You can read the whole story of how I found it, and how to get the report, right here.
I will tell you plainly what I believe and what I do not.
I do not know when the market will wake up to this. Nobody does, and anyone who tells you otherwise is selling you something.
But I know the orders are being signed now, because they have to be.
The permanent power that has to be running by next summer cannot be ordered next spring — at 128 weeks, the machines that get plugged in during 2027 were bought and scheduled a long time before either of us read a headline about them.
That paperwork is moving across desks this month.
By the time it is a press release, the repricing will have already happened.
"The Buck Stops Here,"
P.S. Read that timeline once more, because it is the entire opportunity in two lines. The temporary turbines start coming down this month and must all be gone by next July. The permanent equipment that replaces gear like that carries a two-and-a-half-year lead time. Do that subtraction and you will understand why I think somebody is being paid an enormous premium right now to jump a line that money is not supposed to be able to jump — and why I want you to know their name before the Journal prints it.
Found this helpful? Share it with others.
Written by Dylan Jovine