Hey {{first_name|Investor}} -
Y Combinator's new wish list says the AI buildout is short of transformers and switchgear. I sorted every holding in four power ETFs to see how much of each fund is that gear, and the share ran from about 23% to about 54%.
So the word "power" on a fund tells you very little. AIPO holds slightly more in chips, data centers and bitcoin miners than in gear. VOLT holds almost as much in utilities as in gear. The sort is mine, and the judgment calls are below, so you can move them and see what changes.
What YC asked for
Y Combinator posts a list of ideas it wants founders to bring it. Item 9 on the Winter 2027 list, "Data Center Supply Chain," is by partner Jared Friedman. When I checked it on October 30, it said "Large power transformers have a three-year lead time," and that switchgear, circuit breakers and liquid cooling equipment are sold out. Item 6, "Wall Street for the Physical Buildout," asks for lenders and leasing companies, because "somebody has to pay for all that steel and silicon up front."
Wood Mackenzie's survey, published October 15, 2025, is close to YC's number: power transformers averaged 128 weeks and switchgear 44 weeks. A wish list shows where one investor is looking. It proves nothing about who gets paid.
The money has been arriving ahead of the list. Canary Media reported on February 18 that Heron Power raised $140 million to build solid-state transformers, a newer electronic design. It plans full-scale production in the second half of 2027 and a target of 40 gigawatts a year. On October 21, Heron said Vantage has more than 1 gigawatt of Heron's manufacturing capacity for its data center campuses; the release gives no delivery dates and doesn't say whether that's a firm order. So the startups are already drawing commitments. What they can't do yet is ship at scale, which in my read leaves the equipment going in today with the companies that already have factories.
What the factories are reporting
Five read-throughs, each from its latest reported quarter:
Vertiv ($VRT), in its October 28 8-K for the quarter ended September 30: net sales of $4.31 billion, up 61%, with 55% organic growth, and management said organic orders more than doubled without giving a dollar figure. Deferred revenue rose to $4.12 billion from $1.81 billion at December 31. On the call, management described customers paying ahead of delivery.
Hubbell ($HUBB), on its October 27 call for the quarter ended September 30: data center sales up more than 80% organically.
GE Vernova ($GEV), Electrification segment, in its October 28 release and call for the quarter ended September 30: equipment backlog of $42.2 billion, up 60% year over year, with $5 billion of that from the Prolec GE acquisition. Orders of $5.4 billion rose 6% as reported and fell 9% organically, against a large Saudi order a year earlier. Data center orders were about $1.4 billion in the quarter and more than $6 billion year to date.
Powell Industries ($POWL), in its 10-Q for the fiscal quarter ended June 30: bookings of $934.2 million, up 158%, and backlog of $2.4 billion. One data center project of more than $400 million is a big part of that bookings figure, so I wouldn't read it as a run rate.
Eaton ($ETN), in its November 3 8-K for the quarter ended September 30: sales of $8.9 billion, up 28%, with 19% organic growth. On the call, management said data center orders rose more than 100% on a rolling 12-month basis, data center sales rose 65%, and the Electrical backlog passed $20 billion, up 46%. It also agreed to buy COL Group to add power distribution capacity in Europe.
On its Q3 call, GE Vernova also said it agreed to pay $9 billion for Trench Group, which makes high-voltage bushings and instrument transformers, with closing expected in late 2027. When a company that size pays up for a component maker, I read it as a sign of how it sees that bottleneck.
The rest of this issue is for Insiders.
Below the line: every holding in AIPO, VOLT, GRID and JOUL sorted by what each company makes, the three judgment calls that move the numbers, what fills each fund besides gear, the two companies all four share, what's in the price, and what could break it, Washington included.
This is where it gets interesting.
Insiders get the full research — the emerging patterns, the second-order effects, and the trends I think matter before they're obvious
Unlock the full postA subscription gets you:
- Weekly deep dives on trends before they're consensus
- Operator-level research and frameworks behind every thesis
- Full archive of every past issue and report