Every hyperscaler soared this morning except the one I actually like.

SpaceX is the only hyperscaler I own with real conviction, and it fell to 134 while Alphabet, Microsoft and Amazon ran.

The market’s reading that as a political problem for Musk after last night, and I get it. I also think it’s wrong on a horizon longer than a week, and I’ll explain why in a minute.

First, the thing that matters more.

We knew gridlock was coming for the AI build-out. It got worse than anyone expected. Texas, the state Musk was eyeing for his next mega site, just elected a governor who ran on a moratorium for new data centers. Jefferies put 13.3 gigawatts of announced projects at risk this morning. Another desk said 10 to 15 of the 43 gigawatts due online. When you hear gigawatts, guys, just translate it to “how much compute shows up in 2027.” Less than planned. A lot less.

Demand for that compute is parabolic. It doesn’t care who won. So the gap between what people need and what we can deliver got wider overnight, and the companies sitting on existing capacity are now sitting on something closer to gold in the ground. That’s why the big caps rallied. They own the mines.

Here’s where Musk comes back in. He stands up a data center in a third to half the time anyone else does. When turbine prices climbed, he started building his own. Most of his capacity is outside Texas already, which most people don’t know. And while the neoclouds sold their capacity forward for years because they couldn’t believe someone wanted to lock in that long, he’s been writing short contracts, because he expects the price of compute to keep rising. That’s the right call. I don’t think SpaceX pulls back much from here.

Which brings us to the futures. The CFTC comment period on AI compute futures closes November 9th, and the first contracts print six days from now. This is a global market, with capacity being built in Scandinavia and across Europe, and it’s going to be enormous, for the same reason every futures market gets enormous: a lot of people are mispricing the underlying and need a way to hedge it. $800 billion of capex is about to get its first hedge.

I want one pair on before that first print.

Long Quanta Services. Short Alphabet.

Quanta always gets paid. If we can’t permit new power, somebody has to squeeze more electrons out of what’s already standing, and that’s maintenance, upgrades, grid work. Quanta gets the call every time, and so does Bloom Energy on the generation side, where you push natural gas in and electrons come out with no three-year permitting fight. My squeeze play has nothing to do with a short squeeze. It’s about squeezing more out of a grid that just got harder to expand.

Alphabet is the other side because its high-margin business is cracking at the same moment its spending can’t stop. Search traffic is eroding, and some of that erosion is Google’s own AI overviews cannibalizing the clicks. Nobody goes to websites anymore. Your agent pulls from fifty sources and you never see an ad. Free cash flow went negative, and yesterday’s result makes the build-out more expensive, not less. Two things moving in opposite directions, and neither one points up.

Today the pair looks ugly. Quanta down, Alphabet up on the compute-mine rally. That’s the entry, not the reason to skip it.

One more thread. If the build-out slows, it takes some pressure off inflation and off yields. I have my doubts about how much, but even a little helps the stuff that’s been working outside of AI. Garrett’s been on IWM. I’d look at RSP as well, less yield exposure, same rotation.

Several of these plays don’t trigger until the first contract prints, and a couple sit further out than that. They’re all inside a window that’s done the same thing for 40 years: the index climbs after the midterm and the sectors the election touched carry the gains.

Want to play that and other opportunities in the next 60-days? Mark just released Cycle 11 to let you do just that.

He sends the first Cycle 11 trade Monday morning, strike, entry, stop and exit included, and the compute pair with its ratios goes through that book.

Grab it here before Monday if you want it.

Here for a good time… AND a long time,

Hans