Hey Income Traders,

Thursday took six minutes to erase the entire post-election rally, and then it kept going. The Nasdaq dropped 5.5%, the Nasdaq 100 lost 6.3%, and roughly $2.5 trillion in market value disappeared in one session. Both indexes had their worst day since the tariff panic in April of last year. That’s a different animal than the dips we’ve gotten used to over the past couple of years. This was indiscriminate, position-clearing selling, with levered longs getting the tap on the shoulder and dumping whatever they could. It reset a lot of crowded trades in a hurry.

Underneath it, the story driving the biggest names hasn’t broken. Anyone holding compute capacity still gets paid for it. Owning compute right now is close to owning gold.

It’s scarce, it’s in demand, and it commands a premium, even though meeting that demand means building more of it at steadily higher cost.

For the hyperscalers, that math works. Amazon, Alphabet, Microsoft and Meta are guiding to somewhere around $730 billion of capex this year, and they can fund it out of cash flow and investment-grade paper. Rising build costs are an annoyance to them.

The neoclouds don’t get that luxury. Their model runs on borrowed money, and the credit market has started saying no. CoreWeave is carrying more than $51 billion in debt and obligations, its 2031 bonds now yield north of 15%, and the cost of insuring that debt against default sits above 1,100 basis points. Nvidia’s, for comparison, is around 139. CoreWeave, Nebius and Iren each fell more than 30% in October before this week even started. Capital is now the binding constraint, more than chips and more than power alone, and it’s sorting the players who can fund the buildout from the ones who’d like to.

Permitting and power add another layer, and Tuesday poured concrete on it.

Voters took up 41 local ballot measures on data centers across ten states and limited development in 40 of them. Texas, on track to become the biggest data center market in the world, just elected a governor who campaigned on a moratorium.

That’s the real reason Thursday happened, and I get why people read it as the buildout hitting a wall.

I read it differently.

Tighter rules on new construction and grid hookups make the compute that already exists worth more, and they make building new compute on borrowed money harder.

Both of those favor the operators with the deepest pockets and the longest runway.

OpenAI already blinked, announcing Thursday it’ll pay for its own electricity and infrastructure at every site and sign a local compact with each host community. The jobs-and-growth pitch for the buildout will come, and the rhetoric may thaw, but I’m not waiting on it.

For now, the setup rewards discipline.

Respect the flush and don’t fight it. The compute theme at the top of the quality ladder is intact, and the financing squeeze is doing real work under the surface. The names that come out ahead are the ones that can keep paying for the buildout while weaker balance sheets get forced to the sidelines.

The flush clears the weak hands. It doesn’t change who can pay for what comes next.

We’re going to have ways to play that in Cycle 11, which is a 2-month trading package Mark Sebastian put together to play the 60-days left until the new Congress is seated.

AI-based trades are inevitable.

Tap this link to join,

Here for a good time… AND a long time,

Hans