The nuclear-powered AI data-center trade roars into a new year

You can only shock the world so many times, and a thinner phone with a better camera isn’t always going to cut it.

That, in short, is why Apple has tended to go down on days when it’s introduced a new iPhone to the world, as this great chart from Bespoke Investment Group shows:

On average, the tech giant falls 0.4% on the release date and is negative more than 70% of the time, perhaps a useful tidbit on this, the day of the iPhone 17 launch.

One more thing....

A potentially complicating factor to the aforementioned data is that Apple has often done quite well in the six months leading up to a new iPhone announcement, roughly 5 percentage points better than its typical six-month return, as shown above. That’s not the case this time, with Apple shares up about 5% over the past six months compared to a typical near 20% advance in the prelude to a new iPhone drop.

So it’s not like expectations about how big of a catalyst this can be for the company are sky-high and due for a sharp retrenchment, especially given Apple’s relatively lackluster progress in developing AI capabilities relative to its megacap tech peers. But a seemingly low bar to clear hasn’t necessarily been a boon for the company on the big day, either.

In any event, staring too closely at the minutiae of all this may be missing the forest for the trees.

“While this info may be helpful to traders, we doubt it’s something that long-term shareholders are too worried about given the huge compounding returns the stock has provided during the iPhone era,” Bespoke wrote.

CoreWeave is soaring as Microsoft’s $17.4 billion deal with Nebius shows the immense value and continued demand for all parts of the AI data center ecosystem.

One additional reason for CoreWeave’s jump may be that its pending acquisition of AI data center infrastructure company Core Scientific looks like a great deal compared to Microsoft’s renting of (more broad and advanced) AI data center capacity from Nebius.

CoreWeave’s all-stock deal to buy Core Scientific was initially valued at ~$9 billion, but with the subsequent decline in its shares, it’s worth about 40% less. And in purchasing Core Scientific, CoreWeave is saving $10 billion in what it would have paid the company to lease data center infrastructure over the next 12 years.

As it stands, Microsoft is getting about 300 megawatts in data center power capacity from Nebius, while Core Scientific boasts that its footprint is in excess of 1,300 megawatts. So, on the surface, it looks like an absolute steal for CoreWeave.

But again, this is not an apples-to-apples comparison; not all access to AI computing infrastructure is created equal.

There are differences in the type of AI infrastructure provided by the two: Nebius owns GPUs, while Core Scientific doesn’t, and what it provides in the software layer isn’t offered by Core Scientific as a stand-alone entity. This is the difference between the “full stack” approach (Nebius) and a “colocation” approach (Core Scientific).

That being said, CoreWeave’s acquisition of Core Scientific, once completed, will make the combined entity’s business model look more like Nebius’ model, which, as Microsoft just told us, is something that top hyperscalers are willing to pay a pretty penny for.