CoreWeave, Inc.

But I think the challenge right now is that you could have AI contract at the margin, and there's a lot of fluff and a lot of froth and a lot of stocks out there.

And I think of, like, the CoreWeaves more than I think of the Oracles, but I think there's a l- there's, there's risk where even if at the margin things contract.

You know-

Listen, I think that's a valid point.

And the, and the point is, is here's how I look at it, is if you can, if you can spend your own money and build this out and get these huge rates of return, so like good ROIs, ROIC, return on invested capital, and you're just using cash flow, in general, you wanna be in those kind of businesses.

But when the good businesses, like the hyperscalers like Amazon, Facebook, Google, um, Oracle, CoreWeave, when they have to borrow massive amounts of money 'cause they don't have cash flow, that should be a little-

Yeah

... red flag to say, like, "Okay, we're over-financializing this right now." W- especially when you see the government have to come in and bail it out because there's just nobody, there's no more debt buyers for this stuff, and, and the, and it's, the, it's starting to look wobbly.

Opendoor down five.

Coreweave down five.

Circle down three.

MSU down three.

Here's somewhere closer." "I need this specific powerful compute," or, "I'm okay using a computer from three years ago," uh, et cetera, et cetera, you know? So those are the neo clouds of the world and, and the services they provide.

And then if you look at their earnings, and you can see Nebius recently just came out with earnings for CoreWeave.

I mean, they've been growing at a blistering pace, um, because of their de-- the, the demand.

So when you win the business, what, what sets you apart? Is it based on price or location or how fast you're up and running or, or what?

Now, why are hyperscalers keeping, uh, data centers off their balance sheet? Ultimately, Nvidia could, or you know, or Google or Meta could buy a warehouse, fill it with GPUs, and borrow at 50 basis points to fund it.

Why are they choosing to let, you know, CoreWeave do it, which funds at 500 bps? Uh, one of the reasons is that they don't want their leverage to increase, and if there is excess CapEx or, or lower demand, then, you know, it's easier to restructure an external SPV than to get downgraded on your own debt.

So I think this is one of the reasons.

But look, the real reason why we also started to look at, at this is that we saw that there were companies in the last five years that were going private on their bonds, and we saw that there was a pattern of companies with relatively low, um, uh, disclosure, so very often private companies already, uh, or red flags on their governance, um, high leverage, high spreads, and these companies were not able to fund in public markets.

But it's a very different animal.

So we could also see a slow motion crunch where you have the worst borrowers, the worst data centers that keep borrowing, that keep subordinated bondholders.

Think about CoreWeave.

They've issued first, uh, unsecured bonds, then they issued secure bonds.

Why does Google want to do this in the first place? Why do they have this joint venture? First is for TPU distribution.

NVIDIA has an enormous advantage because customers can rent NVIDIA GPUs from AWS, Azure, Google Cloud, CoreWeave, Nebius, Lambda, and dozens of other smaller providers.

Google's TPU ecosystem is much narrower.

So Crux is creating an independent TPU-focused NeoCloud model, giving customers another way to buy Google Compute besides going directly through the Google Cloud.

They booked 30 billion more of AI contracts in the quarter, and everyone hates the stock.

We're literally, o- of the four hyperscalers, including CoreWeave, Microsoft, Amazon, Google, everyone has said business is booming, margins are better.

Uh, the AI drain on, hey, it's, it's gonna be really expensive, there's no ROI, they, they threw that, uh, out the window.

So what you're seeing is, I think, the same thing from Oracle, which is they're one of the top five infrastructure builders for this cloud era, and they basically raised their EPS guide and reiterated, uh, the fiscal guide for over $90 billion.

This is a really good execution quarter, and that's important because Oracle does not get credit for its AI compute backlog.

It has the biggest backlog with Microsoft.

It has more backlog than Amazon and Google, and yet companies like CoreWeave and Nebius, upstarts without any profitability, are getting full credit for their backlog.

Oracle, at best, is getting a quarter for maybe a quarter of its backlog.

The state has raised concerns around water and air pollution.

The question really for Oracle, does management characterize this project as on schedule, and is this issue isolated or illustrative of a broader challenge facing the AI infrastructure build-out? Comments could have implications for other AI players like the Nebius and CoreWeave, the hyperscalers, Amazon and Microsoft, that are pouring billions into data center development.

Scott, this report does mark roughly one year since Oracle surprised the market with that $300 billion OpenAI deal, which sent shares up by 36% in one day.

Investors will want an update on its relationship with the AI lab as well as its latest financing plans, right, following that $20 billion equity issuance plan.