Oracle 估值偏低（PEG 为 0.65），资产负债表有所改善；押注 Oracle 实质上就是押注 OpenAI 的偿付能力，但交易的经济性尚可。

Oracle just sent a massive warning to SpaceX. And Oracle earnings just gave us a big middle finger to that underwriting.

Oracle OpenAI sound signs 300 billion-dollar deal. But, wait a minute. For how many gigawatts is it? Because if we take a 300 billion-dollar deal and divide it by the gigawatts, divide it by how many years it is, then we're going to find out how close it is to 50.

Oracle contract will require 4.5 gigawatts of power. 300 billion dollars divided by 4.5 gigawatts. That's 66 billion per gigawatt. Fine, but for how many years? Because I want per year. Oh, there it is. For roughly 5 years. Damn it. Divided by five, that number is basically in line with the market comparables. See that? That's 13.3, roughly, billion dollars per gigawatt.

It's really actually good for Oracle. It is though also a gamble, and it's not just me saying it. Literally, the Wall Street Journal uses the word, and it makes sense. You are riding on OpenAI.

The OpenAI and Oracle contract, which starts in 2027, is a risky gamble for both companies. Uh and uh this is uh or Oracle is expected to have to take on debt to actually build this out.

So, let's go look at the Oracle financials and see what we have here.

All right, Oracle, show me your pee-pee. Oracle's pee-pee, damn, still uncertain. Why? It says here that we had revenue growth. Revenue growth was actually pretty decent. We increased revenue by 62% year-over-year.

11.6 billion dollars, 62% growth from last year. Fantastic. Software declined 3%. In fact, uh in in a later piece, you'll see that their licensing revenue fell even more. So, software is getting hit.

And not only is software getting hit, but they're just cutting jobs. Look at this. Services down 4% in expenses, presumably jobs. 12% down in sales and marketing, presumably jobs. R&D down 4%, presumably jobs.

We have a company that is increasing revenues, but its costs of those revenues are actually growing faster than their revenues are. Revenues are growing 62%, costs are growing 77%.

So, we're not yet at the point where we're really growing PP. So, PP isn't growing well, although it looks like they did because their operating expenses are only up 18%, but that's only because they cut jobs.

So, it's a little misleading what we're getting over here. That said, I'm not trying to be bearish Oracle. There's actually some stuff in here that I kind of liked.

Look at their cash flow statement. They actually repaid 4.2 billion dollars of debt. That's nice. In fairness, they issued basically 20 billion dollars of stock.

Now, is it possible they've bottomed out on the stock because they issued 20 billion of stock? Maybe. Because their balance sheet is actually not as horrible as it used to be.

They did spend more than they made, so they had negative free cash flow of like 5.4 billion dollars. It's not great, So their cash has grown to about 36.3 billion dollars. They do have enough cash to pay their bills because they finally raised enough money. They have 36 billion dollars in cash and they have 33 billion dollars in bills. So they finally have enough money to pay their bills.

They've got about 35% of their company's market cap in long-term debt and this is going to go up even more to go finance their open AI contract. So a bet on Oracle is really a bet that open AI is going to pay their bills.

So far I would say there's no sign that they're going to roll off a cliff yet. I think the bigger learning lesson from this isn't that if you're betting on Oracle you're betting on open AI. That's okay. Like that could be a reasonable bet because we could watch what happens with the Anthropic IPO. We could watch what happens quarter over quarter with Anthropic earnings and their margins. Are they starting to fall off a cliff because of you know commoditization from Chinese open weight models or US open weight models. We'll see. The earnings will tell us that. So you'll kind of have a little canary in the coal mine for Oracle.

I think the bigger issue here is that Oracle is signing deals on long-term contracts for like 13 to 15 billion dollars per gigawatt which that puts you closer to a three to four year payback period on the data centers. That's not bad. That's still decent. You're still getting an ROI. If you could pay back a data center in even five years, you're now making money on it.

Focusing though on Oracle, this is not as bad as it used to be. And if you actually look at the valuation of this company, it's dirt cheap. Yeah, this is where we saw license revenues down 14.5%. Oracle revenue in Q2 is expected to be up 30 to 34%. That's the guidance. They've got They're trading for about 20 times earnings, and their forecast growth rate before the opening ideal was 30%. So, we're trading for a 0.65 PEG right now. It's actually trading dirt cheap right now. That's That's actually not bad.

So, I got to give it some credit. This is good for hardware. Uh, it's cleaned itself up. It's signing the right deals. It's firing people where they're losing money, and they're investing more money where they're making more money. They're doing everything they should.

Ironically, I actually like Oracle better here than Restoration Hardware. And I think Oracle's giving us really good transparency on what deal pricing actually should look like for companies like the data center place overall.