Oracle can’t win for losin’ all of a sudden.
Coming off their worst week since 2018 — and their second-worst week since the financial crisis — the company’s shares needed another doubting sources story like Nicolas Maduro needed a US naval blockade, which is to say the stock really could’ve done without a Financial Times report suggesting Oracle’s main data center financing partner pulled out of a deal to back a construction project near Ann Arbor.
Here’s how this works. Alternative asset firm Blue Owl Capital puts up cash and lines up construction loans for data centers which it then owns on the understanding that Oracle will ink long-term leases for the facilities. Oracle then sells the compute produced inside those facilities to its own customers, OpenAI being the largest.
As noted here last week, OpenAI accounts for something like 55 cents out of every dollar of Oracle’s backlog. That’d be a lot of concentration risk even if OpenAI had the revenue to make good on its commitments. Currently, it doesn’t.
So, Oracle’s buildout looks speculative on at least two fronts. First, there’s no guarantee demand for AI will pan out in line with optimistic projections. Second, even if demand does pan out, there’s no guarantee OpenAI will capture and monetize enough of that demand to satisfy its $300 billion deal with Oracle.
The market’s wise to those (and myriad other) risks and so are banks and financing partners. The cost of hedging exposure to Oracle’s debt has exploded in recent months.
There’s the chart again. It’s updated. Oracle’s CDS is out beyond 150bps.
According to the FT, Blue Owl got cold feet on the Michigan project because lenders wanted “stricter leasing and debt terms” in line with suddenly skeptical markets. The more onerous terms made the project look less enticing than earlier joint builds with Oracle.
Without a Blue Owl, Oracle appeared to need a white knight to rescue the $10 billion project. The FT mentioned Blackstone, but said nothing was finalized.
Later, a spokesperson suggested the FT (perhaps inadvertently) mischaracterized the situation. Discussions around financing for the Michigan data center are “on schedule,” Oracle said, adding that its co-developer chose “the best equity partner from a competitive group of options, which in this instance was not Blue Owl.”
Even if investors were buying the excuse, they weren’t buying the stock which was down another 5% headed into the cash close on Wednesday. The decline erased, and then some, Tuesday’s nascent rebound.
As the figure shows, Oracle’s now in a bear market within a bear market: The shares have fallen 19.7% in the short space of a week, and they were already down 33% from their September highs.
I don’t want to be an alarmist, but they gotta figure out a way to stanch the bleeding. This is too much. Pretty soon, people are going to start comparing Oracle’s AI data center plans to MSTR’s Bitcoin reserve strategy.
Wednesday’s FT report came just three business days after Bloomberg said worker and materials shortages may mean delays for other Oracle data centers. (Oracle denied that reporting too.)
Even if the company does secure financing for the Michigan project (and I assume they will), there’s doubtlessly some truth to the implication from the FT article — namely that participating lenders are going to want stricter covenants to participate in any arrangement that involves Oracle, at least until the market calms down.
Remember what I said last week: “Oracle doesn’t enjoy the same sort of financial heft and cushion as the other hyper-scalers, all of which boast balance sheets generally viewed as impregnable.”
In their Wednesday piece, the FT quoted a person familiar with Oracle’s deals as saying lenders are imposing harsher terms on the company compared to the likes of Amazon and Microsoft because they’d “rather work with a hyper-scaler that has more experience and a less speculative project pipeline.”
Could be time to catch a falling knife..Oracle has close ties to the current administration and i would be shocked if DT lets his buddy fail. Some more crony capitalism headed our way.
My rule of thumb for blue-chips says a drawdown of 60-75% deserves a look, almost regardless of the circumstances unless of course there’s some reason to believe the going concern question’s relevant. Take that for whatever it’s worth.
Ha! I was about to post a comment on your very prescient piece from last week. Nicely done!
@Sophist – alas, you may be right. Remember when Republicans were seen as business-friendly hands-off free marketeers? Along with great stewards of government spending?
It’s hard to compete with the megacap hyperscalers, or become more like them. Who knew?
H-Man, this is a poker game where Oracle has declared it is “all in”. If it works, Oracle today is very cheap. If the AI boom is a bust, Oracle looks very expensive. Pick one or the other, and if you have the stomach, give it a ride. On that same note, who is going to provide energy for all of these data centers? That may be a better investment.
XLU is down about 8% in the past two months, so the market doesn’t think so at the moment…
I think it quite possible that AI can still be the next big thing and orcl be a bad investment. Amazon and Google don’t need Oracle.
I live in Ann arbor. I use AI constantly for work, but don’t want a data center increasing my energy bill, ruining my freshwater. There is a lot of grassroots resistance against data centers. It’s a k shaped world babay
In retrospect, Oracle was such an obvious short at its 52 week high at $345. I missed that obvious opportunity. Just like NVDA eliminated any potential China sales from its future revenue projections, ….NVIDIA, and AVGO need to eliminate any potential sales to Open AI and Oracle from their revenue projections unless Open AI or Oracle have already paid. And also announce they don’t have any significant accounts receivable because anything sold to them is only on an immediate cash basis only, from now on. And then publicly announce they have done so ASAP because NVIDIA and AVGO stock are suffering from the guilt by association with Oracle and Open AI. Oracle is uninvestable as long as they have over $100 billion in debt and negative cash flow for the next 2-3+ yrs. The overall accelerated computing/AI industry is doing well as demonstrated by tonight’s astonishing results from Micron.
In retrospect, Larry should have sold $50 billion of his personal Oracle stock when it was $345/share, …and used that $50 billion to pay off half of Oracle’s debt. His personal Oracle stock in the past 3 months, has lost much more than $50 billion of value. Also, Safra Catz resigning at the peak, the same week of Oracle stock reaching 52 week high of $345 was the TELL, that foretold this sordid debt debacle.