Scooby Doo & The Case Of The Mysterious Oracle CDS

Through the lens of Appraisal Rights

Over the past few months, Oracle’s Credit Default Swap (“CDS”) spreads have widened noticeably—from 35bps to 57bps—over the past several weeks. There’s no obvious news event, rating downgrade, or default scare to point to. Someone —maybe several someones—wants insurance. The question is who, and why. Most commentators point to the ballooning debt on Oracle’s balance sheet and its exposure to the AI bubble. In a prior life, I spent decades in the obscure trenches of appraisal rights arbitrage, which is perhaps why the pattern caught my attention—and also, perhaps, why I have a dissident interpretation of Oracle’s CDS.

Appraisal rights allow shareholders to dissent from a merger, reject the deal price, and ask a Delaware judge to independently determine what their stock was really worth, regardless of the deal price. You file your petition the day before closing, when the market has already validated the deal. In the early decades, at least, you were buying a very inexpensive option on being right about something everyone else thought was a settled matter, and your downside was mathematically defined and limited.

Appraisal rights arbitrage is highly technical—and not particularly glamorous—work. It involves litigating valuation models and auction mechanics, searching for hidden management conflicts or deal process flaws that might have depressed the price. But it does teach one a great deal about risk and uncertainty, and what hidden signals might be lurking in market prices like CDS that aren’t on the radar of most observers—because in appraisal rights arbitrage, you might be the one contributing to those signals yourself in a way that is inscrutable to others.

Appraisal arbitrage mechanics taught me lessons about hedging that apply well beyond Delaware, and are perhaps relevant to the mystery of the Oracle CDS. A pending appraisal claim represents unsecured exposure to a leveraged buyer, and the trade timeline may run as long as two or three years. In cases where the exposure was large enough to matter, I’d buy credit default swaps. It was an elegant way to neutralize “gotcha” risks—unseen “black swans” that could ruin an otherwise-attractive thesis. The statutory interest income—Delaware paid a floating rate that was often 7% or higher—covered most of the CDS premium. If the case dragged on, I’d roll the protection forward. If the buyer’s credit deteriorated, the CDS paid out exactly when my appraisal recovery was most at risk.

That experience stuck with me because it revealed that every intelligently structured appraisal, merger arb, or long-tail trade should isolate the core thesis and hedging against unforeseen outcomes that could negatively impact it whenever feasible. The tools may change—Delaware courts in one decade, CDS in another—but in every case you are attempting to protect yourself against “unknown unknowns” and from the possibility that you, the world, or management, or the market has missed something important.

That’s the mindset I bring to Oracle’s CDS spreads. I’m trying to understand who decided to buy protection and why they decided to buy it now.

The most obvious explanation is that Oracle has been leveraging itself to the hilt and that the market is worried about the AI bubble. Oracle is positioning itself as critical AI infrastructure, and the hyperscalers like OpenAI need capacity they can’t build fast enough themselves. Oracle typically agrees to provide massive data-center capacity, the counterparty commits to years of compute spend, and—most relevant to the CDS—Oracle must borrow billions to build out the infrastructure to deliver it. Of late, Oracle has signed increasingly large contracts for AI cloud capacity—first $1.5 billion with OpenAI, then multi-billion dollar arrangements with xAI, AWS, and Google.

But—while we don’t pretend to be AI industry experts, and couldn’t even play one on TV (or FinTwit)—it seems to us that Oracle’s position within this AI ecosystem is fundamentally different from, and perhaps more precarious than, everyone else’s in one important way that could negatively impact its CDS. Microsoft, Google, and Amazon exist in a (for now, anyway) self-reinforcing loop—they build (or invest in) frontier AI models that drive cloud demand that in turn justifies more infrastructure that trains better models. This is (somewhat) self-hedging. If AI spending slows, they’re simultaneously buyer and seller, landlord and tenant. Their credit risk is, in the immortal words of The Bernank, contained within their own ecosystem.

Oracle, however, lacks that natural hedge. The round-tripping that appears circular for the hyperscalers is one-dimensional for Oracle. They don’t have their own frontier AI models—or, as far as we can tell, investments in OpenAI, Claude, or other top contenders—to close the loop. They’re taking on leverage to build capacity for customers whose own demand is contingent on the continued enthusiasm of another layer of customers. It’s vendor financing, but Oracle’s not financing its own ecosystem—it’s financing someone else’s. If the music were to ever stop, Microsoft could slow Azure buildouts and cloud spending simultaneously. Oracle will be saddled with data centers of dubious worth and debt.

The scale keeps escalating. What started as partnerships are now effectively structured finance with equity kickers and revenue-sharing clauses that blur the line between customer and creditor. Oracle’s debt has grown accordingly—each deal requires capital upfront, and the payback runs over years. The bull case is that Oracle becomes indispensable infrastructure and the cash flows materialize. The bear case is that Oracle is the only major player in this trade without the natural hedge of owning both sides. Oracle faces different credit risk than Microsoft, Google, or Amazon. While Oracle is part of the AI feedback loop, they only represent one side of the equation and are unable to close it themselves. The market could simply be pricing in that structural asymmetry.

The straightforward, mechanical explanation could be that banks are hedging their exposures to Oracle’s debt. The banks that arrange Oracle’s financing—J.P. Morgan, Goldman, Citi—also hedge it. The billions that Oracle is spending on data centers and AI infrastructure are borrowed. The banks involved arrange Oracle’s financing, warehouse the loans, syndicate them, and sometimes wrap them in CDS protection before distributing them to yield-hungry funds and other buyers. It’s routine balance-sheet management—the bank maintains the client relationship, the risk gets hedge or passed along, and everyone earns fees. Spread widening could merely reflect prudent hedging, not any sort of judgment on Oracle’s solvency.

But here’s what piqued my attention: if this were simply routine hedging or the market repricing Oracle’s structural position and debt loads, you’d likely expect the spread widening to track Oracle’s debt issuance calendar or coincide with new contract announcements. It doesn’t appear to do so. Someone is buying protection before they need it. That’s the same pattern I learned to watch for in appraisal: protection bought early, when it’s cheap, by actors who may have private information or private exposure the market isn’t yet aware of. Buying CDS on Oracle may be conceptually similar to filing your appraisal petition the day before deal closing, before anyone knows if the deal price was fair or not.

Perhaps the banks’ exposure runs deeper than loan books. J.P. Morgan isn’t merely a banker to Oracle—they’re likely operationally enmeshed with and dependent on Oracle’s software architecture, as well. The banks’ systems plug directly into Oracle Fusion ERP, NetSuite, and Simphony. They’ve built joint solutions: real-time payments inside Oracle Fusion, treasury tools inside NetSuite, supply-chain finance and account validation living natively within Oracle software. J.P. Morgan’s banking rails literally run on Oracle.

Think about what that means if Oracle stumbles—perhaps taking its eye off the ball as a result of its new debt load or some other factor. Software updates slow. Integration support thins. Mutual clients running payments through Oracle Fusion with J.P. Morgan rails underneath start seeing latency, then errors, then escalations to account teams who don’t have answers. For a bank integrated that deeply, the problem isn’t Oracle’s solvency—it’s operational continuity. If Oracle gets wobbly, J.P. Morgan needs to rip out those integrations and rebuild them elsewhere, fast, without explaining to their board why they didn’t see it coming.

That could be part of what a CDS position hedges. Not credit risk in the traditional sense, but the financial capacity to fix the mess quickly—to pay for emergency migration, hiring the developers and consultants, absorbing the disruption without it becoming an earnings problem. The protection buying happens in credit markets, but the risk being hedged could be partially operational.

And if J.P. Morgan is thinking this way, it’s possible that Oracle’s enterprise customers are thinking along the same lines. Fortune 500 companies have welded Oracle ERP into their operational nervous systems. Cloud workloads can migrate; ERP cannot, at least not without years of retraining, recoding, and internal political warfare. Every invoice, every journal entry, every close process lives in there. If Oracle’s debt became a distraction—not a crisis, just an overhang that slowed development or triggered management turnover—those customers would be among the casualties.

For a large enterprise CFO, buying CDS on Oracle—or having their bank embed that protection into a financing wrapper—could help insure that you have the financial firepower to ensure your own operational continuity. The payout could at least partially fund your exit from the Oracle ecosystem—and a commensurate hit to your earnings power—should it ever come to that. You’re acting behind the scenes because the last thing you want is Oracle asking why you’re hedging your exposure to them.

This version of the CDS explanation is perhaps the most interesting, as it could partially explain both the timing and the discretion. The protection isn’t necessarily being bought by people who think Oracle will default—it’s being bought by people who can’t afford to be surprised one way or the other.

There are other possibilities, as well. A REIT financing Oracle-anchored data capacity—say, a $2 billion lease commitment tied to Oracle Cloud expansion—faces genuine concentration risk. If Oracle slows its buildout or renegotiates lease terms, the REITs cash flows could gap down immediately. Buying CDS on Oracle’s debt is an obvious hedge: the protection pays out exactly when lease income disappoints. Maybe several REITs are doing this simultaneously, each with $50 or $100 million in notional exposure. Not enough to move markets individually, but collectively it adds up.

There’s a more mischievous possibility. Maybe a (offensive minded and/or geopolitically disruptive) competitor—AWS, Azure, Google—decided to seed doubt because a little market anxiety would help tilt future contract negotiations. Buy enough CDS to move the spread a few ticks, not enough to trigger regulatory reporting. Route it through several prime brokers to obscure the origin. The out-of-pocket cost is small. The benefit is reflexivity: Oracle suddenly looks a bit more levered in every credit presentation, every vendor diligence call, every CFO’s quarterly review. When your sales team walks into the next bake-off against Oracle, there’s a nagging question in the buyer’s mind about Oracle’s long-term stability. Heck, the position itself might even be profitable if the spread keeps widening.

There’s precedent for using credit instruments in this way to manage perception. In the mid-2000s, bond insurer MBIA allegedly (according to @billackman and others, if we remember correctly) put on certain trades that would have had the effect of making their financial position appear stronger than it. The mechanism for Oracle would run in reverse—using credit instruments to suggest fragility. I don’t know if this is happening, but the cost is likely trivial relative to a cloud contract worth hundreds of millions, and the incentives are obvious. If I were running competitive strategy at Azure, I’d at least model it.

Finally, macro funds have also been circling like vultures around the “AI bubble” more broadly, the thesis boiling down to too much leverage chasing too much hype with too little margin for error. Buying CDS on Oracle, AWS’s debt, or the high-yield tranches of data-center CLOs is one way to own cheap convexity on that unwind. If the trade works, you make multiples on a small premium. If it doesn’t, you’re out basis points.

The reality is that all of these explanations can be true simultaneously. Banks hedging loan exposure, operational partners insuring continuity, customers protecting against distraction, REITs offsetting concentration, macro funds trading themes, competitors nudging perception in an effort to set off a reflexive spiral. Like all prices, CDS is where every motive collapses into a single price. The market price doesn’t distinguish between hedging and speculation, between prudence and positioning. It just sees demand for protection relative to supply, and spreads adjust accordingly.’

The interesting question is less about whether Oracle’s debt is sustainable—the company generates enormous cash flow, the debt is manageable for now, and for all we know the AI capex bet might pay off spectacularly. What’s of greater interest to me is why multiple sophisticated actors, for different reasons known only to them, decided to hedge or speculate on Oracle CDS now. Speculators, or individuals deep within Oracle’s network—lenders, operational partners, enterprise customers, maybe competitors—with detailed knowledge of the exposure are buying insurance while it’s still cheap. Pure speculation on our part, but we wouldn’t put it past Michael Burry or Masa Son (or a stealth interested player with their own agenda such as China) to put on protection trades for themselves.

The price of protection is the market’s aggregation of private doubts. When it moves without obvious catalyst, that could be an interesting signal. Someone who can’t afford to be surprised might be hedging risks.

CDS is just protection against the tail. And right now, the tail is cheap. The price of protection rises, social media and journalists notice, and the narrative follows. The market starts treating the CDS move as evidence of concern, even if it might have originated as simple risk management, and reflexivity and momentum continued the trend. In the credit world, protection buying and storytelling are indistinguishable. The insurance can reflexively become the signal.

If Oracle adjusts—slows the capex, preserves cash flow, proves the skeptics wrong—the protection expires worthless and everyone moves on. But if the hedgers and speculators are right, they’ll have been paid for being early. By then, of course, the spreads will have moved, the market will have listened, and the window will have closed.

Just like in appraisal arbitrage, you isolate your thesis and hedge the known risks and unknown unknown risks. In appraisal, you file your petition when the market has already validated the deal price. In credit, you buy protection before consensus the market flashes risk warning signs. Both are expressions of skepticism about the probabilities and expected values priced into the market.

If this piece meant something to you, share it with someone who could benefit.

I’ll be writing more on governance, activism, dark arts trades, and investment process. I’ve lived and breathed these for years. I’ve also written about tokenization, Formula One, the NBA, and sometimes just tell a story worth remembering. Stay tuned.

You can also find me on Twitter at @MrMojoRisinX and @BeWaterltd.