A delayed return from a trip to Australia 🇦🇺, where a tech scene is booming, and here we are! The week ahead is punctuated by important earnings from Broadcom, Oracle and Synopsys, all right in the Epistrophy wheelhouse. The Fed and an ever-changing calendar of post-shutdown economic releases should make for some holiday treats as well.

We’re ready!

The full archive and Epistrophy research database is live here: https://epistrophy.beehiiv.com (your email address works as a password). Check it out.

As always, I’m focused on three things:
1) Technology-driven change;
2) the latest in innovation and startup trends, and;
3) stock fraud.

Companies Discussed

In This Note:

1. Companies Discussed

2. Oracle: Comfortably Plumb

   1. Oracle: New Kid On The AI Block

   2. Debt: Oracle’s Swiss Army Knife 

3. Tweet O’ The Week

4. Epistrophy In The News

5. 📆 of Epistrophy Events

6. Availability This Week

Oracle: Comfortably Plumb

Holding the line on leverage as Silicon Valley pines for zero-debt purity.

Three months after landing the biggest contract in the history of contracts. Oracle (ORCL: NASDAQ) has been treated as if scale demands scorn. 

With fiscal year Q1 earnings, the company reported $455 billion in remaining performance obligations, a 771% increase from the previous quarter, $300 billion of which: a reported deal with OpenAI.

The stock is now 9% down since then. 

All the more-established cloud providers were fighting for that OpenAI contract. But Amazon Web Services (AMZN: NASDAQ), Google Cloud (GOOG: NASDAQ) and Microsoft Azure (MSFT: NASDAQ) didn’t get the biggest gig ever. Oracle did. And yet a general fear of an “AI-bubble” made Oracle into the poster-child of a feared AI equity pandemic. Credit-default swaps widened, equity softened and the bubble analogy spread beyond Oracle with a central worry: debt is danger. 

In any other context the firm landing the central tenant of the AI era would have earned the valuation premium. Instead Oracle became a pariah. And yet, that business is unprecedented, but the tools that Oracle employed are not as unfamiliar to Oracle as it might seem. 

Below: a closer look at Oracle and the technologies it’s debt has enabled so far. 

(A reminder: this is not investing advice. We are Oracle shareholders, have worked with Oracle and much love for many things Oracle. We’re biased. As the Buddha said to Kālāma Sutta: “Don’t believe what I say.”)

Oracle: New Kid On The AI Block

At Oracle AI World in Las Vegas on October 15, 2025, new Oracle co-CEO Clay Magouyrk reminded us all how new the Oracle Cloud Infrastructure business is. “Joining Oracle in 2014 to work on building a new cloud was not an obvious decision,” he said to some laughter. “Many people, in a polite way, were asking: ‘Well, do you really need a fourth cloud provider? Is this really what you should be spending your time on?’”

The cloud incumbents spent a decade building for pre-AI traffic: distributed enterprise workloads, cloud migrations, content delivery and SaaS elasticity. By the time OCI was rolling out, AWS already had roughly a 10-year lead. Google Cloud had 8 years of public-cloud presence. Azure had about 6 years.

To train large models at scale Oracle needed GPU-dense regions, liquid and immersion cooling, higher-watt racks, lots of power, optical backhaul and unimaginable remote-direct-memory-access fabric. And it needed a lotta cash.

The Marshall Plan cost $13.3 billion in 1948 (about $179 billion in today’s dollars). The Federal-Aid Highway Act of 1956 cost $25 billion for the initial 13-year build of roughly 41,000 miles of interstate highways (about $285 billion in today’s dollars).

Oracle’s multiyear AI-infrastructure contracts are already at $455 billion. Yeah, that will take some debt, but a whole lot less than Oracle is used to. 

Debt: Oracle’s Swiss Army Knife 

Tech once walked through the shadow of Silicon Valley death, when its most-admired companies nearly bought it. And the Valley remembers.

Intel’s near-breaking point came in 1985, when a brutal DRAM price collapse forced it out of the memory business that had defined its birth. The company nearly failed. Apple’s followed a decade later in 1997, when horrible management decisions left it with just weeks of cash. Through desperate measures, these icons survived. But it left a mark on the Valley – it remembered that boom can become bust and cash reserves are the only anecdote.

The next iteration of big-tech, the likes of Facebook and Google, generated so much free cash flow they never had to consider debt to chase their aspirations. Their investors, accustomed to share price gains, didn’t demand dividends from the rising cash hordes. 

But Oracle has never been a stranger to debt. 

Through the late 1990s and early 2000s the company ran with essentially negligible borrowings, then levered up in lurches to fund hostile and strategic takeovers. 

Oracle’s total debt went from well under $1 billion to about $3 billion in 2005 as it absorbed PeopleSoft for $10.3 billion, to nearly $6 billion in 2006 after the $5.85 billion Siebel deal, then past $11 billion in 2008 alongside the $8.6 billion BEA Systems acquisition and to roughly $15 billion by 2010 as Oracle closed the $7.4 billion purchase of Sun Microsystems. 

By 2014–2016 Oracle had moved beyond “acquisition-only” leverage and into balance-sheet engineering, issuing $10 billion of long-term notes in 2014 and lifting total debt from about $18 billion in 2013 to $42 billion in 2015 and $44 billion in 2016, with explicit language that proceeds could be used for stock repurchases, dividends and future deals rather than just a single transaction. 

The NetSuite acquisition in 2016, at $9.3 billion, and a long run of buybacks pushed debt into the $55–60 billion range by the late 2010s, then another step-up into the low $70 billions around 2020 as Oracle leaned harder on cheap credit markets. 

The Cerner acquisition in 2022, at roughly $28.3 billion, pushed the model to its logical conclusion: by fiscal 2023–2025 total debt had climbed into the $90–105 billion band, making Oracle one of the most levered investment-grade names in large-cap tech and cementing a pattern in which the company routinely uses the bond market to fund both transformational M&A and aggressive capital returns. 

Throughout, Oracle has remained investment grade, but it has done so while steadily teaching the Street that, unlike the cash-rich cloud incumbents, it is perfectly willing to carry a utility-style debt load if it buys it time and scale.

In recent years, the debt financed physical capability: OCI Supercluster with GPU adjacency engineered for model training, Exadata X10M for memory-intensive loads, Cloud@Customer for sovereign AI, Alloy for partner-led region multiplication and the optical backbone that makes cross-region training possible. The leverage extended reach across three axes simultaneously: compute density, global region count and on-premises tenancy.

Oracle’s current debt-to-equity ratio sits below its pandemic-era peak of 88.9x, when equity compressed after the Cerner goodwill adjustment and total borrowings crested in the low–$100 billion band. The distortion in 2022–2023 came not from fresh leverage but from the denominator: equity fell because Oracle chose to mark Cerner’s legacy software honestly rather than nurse its book value. Once earnings rebuilt and buybacks slowed, equity expanded and the ratio eased. The optics were violent; the capital structure never was.

For context, Oracle has lived with a higher ratio before, and not during an AI build but during a conventional M&A decade. The 2014–2016 window, when NetSuite closed and Oracle issued $10 billion in long-dated notes to fund buybacks, produced a sharper leverage signal from a weaker strategic need. That period carried a higher debt load relative to equity while producing no commensurate infrastructure leap. Today’s ratio—still elevated by tech-sector standards—is lower than that mid-2010s peak and is underwriting architecture, not cosmetics.

The present structure therefore represents an inversion of the usual risk story. The balance sheet looks calmer precisely when Oracle is doing its most capital-intensive work. Debt funded GPUs, cooling, transport and tenant capacity, not dividends and photo-op repurchases. A decade ago the company borrowed to defend valuation; today it borrows to deliver compute. For an investor conditioned to fear leverage, the data cuts the other way: Oracle has carried a higher debt-to-equity ratio in quieter years than it carries now, in the first true infrastructure cycle of its corporate life.

Tweet O’ The Week

Epistrophy In The News

Just before Thanksgiving and my long trip, I was able to join NewsNation and Connell McShane to talk about what wonders Jony Ive and OpenAI might be creating with an AI device. Good fun to start the holiday season (no I’m not counting Halloween) with an upbeat notion of what AI can do for us. Check it out.

📆 of Epistrophy Events

Availability This Week

I’m back! In San Francisco all week (with a quick client visit to the Valley mid-week). Feel free to text for time-sensitive items; email works too, replies later in the day.

Written reports are available to clients, with video summaries on YouTube, and of course our popular summaries of the summaries on Instagram, TikTok, and YouTube Shorts.