Last week, Oracle redrew the map of AI infrastructure with an order book swollen by OpenAI’s commitments. This week offers no earnings to parse and my spreadsheets get a break — my clients not so much: I’ll be meeting with investors and taking on some company visits in Silicon Valley.

Hey, take a look at our updated website. It features an ever growing repository of past notes and our searchable research database at epistrophy.beehiiv.com.

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:

• Companies Discussed

• Oracle’s RPO Tsunami

  • FY2026 Q1 Earnings Blow The Roof Off

• Tweet O’ The Week (twofer!)

• Epistrophy In The News

• 📆 of Epistrophy Events

• Availability This Week

Oracle’s RPO Tsunami

FY2026 Q1 Earnings Blow The Roof Off

Oracle (ORCL: NYSE) just delivered a quarterly earnings haymaker that nearly defied language and—by Larry Ellison’s reckoning—heralded “a seismic shift in computing.” In a climate where financial punditry routinely misproduce the same handful of exhausted metaphors—“blowout,” “breakout,” “reshuffle”—this quarter’s $455 billion in remaining performance obligations (RPOs), the single largest recorded leap in the company’s history, deserves its own taxonomy. 

Why care now? Microsoft (MSFT: NASDAQ), Amazon (AMZN: NASDAQ) and Google (GOOGL: NASDAQ) have long sparred for the public cloud crown, with Google panting a distant third. Oracle was cast as old tech—legacy code for legacy customers. That theory died this quarter, quietly euthanized by a cohort of AI heavyweights, including AMD (AMD: NASDAQ), Meta (META: NASDAQ), NVIDIA (NVDA: NASDAQ) and xAI (private) each signing “significant” contracts for Oracle’s cloud muscle, and the mother of all contracts a reported ~$300 billion deal with OpenAI (private).

“Oracle has become the go-to place for AI workloads,” CEO Safra Catz said on the Q1 2026 call, almost underselling the force of the transformation. Those words, spoken with the simple severity of a statistician watching a tidal wave register on a seismograph, should not be ignored.

Here’s the technical heart: Oracle’s critical leap resides not merely in storage, compute or even AI training capability, but in the industrialization of data center infrastructure at scale. For a decade, Oracle invested compulsively in custom hardware, special-purpose networking and vertically-integrated stacks—choices that once looked anachronistic compared to the developer-friendly, mix-and-match ethos of AWS or Google Cloud. Now those bets are maturing. The result? Oracle’s gigawatt-scale data centers are rapidly emerging as the infrastructure of record for AI’s largest workloads. With $455 billion in RPO—up 359 percent year over year and a dramatic $317 billion in a single quarter—the technical roots run deep: fast, specialized networking, rapid customer onboarding and a volume-based equipment strategy that skirts the old cloud dogmas. RPOs are now 771% larger than Oracle’s sales in the last year.

Oracle’s market positioning is now as much about physics as software. Clients want inferencing capacity, not just the ability to train language models in the public cloud, and Oracle’s dense network design enables outsized speed and cost-efficiency. “If we can move data faster than the other people, if we have advantages in our GPU superclusters, if we’re twice as fast we’re half the cost,” Ellison said on the call. Oracle’s multi-cloud capabilities—deploying Oracle Database and AI services directly inside Azure, AWS and Google Cloud—delivered 1,529 percent annual growth this quarter, a figure so implausible it required reiteration by both Catz and Ellison to verify its provenance.

For the layperson, Oracle is still a database company—think Oracle Database, Oracle Autonomous Database and Oracle Exadata. For the sophisticated, the company’s emergence as a data center superpower is the real headline. Over the last decade, Oracle’s pivot has been methodical: ten years of streamlining Oracle Cloud Infrastructure (OCI), modernizing with fresh deployments of Fusion Cloud ERP, SCM and HCM, scaling Oracle Cloud@Customer, advancing the AI Database, and enhancing services such as MySQL HeatWave, Kubernetes Engine and Cloud Integration Services. Hardware tangents were phased out while investments poured into specialized networking—especially OCI’s Gen2 architecture, which isolates customer networks at the physical level and provides dedicated, high-throughput, low-latency fabrics for each client. This eliminates noisy neighbor effects and ensures workloads don’t compete for bandwidth, making performance not only more consistent but also more predictable—a critical requirement for AI training, inferencing and high-performance HPC deployments. With features like Virtual Cloud Networks (VCN), Dynamic Routing Gateways, FastConnect and native support for distributed superclusters with over 32,000 GPUs, Oracle engineered onboarding processes that connect hyperscalers and large enterprise clients in days, not months. Control plane/data plane separation further improves uptime and security, while the ability to rapidly spin up dedicated, private environments for regulated or latency-sensitive workloads has become a key differentiator in winning contracts with both cloud-native disruptors and risk-averse incumbents.

Among the client wins called out this quarter:

• Nokia (NOK: NYSE) tapped Oracle Database@Google Cloud to overhaul data center strategy and optimize licensing.
  

• Cohere (private) deepened its OCI Generative AI partnership, leveraging OCI for scale, reach and custom enterprise AI models.
  

• Tensor (private), chasing Level 4 Robocar autonomy, selected OCI for its explosive AI training needs.
  

• Coronis Health (private) turned to Oracle to modernize virtual desktop infrastructure—expecting a 35% cost reduction.

These wins reflect Oracle’s ability to serve everyone from global telecoms and AI startups to healthcare and infrastructure.

The financials reveal an operation ramping into new territory. Q1 revenue rose 12 percent year-over-year to $14.9 billion, with cloud infrastructure revenue up a category-defying 55 percent. Gross operating income grew to $6.2 billion (non-GAAP), though GAAP net income held steady at $2.9 billion. The company’s non-GAAP EPS of $1.47 beat last year by 6 percent. CapEx ballooned, but with stated intention: to convert the “large RPO backlog into accelerating revenue and profit growth.” Importantly, Catz’s guidance for Q2 projects revenue growth of up to 16 percent and cloud revenue acceleration of up to 37 percent.

Here’s a snapshot comparison of key metrics:

The impact is systemic. Oracle’s role as custodian of “millions of databases” positions the company uniquely for the age of generative AI. Security, speed and locality—three properties underappreciated in hype-driven cloud evangelism—have become paramount now that CEOs and heads of government, says Ellison’are demanding inferencing capacity atop their most private data. 

What does it mean? For one, the inferencing market—running AI models in production, automating workflows, “placing bets in financial markets, automating legal processes”—will dwarf the already multitrillion-dollar training opportunity, Ellison said. Oracle’s newly announced AI database is a bid to make this “ChatGPT-for-the-enterprise” moment real, allowing customers to vectorize private data then ask LLMs any business question—a function other clouds cannot yet offer with comparable privacy guarantees. The technical vision is immense but not ethereal; the infrastructure is being delivered, consumed and (crucially) paid for.

Skepticism is still in order, particularly around sustainability of CapEx, potential for customer churn if workloads shift, and the perennial risk of hyperscaler partnerships souring. Investors who blanch at negative free cash flow are not wrong to spot the risk, but the directional signals—the immediate conversion from CapEx to revenue, the speed of customer ramp, the repeatability of success across both public and “dedicated region” cloud models—are, for now, working as promised.

We expect Oracle to sign more multi-billion dollar contracts in coming quarters. RPO is likely to exceed half-a-trillion dollars. Cloud infrastructure revenue—already projected to rise 77 percent to $18 billion this fiscal year—may hit $144 billion in four years, much of it already booked. 

It’s all about getting these RPO converted to revenues and cash flow, and fast. “As we bring more capacity online, we will convert the large RPO backlog into accelerating revenue and profit growth,” Catz said, projecting not hope but conviction. In an industry defined by vapor and velocity, that’s the signal to watch.

Tweet O’ The Week (twofer!)

Epistrophy In The News

On NewsNation with Connell McShane, I was live as Oracle’s numbers hit the wire, saying the AI data center game changed in real time—a viewer noted I “looked shook.” The segment captured the scale of the moment.

Later in the week, Yahoo Finance’s Josh Lipton asked me to parse what could still go wrong for Oracle, even as its dealbook swells. Within the hour, they had me back on to dissect Adobe’s results—two very different companies, one question: how much of tech’s future is being pulled forward by AI demand?

📆 of Epistrophy Events

Availability This Week

Available throughout the week, in an unusually hot San Francisco, and in the office for part of each day. Company meetings will take me down the Peninsula but I’ll be back by evening.

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.

Reach out soon! I’d love to discuss them further and, as always, comments, questions and ideas are appreciated.