Dear WIMM Reader,
We continue our week:
18 March - [Essay] Lessons from Koenigsegg
19 March - [Analysis] The disruption of the German car industry
20 March - [Market updates] Oracle
21 March - [Deep dive] Micron
22 March - opening my 3rd portfolio (for Premium/paying subscribers only)
TL;DR
Oracle’s quarter showed that AI contracts are translating into real growth and a massive backlog.
The opportunity is obvious: Oracle can turn its database moat into an AI infrastructure and multicloud control-plane business.
The risk is just as obvious: capex, negative free cash flow, and rising net debt mean execution has to stay near perfect.
The next chapter will answer the question “Can Oracle turn RPO into revenue, revenue into cash flow, and cash flow into confidence?”
Just one month ago, I wrote in the post called Oracle’s AI Backlog Problem the following:
The real story is the cash flow inversion. Operating cash flow in Q2 was $2.1B, but FCF was -$10B and CapEx was $12B. Management is basically saying: “We’re not losing money; we’re pre-buying revenue-producing equipment late in the data center cycle”. They also emphasize leases cover much of land/buildings/power timing, and they’re exploring customer/supplier financing structures (bring-your-own-chips; chip leasing).
They’re defending their ability to fund and deliver supply while staying investment-grade. That’s a very specific kind of corporate flex and also a tell that the constraint is execution, not sales.
From Bloomberg:
Oracle Corp. shares gained almost 10% in extended trading after the company posted strong results and gave an outlook that suggested there is little letup in demand for AI computing. Revenue in Oracle’s closely watched infrastructure business increased 84% to $4.9 billion in the period ended Feb. 28, the company said Tuesday in a statement. That marked a faster jump than the 79% anticipated by analysts and a 68% sales rise in the previous quarter. Total revenue will reach $90 billion in the fiscal year beginning in June, Oracle said. Analysts, on average, estimated $86.7 billion.
The company is working to deliver on massive cloud infrastructure contracts with customers like OpenAI and Meta Platforms Inc. Known for its namesake database software, Oracle has found success with its cloud business by providing chip-filled data centers and other equipment for training and deploying AI models. That effort comes with huge costs. Capital expenditures, a metric of data center spending, were about $18.6 billion in the fiscal third quarter, higher than the $14 billion anticipated by analysts. The company maintained its outlook for $50 billion of capital expenditures in the current fiscal year, which “could address concerns about overspending that have plagued Oracle and other cloud infrastructure providers,” wrote Anurag Rana, an analyst at Bloomberg Intelligence.
Oracle is quickly delivering cloud capacity to customers, with 90% in the quarter provided on or ahead of schedule, co-Chief Executive Officer Clay Magouyrk said on a conference call after the results were announced. Remaining performance obligation, a measure of bookings, was $553 billion, compared with the $523 billion reported in the prior quarter. Most of this increase came from large-scale AI contracts in which the customers will fund the up-front purchases of semiconductors, the company said in the statement.
My bull-ish feedback was mainly related to the fact that the company has to commit now the money for a demand that is in the future:
…and this one:
So, what happened? Basically, Oracle is capitalizing on the AI wave, which is also a main thesis at Where is my moat? - meaning, tech companies will become bigger and better.
Ben Thompson explained in a post what the 3 major changes in the AI boom:
The first inflection point was the emergence of LLMs — call this the ChatGPT moment. In this first paradigm tokens were generated by GPUs and presented as the answer to a question.
The second inflection point was the emergence of reasoning models — call this the o1 moment. In this paradigm there are a very large amount of tokens that are generated to figure out the answer before the answer is actually generated; this was an exponential increase in the addressable market for tokens.
The third inflection point was the emergence of functional agents — call this the Opus 4.5 moment. In this paradigm those reasoning models are not triggered by humans asking a question, but by an agent solving a problem. This increases the market in two directions: first, humans can run multiple agents, and secondly, agents can leverage reasoning models multiple times to accomplish a task. This isn’t just an exponential increase in the addressable market for tokens, it’s two exponential increases squared.
Oracle’s quarter was strong
Oracle is becoming a leveraged infrastructure company that is trying to convert database captivity into cross-cloud relevance and AI demand into long-duration cash flows. That is a much bigger opportunity than just an ERP company. It is also a much riskier setup.
The bull case starts with one simple fact, which is, the demand is real. Oracle’s management made clear that GPU and CPU demand still exceeds supply, and that much of the new RPO growth is tied to large-scale AI contracts. Clay Magouyrk (co_-CEO) also said Oracle delivered more than 400 megawatts to customers in Q3, with 90% of committed capacity delivered on or ahead of schedule. On the delivered AI capacity, Oracle said gross margins remained above its 30% guidance, at 32%.
That matters because the market’s biggest fear was not whether Oracle could sell AI capacity. The fear was whether Oracle was building too much, too fast, with too much debt. This quarter offered a partial answer. Oracle is trying to reduce that financing burden by shifting part of the upfront cost to customers. Management said it signed more than $29 billion of contracts using models such as bring-your-own-hardware and upfront customer payments, precisely to expand AI infrastructure without forcing negative cash flow on Oracle for those contracts. Bloomberg also reported that most of the latest RPO increase came from large AI contracts where customers fund the upfront semiconductor purchases.
This is why the quarter was better than expected. Oracle showed that demand is strong enough for customers to help finance the buildout. That is a very different situation from a speculative supply push.
Still, Oracle’s moat is evolving in a very specific way. Historically, Oracle’s strength came from switching costs. Databases, workflows, audit trails, industry-specific processes, regulatory requirements, and decades of embedded logic made Oracle hard to rip out. That old moat is still there. What is changing is that Oracle is now using AI and multicloud to widen the perimeter. The company’s Multicloud Database business grew 531% year over year, and Oracle now has broad region coverage across Microsoft, Google, and AWS. The strategic point is obvious, where Oracle no longer wants to be only the place where data lives inside Oracle. It wants to be the database layer and private-data control plane across all clouds.
That is what makes this quarter more interesting than a simple “OCI [Oracle Cloud Infrastructure] had a great quarter” story. Oracle has two engines working at once:
a) The first is AI infrastructure, where scarce compute and power create a window for Oracle to matter.
b) The second is multicloud database, where AI adoption pushes enterprises to move their most valuable private data into cloud environments that Oracle can serve, regardless of where the model runs.
In that sense, OCI is a lead generator for the broader stack. Management explicitly described a halo effect: AI infrastructure is pulling through database, applications, sovereignty, and multiproduct deals.
What comes next?
Over the next few quarters, the key issue is not demand discovery. It is conversion and discipline.
Investors need to see CapEx turn into revenue with shorter lag. My earlier Oracle analysis framed this perfectly, where backlog is the proof, CapEx is the toll. The near-term catalyst is OCI growth accelerating in a way that justifies the capital intensity. If CapEx stays huge but monetization slips, the stock will struggle. If revenue catches up fast enough, the market will forgive the spending.
Watch RPO [Remaining Performance Obligations] quality, not just RPO size. Oracle’s $553 billion RPO is enormous, but the debate is about concentration and duration. My previous analysis argued that a very large share of backlog growth may be tied to one major counterparty, likely OpenAI, and that Oracle may be taking on longer-dated obligations than some customer commitments justify. It also noted concerns around lease obligations, duration mismatch, and the fact that Oracle has moved into negative free cash flow territory. Those concerns do not disappear because one quarter was strong. They are merely postponed if Oracle keeps executing.
A multicloud database may become the cleaner, higher-quality growth story. AI infra gets the headlines, but multicloud database may ultimately be the better business: higher margin, less balance-sheet stress, and deeply connected to Oracle’s legacy advantage in enterprise data. If enterprises increasingly want frontier models to reason over private data securely, Oracle is well-positioned to be the layer that makes that possible.
Software will matter more than the market thinks. Oracle is trying to tie applications, database, infrastructure, and AI tooling into one enterprise automation stack. If that works, Oracle stops being “late to cloud” and starts looking like a rare company that can monetize AI at every layer: compute, data, workflow, and industry-specific automation
Concluding:
The quarter proved that demand is real, delivery is improving, and the business model is adapting faster than bears expected. But Oracle is still running a very aggressive transformation financed with heavy capital commitments. The upside is enormous because Oracle could become the neutral private-data utility for the enterprise AI era. The risk is also obvious, that is, if backlog quality disappoints, if AI demand shifts, or if monetization lags the buildout, the balance sheet becomes the story again
So the right framing for Oracle today is not “software company with a cloud option”, but “enterprise data moat trying to become AI infrastructure plus cross-cloud control plane”. That is why the stock is now about execution, not narrative. For the next few quarters, execution means four things: (a) revenue conversion, (b) financing discipline, (c) multicloud expansion, and (d) proof that Oracle can turn this AI rush into durable, high-quality cash flows.
Verdict: Buy (change from Buy-ish)
“Where is my moat?” is designed for individual readers, though the occasional forward is absolutely fine. If you’d like to set up multiple subscriptions for your team with a group discount (minimum 5 seats), reach out to me directly.
Thanks for your support & have a wonderful day!
Thank you for being a supporter of WIMM
