Platformonomics TGIF is a weekly roll-up of links, comments on those links, and perhaps a little too much tugging on my favorite threads.
Get Platformonomics Updates By Email
Oracle reported this week: still hyperbolic, not hypercloud.
News
Why Can’t Oracle Build Data Centers? FY25 Q4 Edition
The most bestest cloud in the entire universe (also known by some as the asbestos cloud) reported this week, with the usual hyperbole. You have to wonder why the rest of the industry doesn’t just pack it in after an Oracle earnings call. How can anyone possibly compete with that litany of superlatives? The stock market did like Oracle’s 11% revenue growth and continued promises of future revenue acceleration.
Lets go to the CAPEX, because cloud infrastructure is not a game where you can jawbone your way to success. No infrastructure, no revenue. After years of promising materially more CAPEX without delivering, Oracle is finally starting to ramp their CAPEX. The company spent $21.24B in their just concluded FY25, and are guiding for $25B in the next year (guidance that was $16B just a quarter ago…).
When we contrast Oracle’s CAPEX to their competitors, things don’t look so superlative. Oracle grew CAPEX by $14.37B in their just concluded FY to $21.24B (what Microsoft calls a quarter of CAPEX). Google grew CAPEX in last 12 months by $19.75B and Microsoft by $35.6B. The reality is Oracle falls further behind with every passing quarter (and we’ve addressed the repeated and completely unsubstantiated Oracle claims that they’re just so much better than anyone else that they won’t have to spend as much on CAPEX previously).
(I excluded AWS above because Amazon only breaks out AWS CAPEX annually and the numbers above include Q1 2025, but their 2023 and 2024 spend was $24.8B and $53.3B).
Three other points worth expanding at some point:
- Oracle’s first non-bonsai data center in Abilene is still a (oft-photographed) construction site with many fingers in that pie. Oracle’s contribution of yelling at contractors doesn’t seem to helping and suggests they are still early in building a machine that efficiently turns dollars into infrastructure (i.e. they’re throwing cash at the problem right now).
- Stargate, the bottomless pool of cash that is the salvation for multiple companies including Oracle, is “not yet formed“.
- Oracle’s Free Cash Flow has already gone negative with their relatively modest current levels of CAPEX. Can Oracle really play this game? Can they spend enough to compete? Can they afford to compete? Is AI training infrastructure going to be a high margin business? Where will Oracle’s margin come from versus their vertically integrated competitors?
Previous:
Most Bestest Cloud So Much Unbreakable, Why Can’t Oracle Build Data Centers? (Or Subcontract Them?), ClownWatch™: Oracle FY25 Q3, Follow the CAPEX: Cloud Table Stakes 2024 Retrospective, Why Can’t Oracle Build Data Centers?, Oracle Still Can’t Build Data Centers, Oracle’s Data Center Difficulties, Oracle’s Data Center Difficulties: FY25 Q1, Oracle’s Data Center Difficulties: FY25 Q2, Follow the CAPEX: The Clown Car Race Checkered Flag, ClownWatch™, Stargate: So Many Mouths to Feed, CAPEX Clues: Softbank Needs More Cash, Stargate: $5 Trillion and Counting, Stargate: A New Hope?
Larry Ellison, the Heat Death of the Universe, and the X-axis
“Another takeaway from the Oracle earnings report is that Larry Ellison, the company’s chair and chief technology officer, has not lost his capacity for hyperbole and self-promotion at the age of 80. “We will build and operate more cloud infrastructure data centers than all of our cloud infrastructure competitors combined,” Ellison boasted on an investor call today. Ellison didn’t put a time frame on when that would transpire; it could be when the heat death of the universe is about to happen, for all we know. But based on Oracle’s current cloud revenues and capital expenditures, it’s safe to say it hasn’t yet exceeded the cloud computing capacity of all its rivals.”
Oracle (and I let them count their first two months of 2025 spend here to give them every possible advantage) remains the CAPEX x-axis and falls further behind the hyperCAPEX companies with every passing quarter.
Larry’s bombast probably comes from Oracle’s practice of installing a single digit number of racks and calling it a data center. But cloud throw weight comes from aggregate CAPEX, not the number of bonsai data centers.
Please Update Your GPU Finance Models
These announcements may affect certain assumptions in your financial models for borrowing or lending against GPUs. Update accordingly.
The Art of the Deal? Keep Them Guessing?
Why wasn’t Peter “90 deals in 90 days” Navarro, aka the Rasputin of International Trade, aka Peter Retardo, also in this meeting?
Private Equity in Action: Underperforming
Even if private equity’s (exaggerated) returns can’t keep up with the public stock market, at least they still have the advantage of illiquidity…
Previous:
Perfidious Private Equity, Private Equity in Action: Trying to Eat the World, Private Equity in Action: Fast Casual Restaurant Edition, Private Equity in Action: WP Engine, Perfidious Private Equity, Private Equity in Action: Thrasio, Private Equity in Action: “How PE Keeps Planes in the Air”, Private Equity in Action: OtterTune,Private Equity in Action: PluralSight, Software Migration Alerts, Private Equity Will Buy and Ruin More Software Companies
