Platformonomics TGIF #84: March 14, 2025

By

Platformonomics TGIF is a weekly roll-up of links, comments on those links, and perhaps a little too much tugging on my favorite threads.

Lots of database vampire and antitrust incoherence this week. No missive next week.

News

ClownWatch™: Oracle FY25 Q3

Database vampire, the x-axis of cloud CAPEX and most amazeballs company ever (to hear them tell it) Oracle reported Q3 earnings. Revenue growth was — wait for it — 6.4% (their much bigger competitors all grew at least twice as fast, but Oracle did beat out both fellow cloud clown IBM and inflation). The company missed on both revenue and profit expectations.

But future so most glorious. Again Oracle talked up faster growth out on the horizon and RPO (which for Oracle means selling capacity they don’t have) while downplaying current performance. They’re now hyping expected FY27 growth rates! We’ll see…

There was no Seemingly Unscripted Moment™ from Larry Ellison on the conference call. Alas. The small modular reactor incident may haunt them as everyone has accounced concrete nuclear plans except Oracle.

But most important: CAPEX. Oracle spent $5.86 billion in the quarter, a record high for the company, and are guiding to “around $16 billion” for FY26 which starts in June. They will make their CAPEX guidance for the fiscal year (after a bunch of CAPEX misses 90 days after guidance). Yay Oracle!

With that brief congratulatory interlude completed, lets put those numbers into context. Oracle still falls further and further behind the hyperCAPEX companies with every passing quarter.

AWS, Google and Microsoft’s incremental CAPEX spend in the last year each exceeded Oracle’s total spend in the last year (and Meta just missed being in that club). Those three companies spent more on CAPEX in 2024 than Oracle has spent cumulatively in its nearly 50 year existence. All four hyperCAPEX companies will spend more next quarter than Oracle has guided to for the next year. But sure, we can pretend like Oracle is somehow catching up.

Oracle is taking a lot more financial risk than our other contestants. Their debt grew by 10% to $88 billion in the last quarter. This quarter’s $5.86 billion in CAPEX just about wiped out their $5.9 billion in operating cash flow. And Oracle announced they’re increasing the dividend by 25% (Larry needs cash? See below). Softbank isn’t the only company betting on the rosiest of AI futures.

Stargate: So Many Mouths to Feed

Abilene, Texas claims to host the world’s largest BBQ. Experience feeding multitudes will come in handy, with all the different companies taking credit for, and hoping to profit from, the new data center in town (this one data center gets more PR than all other data centers combined, perhaps because there so many interested parties each trying to make their narrow slice seem bigger than it).

Lancium owns the “data center campus”. Crusoe has leased property, is the general contractor, and owns the data center (a co-lo basically). Oracle is leasing the data center and populating it with servers. Open AI will pay to train its models on Oracle’s infrastructure. And Softbank is promising to rain cash down on everyone involved (along with a variety of other financial investors).

This horizontal structure stands in marked contrast with the vertically integrated hyperscaler approach. Who gets margin and who doesn’t? Are there Coasian coordination costs? I’m too lazy to label all the layers, but you get the idea.

Tik Tok, Swamp Creature

Vague, but incredibly disappointing if true on two fronts. It seems the CCP would retain control (never mind that divest-or-ban law passed by a bipartisan Congress, signed by the president and upheld unanimously by the Supreme Court). And it is another data point for the argument we’re turning into an oligarchy. I await an emergency khakiocracy podcast explaining why actually this is fine for the rule of law and really good even for Little Tech!

The Oracle dilemma of both needing to retain its existing Tik Tok revenue but not having nearly the cash to purchase Tik Tok may lead us to this worst of all worlds outcome, if Oracle is at the top of the oligarchic pecking order for this issue. It isn’t clear whether the Tik Tok interest beyond hosting is an Oracle thing or Ellison family thing (Larry and son Bronny, who runs Paramount) thing. The Oracle dividend increasing to put $2 billion a year into Larry’s pocket might tell us something.

Question for legal scholars: what kind of standing do you need to sue the likes of Akamai, Apple, Google and/or Oracle for being in violation of the Protecting Americans from Foreign Adversary Controlled Applications Act by helping Tik Tok? Let me know (responses will be kept anonymous).

Going Nuclear: Yay Nuclear Edition

Not sure whether there is any actual substance to this, but I like the sentiment. Some players are insufficiently sentimental:

Microsoft and Apple, which also use significant amounts of energy and have committed to reducing their carbon footprint, did not sign the statement.

Amazon has come far since “No Team at Amazon is Working on Nuclear Power“:

Amazon said it had invested more than $1bn in the nuclear industry in the past year and that speeding up new power stations would be “critical” for US security, meeting growing energy demands and helping combat climate change.

And yes nuclear is an investment for the 2030s. But the wait times for gas turbines are into that decade as well. There are no quick energy fixes.

John Ketchum, chief executive of NextEra Energy, which operates one of the largest nuclear power fleets in the US, told the Financial Times that new nuclear power would take until “2035 or later”, noting that “we still have to develop the first of a kind advanced nuclear unit that works”.

Brian Savoy, chief financial officer of utility Duke Energy, said after his company’s earnings last month that it did not expect SMRs to play a meaningful role until the late 2030s.

Antitrust Incoherence: iRobot Elegy Edition

Take a victory lap Lina Khan and the EU! Yay competition!

Antitrust Incoherence: Existential Amazon Questions Edition

Why are we here? Who are trying to protect? From whom? It is almost like antitrust needs a core set of consistently applied principles to help determine what is and is not anti-competitive behavior, as opposed to treating every company as a snowflake and making up the rules each time. Perhaps a doctrine?

And I would be remiss if I did not reiterate that Lina Khan made her name and got her FTC job by claiming Amazon’s prices were too low.

Antitrust Incoherence:Dodging DOGE Edition

Would love to see the internal FTC email thread from the hours between those two headlines.

Antitrust Incoherence: The Emerging Trump Doctrine?

The Biden administration antitrust doctrine of “Big Tech Bad!/Break Them Up!/All Acquisitions are Bad!” is starting to look coherent (or at least consistent) compared to the emerging Trump administration doctrine of “Assist Elon with his Vendettas” and/or “Antitrust is Just Another Source of Leverage”. This case is another example of Trump FTC continuity with Biden FTC policies and priorities, which I vaguely recall was not popular amongst the khakiocracy.

An antitrust doctrine provides clear lines on the field for all the players, and mitigates against arbitrary and inconsistent enforcement. After spending decades of climbing out of the abyss of arbitrariness, antitrust law is falling back into its primordial politicized swamp.

Prediction Victory Lap: Ireland Joins Enemies List

Never mind that the imbalances stem from Trump’s 2017 Tax Cuts and Jobs Act, which incentivized IP companies (particularly pharma and tech) to move their profits to Ireland. It is more tax avoidance than trade imbalance, but we’re long past such nuances. And there are few prizes for picking enemies from an infinite list.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Get Updates By Email

Discover more from Platformonomics

Subscribe now to keep reading and get access to the full archive.

Continue reading