Platformonomics TGIF #54: June 14, 2024

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.

An interesting week as the strategic fault lines creak and shudder.


My Writings

Why Can’t Oracle Build Data Centers?

I asked before Oracle’s earnings announcement why the company can’t seem to build data centers. Is Johnny-One-Rack struggling to scale up?


News

Oracle Still Can’t Build Data Centers

Oracle’s FY24 Q4 CAPEX spend was $2.8B and $6.87B for the fiscal year (a decline of 21% from the prior year). The company missed the low end ($7B) of its twice-lowered CAPEX guidance, most recently revised after the third quarter.

The company missed both revenue and earnings expectations, but the stock soared after the company aggressively chanted “AI, AI”, raised revenue growth guidance for next year from mid-single digits to 10%, and highlighted 44% growth of RPO (Remaining Performance Obligation, which is revenue they have booked but not delivered).

It sure looks like Oracle is selling cloud capacity they don’t have, and unless they quickly get on a vastly steeper learning curve for building data centers, are unlikely to have that capacity any time soon. CAPEX is required to turn RPO into revenue and that never happens with the two metrics going in different directions.

The Siri Black Hole

Apple laid out their vision for the personal assistant and reframed the industry AI discussion with Apple Intelligence. They were very true to the personal computing ethos and maximized Apple’s home field advantage. The discussion of whether Apple has missed the boat on AI should now move onto whether they can deliver.

Apple is the company I most trust to deliver a personal assistant. If they deliver, even just partially, they will have the strongest iPhone upgrade proposition in years.

From an application ecosystem standpoint, the personal assistant gets better with more relevant personal information and context. This will change the dynamics for Apple’s own applications and the whole App Store ecosystem, as Apple seeks to make Siri the epicenter of user interaction. Third party apps will need to decide whether to join or try to exist standalone. Either way, Apple is likely to invest more in its own applications to drive that virtuous cycle of more information and context. The App Store equilibrium is over.

Apple finessed the frontier model question brilliantly and embraced a pluggable model as I previously predicted.

Apple Intelligence will also pressure Android. Google, already sowing disequilibrium in the ad-supported web ecosystem as the company moves from organizing the world’s information to summarizing it, has another battle to fight.

Apple Intelligence Infrastructure

I historically have not closely tracked Apple’s CAPEX because they don’t have a public cloud, their spend has not kept up with the hyperclouds despite being an enormous company, and a bunch of what they do spend goes to manufacturing tooling.

Perhaps the biggest risk to pulling off Apple Intelligence is the back-end infrastructure. They are building what they claim is a privacy-preserving cloud infrastructure that allows them to do cloud processing while remaining true to their privacy promises. Apple intends to implement that architecture on an all-new cloud infrastructure built on Apple silicon (which historically has been optimized for clients). Apple has limited experience building hyperscale infrastructure and will not be able to lean as much on the existing data center ecosystem. We’ll monitor Apple CAPEX going forward, because CAPEX is such a good perspective on what is happening around AI.

Elon Musk Threatens to Ban Oracle?

Thin-skinned fractional CEO Elon Musk had a tizzy after Apple announced ChatGPT integration. We await his similar reaction to Oracle announcing Microsoft was vacuuming up some of their GPUs to support ChatGPT (I’m pretty sure Oracle misrepresented this arrangement, given it was a unilateral press release and Oracle is gonna Oracle).

But that may explain why Larry spent so much time kissing Elon’s ass on Oracle’s quarterly conference call (which is atypical to say the least). Larry announced Elon is close to FSD approval in China and did a soliloquy (a silly-quy really) about how Oracle is automated like Starlink.

One response

  1. Hilarious read as always 😂

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