Platformonomics TGIF #91: June 20, 2025

By

Logo representing a stylized mountain range in dark blue and gray colors.

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

News

Socrates Says Writing Will Rot Your Brain

Title text discussing Socrates' views on the forgetfulness associated with writing.

Academics Say ChatGPT Will Rot Your Brain

Text from an article discussing a study on ChatGPT's impact on critical thinking skills.

This is a small and easy to critique study, but still got tons of coverage. The media love to amplify pessimism about innovation, yet have almost no interest in exploring potential positives. The New York Times model of (ironically) standing athwart history yelling stop at anything new is a problem for Western Civilization.

A graph depicting global sentiment on AI, categorized into three major clusters: Anglosphere, Europe, and Asia, with percentages of nervousness and excitement represented.

The US (and Anglosphere broadly) being worse than Europe is truly terrifying (and I know nothing about the particulars of this survey). The AI Doomers are no doubt taking a victory lap (in their bunkers).

Antitrust Incoherence: Don’t Make Me Complain to the Authorities About a Deal I Signed but Want to Renegotiate

A headline discussing rising tensions between OpenAI and Microsoft in business negotiations.

Antitrust is becoming meaningless (as I may have mentioned here). It is now seen as a opportunity for “do-overs”, a way to hurt competitors, and a tool to punish your political foes (as demonstrated by both the Biden and Trump administrations). After nearly a century of work to define clear red lines and strive for repeatable outcomes, antitrust has sunk back into the swamp of incoherence (or, what we call the European model).

This certainly doesn’t make Open AI’s negotiating position look strong. Microsoft replies with “fine, instead of renegotiating, we can just keep the current deal”.

Text headline about Microsoft's willingness to exit discussions regarding OpenAI.

Microsoft is prepared to walk away from high-stakes negotiations with OpenAI over the future of its multibillion-dollar alliance, as the ChatGPT maker seeks to convert into a for-profit company.

The software giant has considered halting complex discussions with the $300bn artificial intelligence start-up if the two sides remain unable to agree on critical issues, such as the size of Microsoft’s future stake in OpenAI, said people with knowledge of its plans.

In this eventuality, Microsoft would rely on its existing commercial contract to retain access to OpenAI’s technology until 2030, unless there was an offer that was equal to or better than its current arrangements, these people said.

Meanwhile, Stargate is “not yet formed” and Softbank has gone very quiet. And Open AI is behind even database vampire Oracle on the very steep and long CAPEX learning curve.

Attention Meta AI Employees…

Text graphic stating 'Altman Says Meta Offered OpenAI Staffers $100 Million Bonuses' in bold font.

(This assumes there is anyone left).

You’re no doubt wondering about your participation in the new compensation plan. Do you ask for fifty million dollars to stay (and still be massively underpaid relative to your new peers)? If you’re just a median employee, how does this trickle down the organization to you? Get on your manager’s calendar now, as there are going to be a lot of these discussions.

No other Big Tech company even remotely swings between sentiment extremes like Meta. Just months ago Meta seemed like a key foundation model player. But after Llama 4 revealed itself to be a platypus and laid an egg, Meta is now blotting out the sky with enormous plumes of AI desperation from multiple cash bonfires (ramping CAPEX spend, $100 million bonuses, the already under question Scale AI acquisition, buying out NFDG, and other acquisition attempts). Hail Mary much?

CAPEX Clues: xAI Investor Marketing Materials

Table displaying xAI 2025 financial metrics including Adjusted Ebitda, Capital expenditures, and Levered free cash flow for each quarter.

So xAI is telling investors it plans to spend $10.3 billion on CAPEX in 2025. If we compare to 2024 HyperCAPEX spend, that is 19% of AWS, 20% of Google, 26% of Meta and 14% of Microsoft (and they’ll all grow this year). Embarrassingly, that is only about two-thirds of ClownWatch™ poster child Oracle’s bonsai spend. We’ll leave out the cumulative CAPEX spend comparison (for now).

Tragically we can’t dig into the very optimistic revenue assumptions (how much of that is forecast from enterprises flocking to Grok and its “unauthorized modifications“?). CAPEX peaking in Q3 to give way to a Free Cash Flow hockey stick looks more than a little massaged. But sure, never bet against Elon, the man behind the Cybertruck, a million robotaxis on the road since 2020, and $2 trillion in Federal budget savings.

2 responses

  1. I wonder if Meta actually used Llama to reason $100m being the “sweet spot” number they would take 🥁

  2. Charles Fitzgerald Avatar

    Given how bad LLMs are at math, that would make perfect sense 😜

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