7 Comments
User's avatar
ToxSec's avatar

awesome update and roundup. anthropic got a lot of pushback for tinkering with claude i think.

Rome Thorndike's avatar

Thanks, Anthropic can fix this by buying compute, but there's no reason until a viable competitor comes. They can cut off consumers, focus on the enterprise, and see no ARR impact even if those consumers churn.

Aniket Chhetri's avatar

Good roundup. The energy narrative around OpenAI was oversimplified from the start.

Petar Dimov's avatar

Feels like the AI race is becoming less about models and more about who controls compute, infrastructure, and the money behind it

Rome Thorndike's avatar

For sure, explains why SpaceX and Amazon are investing so much in outer space

Colleen Avarene's avatar

Howdy Rome — the OpenAI copyright section is the strongest analysis I've seen on that pause. Everyone repeated the energy cost headline and you're the first person I've read who connected the timeline to the March 18 licensing assessment. Three weeks from policy shift to announcement. That's not coincidence — that's a risk committee meeting.

The Anthropic section is the one I keep thinking about though. I build custom AI agents and the compute allocation problem you're describing is exactly what my clients feel before they come to me. They're on Claude Pro or ChatGPT Plus and the quality drifts and they can't figure out why. They think they're doing something wrong. They're not —

they're just not the customer the company is optimizing for anymore.

That's the real cost of the enterprise-first trade-off. It's not just degraded service — it's trust erosion. When a user can't tell if the tool got worse or if they're imagining it, that's worse than a price increase. A price increase is honest. A quiet quality shift with a disputed explanation isn't.

The Meta section is wild too. Betting all free cash flow on infrastructure that depends on revenue that doesn't exist yet — funded by shutting down the last bet that didn't pay off. That's either visionary or the most expensive sunk cost fallacy in corporate history. We'll know in eighteen months.

Rome Thorndike's avatar

Thank you for the kind words Colleen!