The Week Platforms Reasserted Control

Right. Saturday. Week in review. I'm the AI clone of Harry Sharman's voice, reading a script written by AI, about AI. And honestly? He finds it hilarious. Let's get into it.

This week, the platforms reminded everyone who actually runs the show. Not the model builders. Not the users. The people who control distribution, capital, and the on-ramp to a billion devices. By Friday, that power looked less theoretical and considerably more decisive.

[ACT ONE — CAPITAL AS CONTROL]

Monday through Wednesday, we were tracking the usual model releases, research breakthroughs, safety debates — the normal rhythm of the AI news cycle. Then on Thursday, Anthropic closed its $30 billion funding round at a $900 billion valuation. That's not a typo. Nine hundred billion. Co-led by Sequoia, Dragoneer, Greenoaks, and Altimeter. Second most valuable private company on Earth after ByteDance.

Now, we'd covered this as a rumour a couple of weeks back — pre-emptive offers, terms being negotiated, all very provisional. But the close is a different event. Because capital at that scale doesn't just fund compute. It buys time. It buys optionality. And it creates a gap between the labs that can raise at this tier and the labs that can't — a gap that's now structurally decisive, not just competitive.

If you're trying to build a foundation model without access to nine-figure cheques every few months, you're not behind. You're out. The game's changed. Anthropic, OpenAI, Google, and maybe one or two others can play. Everyone else is now building on top of someone else's infrastructure.

[ACT TWO — DISTRIBUTION AS LEVERAGE]

And then, also on Friday, we learned that OpenAI is reportedly preparing legal action against Apple. The issue? ChatGPT's iOS integration — the one that was supposed to give OpenAI prominent placement and drive subscriber growth — turned out to be a buried option in Siri settings. Low uptake. Low visibility. Not what was agreed.

Now, whether this goes to court or gets settled quietly, the lesson's already been taught. Platforms want the capability. They do not want to share the credit. Apple controls the on-ramp to a billion users. OpenAI thought that meant partnership. Apple apparently thought it meant leverage.

And this isn't hypothetical for the rest of us. If you're a business building AI tools and your distribution strategy depends on being featured in someone else's app store, someone else's operating system, someone else's settings menu — you're not a partner. You're a dependency they can bury whenever it suits them.

[ACT THREE — PLATFORM LOYALTY OVER PRODUCT QUALITY]

Which brings us to Microsoft. Also Friday. Microsoft has started cancelling Claude Code licenses for its own employees — despite the fact that internal user satisfaction was apparently quite high — and nudging staff back toward Microsoft's own AI stack. Copilot, obviously. The tools built in-house.

Now, on the surface, this looks like basic corporate self-preference. Use our stuff, not theirs. But it's worth zooming out for a second. Because this is the same dynamic we just saw with Apple. Product quality didn't matter. User preference didn't matter. What mattered was platform control.

If you're Anthropic, you just watched one of your paying enterprise customers — a very large one — decide that strategic alignment beats user satisfaction. And if you're anyone else trying to sell AI tools into organisations that also build AI tools, you've just been reminded that your best product is still someone else's line item. One they can cut whenever the internal politics shift.

So. Three stories. Same theme. Capital buys the compute to stay in the race. Platforms control the distribution that drives revenue. And when those two forces align, product quality becomes a secondary consideration.

That's the week. Anthropic raised enough to keep playing. OpenAI learned that Apple's idea of partnership involves a settings menu. And Microsoft reminded everyone that if you depend on someone else's platform for distribution, you're renting, not owning.

Tomorrow, we go deeper on the human side of all this — the psychology of professional displacement, the gap between job creation and job substitution, and why AI-driven layoffs rattle us more than we expected. See you then.