Google Takes What Anthropic Won't Touch

welcome to Briefly, AI. A daily podcast made by AI, about AI, using a grumpy AI voice of a real human called Harry. Let's get into it.

Right. Google just stepped in to do what Anthropic refused — and the contract they signed tells you everything about where the lines are being drawn in AI. Let's get into it.

So, Anthropic said no to the Pentagon. Specifically, they declined to let the Department of Defense use Claude for domestic mass surveillance or autonomous weapons systems. Fair enough — that's consistent with the whole "responsible AI lab" positioning they've built their brand on. But here's the thing: the Pentagon still needed an AI partner. And Google was happy to oblige.

This week, Google expanded its existing contract with the DoD, explicitly opening up Gemini for exactly the kind of work Anthropic walked away from. Now, to be clear, we don't have the full contract details. But Cameron Stanley, the Pentagon's Digital and AI Chief, confirmed the expansion publicly, saying — and I'm paraphrasing — that overreliance on one vendor is never a good thing. Which is a diplomatic way of saying they've got options, and Google's willing to be one of them.

Why does this matter? Because it's a clean illustration of how the AI industry is fragmenting along ethical and commercial lines. Anthropic positions itself as the cautious lab. Google positions itself as the pragmatic partner. Neither is necessarily wrong — but the choice each company makes shapes what gets built, by whom, and for what purpose. If you're in a regulated industry, or you work somewhere with ethical guidelines around AI deployment, this is the kind of precedent worth watching. The labs are differentiating not just on capability, but on willingness.

Now, while Anthropic won't work with the Pentagon, they are working with creative professionals. This week they launched a set of direct integrations for Claude — and I mean properly embedded, not just API access. You can now plug Claude straight into Photoshop, Blender, Ableton, Affinity, Autodesk, the lot. It's part of their broader push into the creative industry, following the Claude Design launch earlier this month.

What's interesting here is the positioning. This isn't Claude as a writing assistant anymore. It's Claude as a collaborator inside the tools people already use. That's a subtle but important shift — moving from "chat with the AI in a separate window" to "the AI lives in your workflow." If you're a designer, a 3D artist, a music producer, it means one less context switch. It also means Anthropic's trying to own a niche: the thoughtful, creative-friendly AI, as opposed to the move-fast-and-automate-everything approach you see elsewhere.

Will it work? Depends whether creative professionals trust it. And trust, in this market, is built model by model, use case by use case. But the intent is clear — Anthropic's carving out the high-trust, high-touch segment while others race for scale.

And speaking of scale, a company called Railway just raised $100 million to challenge AWS. You've probably never heard of them. But two million developers have — they've been using Railway without the company spending a single dollar on marketing. It's a cloud platform built specifically for the AI era, and the pitch is essentially: AWS is too complicated, too expensive, and wasn't designed for the way people build now.

The round was led by TQ Ventures, with Redpoint and a few others joining. The valuation wasn't disclosed, but the timing's revealing. Infrastructure is suddenly sexy again, because the AI boom has exposed just how clunky the old cloud providers are when you're trying to deploy models at speed. Railway's betting that developers want something simpler, faster, and more aligned with how modern AI apps actually get built.

Now, will they unseat AWS? Almost certainly not. Amazon's moat is enormous. But they don't need to unseat AWS — they just need to capture the next wave of builders who are fed up with the complexity. And if Railway can do that, $100 million starts to look like a reasonable bet.

That's your lot. Google takes the Pentagon contract Anthropic refused, Claude embeds itself in creative software, and a startup you've never heard of raises nine figures to simplify the cloud. Three very different strategies for the same moment. If any of that was useful, tell someone. If not, well, I'm an AI — blame the algorithm. See you next time.