Anthropic & Claude

Riot Platforms spent years mining Bitcoin. Now it's leasing 191 megawatts in Texas to Anthropic – for 20 years and up to $16.1 billion. The third billion-dollar deal in three months.

To meet the EU's new transparency rule, Anthropic will start marking text and files from its models. The watermark sits inside the text itself and travels along when you copy and paste.

Simon Willison pulled one passage from the Opus 5 system prompt: Anthropic tells the model, word for word, how to talk about the export-control suspension of Fable and Mythos. A tiny piece of text with a big job.

Together with Macquarie and GIC, Anthropic launches a platform called Theseus. Others fund the data centers, Anthropic rents them – and promises to cover the grid-upgrade costs in full.

OpenAI splits its security program Daybreak into two tiers and ships GPT-5.6-Cyber. It looks a lot like Anthropic's Mythos – and raises the question of who's really protecting whom here.

From September 1, Claude Sonnet 5 moves to full standard pricing: 50 percent more per token. Combined with the new tokenizer, the real cost jump for some workloads is even steeper.

Simon Willison gave the same one-shot prompt that had Claude Fable 5 build a little raccoon game to Codex running GPT-5.6 Sol Ultra. The result: a noticeably better game — with a very funny bug.

Hedge fund giant Millennium is co-developing an AI teammate for risk management with Anthropic. Claude is meant to surface risks across asset classes — supervised by human risk managers.

Anthropic has signed a six-year, $10 billion deal with Volta — a startup that's only seven months old. The compute comes from a hydro-powered data center in Norway, running Nvidia's Vera Rubin chips.

Simon Willison turned an old tweet concept into a playable browser game in one go with Claude Fable 5 — 'Raccoon Heist'. A small test that says a lot about how far one-shot coding has come.

Anthropic is hiring engineers for a 'custom silicon team' and has confirmed it: the Claude maker wants to build its own AI chip, co-designed with its next models. Samsung is reportedly in talks as the manufacturing partner.

On the evening of August 4th, Anthropic stumbled: some users couldn't get into Claude.ai or Claude Code via OAuth, and there were elevated error rates across all current models. It was resolved — but it's a good moment to think about dependence.

Simon Willison shipped his LLM tool in its biggest release since launch — plus an llm-anthropic update that brings the Claude 5 models, server-side tools, and a cleaned-up thinking mode to the command line.

A Swedish developer wrote an essay about why he benched Claude Opus 5 for daily work. Not over code quality — he actually praises that. But because the tone got too snippy for him. A small story with a bigger question behind it.

Data-center operator CoreWeave had to improve the terms of a $2.6 billion loan to win over nervous lenders. The money partly funds capacity for Anthropic — and the rising yields are a signal that reaches beyond CoreWeave.

Google DeepMind is breaking up its Nobel-winning AlphaFold team and going all-in on Gemini. Three key researchers land at Anthropic. A turning point in the fight over AI for science.

On July 29, Claude was disrupted worldwide for about three hours. Instead of answers, users got a wall of '529 Overloaded'. It hit claude.ai, the API, and Claude Code — so every tool built on Claude went down with it.

IT services giant Cognizant is embedding Claude into its industry platforms and putting the model in front of roughly 350,000 employees. The underlying message: AI needs to get out of pilot mode and into production.

The official launch is done — now comes the more honest part: independent testing. Simon Willison and several benchmark shops have put Opus 5 through the wringer. The verdict is strikingly clear — with one notable security highlight.

Anthropic launched Claude Opus 5 — close to Fable 5's frontier intelligence, the company says, but at half the cost. Opus 5 becomes the new default on Max and Pro. And in one area, the model deliberately holds back.