- The midterms turned data centers into a losing political bet
- Google's $17B Wyoming data center is actually a rescue takeover
- A US intelligence memo alleges a Chinese spy got close to Nvidia's CEO
- Microsoft is quietly turning off AI capacity it rents from Oracle
The week AI hit a trillion-dollar valuation - and its first real doubts
An archive of the AI briefings from the week of November 2 to November 8, in plain English. The home page keeps the current week, Monday to Sunday; every completed week stays archived under Past weeks.

The week at a glance
This archive covers the seven editions from November 2 through November 8. The week was about the AI boom's staggering price tags - and the first real doubts about them.
The money was enormous. OpenAI closed the first $11 billion of a new round at a $1.25 trillion valuation, Fal raised $1.3 billion at a doubled $16 billion valuation, and Salesforce agreed to buy voice-AI maker Cresta.
The doubts arrived alongside. Investors questioned Anthropic's $1.5 trillion IPO valuation after the year's sharpest tech selloff, data showed business AI spending growth stalling, and Microsoft began winding down AI capacity it rents from Oracle.
The buildout met politics. Voters approved more than 20 midterm measures restricting data centers, Google backstopped a $19 billion Anthropic lease and announced a $17 billion Wyoming campus that turned out to be a rescue takeover of a stalled Crusoe project.
And the week ended on espionage: a US intelligence memo alleged a Chinese spy got close to Nvidia's CEO - unverified, but enough to draw a congressional briefing request.
Key stories of the week
- OpenAI closed the first $11 billion of a new funding round at a $1.25 trillion valuation, expected to reach about $20 billion - below the $30 billion target reported in September.
- Doubts surfaced around Anthropic's planned IPO: potential investors question whether it can hold a $1.5 trillion valuation after the year's sharpest tech selloff, while data showed business AI spending growth stalling.
- OpenAI released GPT-6 Cyber, a security model it rates in its highest risk tier, built for vetted defenders to find and validate software vulnerabilities.
- Voters approved more than 20 midterm ballot measures banning or restricting data centers, and candidates who campaigned on restrictions won - the industry's buildout now faces organized political resistance.
- A reported US intelligence memo alleges a Chinese intelligence service recruited a tech executive who grew close to Nvidia CEO Jensen Huang - unverified, with no charges, but enough to draw a congressional briefing request.
The week's briefings
- OpenAI closes the first $11B of its new round at a $1.25T valuation
- Anthropic explains how Claude's memory works - dreaming included
- DeepMind's AI agents find much faster attacks on legacy encryption
- Doubts surface around Anthropic's IPO as AI spending growth stalls
- OpenAI releases GPT-6 Cyber, a security model it rates in its highest risk tier
- Outside researchers: AI has doubled Anthropic's pace of progress
- Meta's Muse agent gets a memory, follow-through, and Spaces
- Claude can now read your bank account - but not touch it
- Google plans a $17B Wyoming data center with its own power plant
- Google backstops a $19B lease for Anthropic's Kentucky data center
- OpenAI tells data-center towns it will pay its own power bills, with no more NDAs
- Moonshot AI releases Kimi K3.1 Pro's weights for anyone to download
- ChatGPT can now make animated GIFs
- Google launches Gemini Omni 1.2 Flash: 15-second AI video, 30% cheaper
- Anthropic signs its first news publisher: the Associated Press
- Tencent launches Hy4, a new open-source heavyweight
- Fal raises $1.3B to run other companies' AI, doubling its value to $16B
- OpenAI says ChatGPT now handles a fifth as many searches as Google
- Anthropic updates its IPO paperwork, and the numbers are striking
- Salesforce agrees to buy an AI call-center company
- Researchers got an AI to edit another AI's brain
- Apple's AI features are finally coming to China
- ChatGPT and Codex will share one usage allowance
- OpenAI wants data centers rebranded as “compute campuses”