November 1: Anthropic junk rating, Meta beats Google, Intel

The weekend was mostly about money: who pays for AI, who profits from it and who takes the hit. Here are the 7 stories to know in 5 minutes.

Anthropic could get a "junk" credit rating

According to the Financial Times, picked up by TechCrunch on October 30, Blackstone (a major investor in a $63 billion deal to finance Anthropic's chips) expects a "junk" rating at the high end of that tier after the company's IPO. A "junk" rating signals higher risk to lenders. It's only a forecast, not an official rating, and both Anthropic and Blackstone declined to comment. Why it matters: borrowing would cost more for the company that makes Claude, and it shows that AI is increasingly being financed with debt. Source: TechCrunch

Meta pulls ahead of Google in advertising

Last quarter, Meta's ad revenue grew 26% ($63.1 billion), compared with 12% for Google Search ($63.4 billion), according to CNBC on November 1. That's the widest gap since 2021. Meta says its AI tools help advertisers; on Google's side, chatbots are pulling some users away from classic search. Why it matters: the way you look for information is changing, and ad money follows. Google is still doing well elsewhere, with a cloud business that more than doubled. Source: CNBC

Intel crashes, and memory prices are to blame

On October 30, Intel's stock lost about 28% after a disappointing quarterly forecast ($15 to $16 billion in expected revenue, versus the $17 billion hoped for). CEO Lip-Bu Tan says customers expect PC sales to slow because of the price and availability of memory, which the AI industry is soaking up in bulk. The U.S. government, which owns about 10%, saw its stake lose $12 billion. Why it matters: AI data centers can push up the price of your next computer. Source: TechCrunch

A Google satellite runs AI in space

Launched in early October, Google's Suncatcher prototype is up and running: on October 30, the company announced that its TPU chip in orbit had gotten the Gemma 4 model to write a haiku. The satellite carries four chips but runs only one at a time, for about twenty minutes, because of limited power (about 1 kW). Two more satellites are planned for 2027; the target for a profitable system would be the 2030s. Why it matters: it's a real test, but very far from replacing a data center on the ground. Source: TechCrunch

xAI will have to answer in court for its turbines' pollution

On Friday, October 30, a federal judge in Mississippi refused to dismiss the NAACP's lawsuit against xAI (part of SpaceX). The civil rights organization accuses the company of running 27 gas turbines without air permits between August and December 2025 to power data centers near Memphis. The judge rejected all three of the defense's arguments, including the one about "mobile" turbines. The trial hasn't happened yet: nothing has been decided on the merits. Why it matters: AI's electricity has a local cost, in the air people breathe. Source: CNBC

Cohere raises up to 7 billion Canadian dollars

According to Le Monde Informatique (a French IT news outlet) on November 1, the Canadian enterprise AI company is announcing a first close of a funding round that could reach up to 7 billion Canadian dollars, at a valuation of 35.7 billion. German group Schwarz is leading the round, and the Canadian government would invest up to 1.6 billion, in exchange for a board seat and investments that stay in Canada. The amount is a ceiling: the final total may differ. Why it matters: governments are now funding their own "sovereign AI" so they don't depend on the American giants. Source: Le Monde Informatique

Mistral puts its AI to work on cybersecurity

Mistral, the French AI startup, has just opened up the weights of its Large 4 model (its parameters, which anyone can download). According to Le Monde Informatique on November 1, Orange Cyberdefense is also getting early access to a specialized version for penetration testing, vulnerability analysis and incident response, with human oversight and computing done in Europe. No rollout date has been given. Why it matters: cybersecurity is one of the first fields where AI really helps, and this is a European AI. Source: Le Monde Informatique