Felix Laurent
The weekend's AI news centered on Nvidia and Anthropic. Two large Nvidia investors asked the board to slow its CEO search and keep Jensen Huang's return on the table. Anthropic heads into its IPO week, with pricing set for Wednesday and trading for Thursday. Below are the stories of Saturday and Sunday, plus Friday news that came after our last digest, and the weekend's rumours.
Nvidia investors ask the board to slow down
Altimeter Capital and Atreides Management asked Nvidia's board for a "slow, measured, and deliberate process" on leadership. Their letter leaves open the potential return of Jensen Huang, CNBC reports. The two funds disclosed about $2.1B of Nvidia stock between them.
Inside the company, a suggestion box post asking the board to weigh staff sentiment drew more than 1,600 comments. Most favor Huang's return, WIRED reported; more in our article.
The board has not said whether it would consider reinstating Huang. CFO Colette Kress remains interim CEO and leads Nvidia's earnings call on Tuesday, November 17.
Anthropic's IPO week begins
Anthropic expects to price its IPO on Wednesday, November 18, and start trading on Nasdaq as ANT the next day. Through the offering and a private placement it could raise up to $90.4B at about a $1.4 trillion valuation. Demand already exceeds the shares on offer several times over, as we reported here.
Speaking with JPMorgan's Jamie Dimon on Friday, Dario Amodei said Anthropic wants to "underpromise and overdeliver". He called buying compute without overspending the company's deepest business dilemma, Yahoo Finance reports.
Each co-founder is expected to sell about $250M of stock in the offering, CNBC reports. The co-founders also made a non-binding pledge to give away 80% of their personal equity wealth.
Smaller updates
Seven smaller stories came out on Friday and over the weekend.
- Anthropic published 10 recommendations for building AI safety evaluations, with numbers from its own testing. We read them here.
- Claude Code on desktop got custom themes that recolor the Code tab and add animated backdrops. Details are here.
- llama.cpp and ggml reached version 1.0.0, with a promise that breaking changes will now bring a new major version. More in our article.
- Google DeepMind released PARDIG and SatWeave, the rainfall and satellite datasets behind WeatherNext 3. We covered them here.
- LangChain, Pipecat, OpenRouter and Vercel added Google's new Gemini Decision 1 model within two days of launch. Our breakdown is here.
- The official Grok account said a soft rollout of Grok 4.8 to Grok Bot users is underway. No benchmarks or model card have been published.
- We also published a guide to running DeepSeek V4.1 Flash with Ollama, which is here.
Four of the seven are tools for developers who build on these models.
Rumours
Three more stories circulated over the weekend, but we could not read the primary report or confirm them with the companies.
- xAI has reportedly cut about 60 people from its Grok Imagine video and image team, The Information reports. Grok Imagine stays available.
- The White House AI task force reportedly urged agencies to ease liability rules so that AI companies report critical incidents, Bloomberg reports.
- Claude Code Projects can reportedly run parallel cloud workers under a coordinator, according to third-party sites. We found no Anthropic announcement.
Treat all three as unconfirmed until the companies announce them.
What the weekend was about
The weekend was a pause before a heavy week. Nvidia's investors and staff pushed the board to move carefully, and the company has its first earnings without Huang on Tuesday.
Anthropic used the same days to set expectations low before Wednesday's pricing. On November 14–15 the two biggest stories of the month waited for their next step, and both arrive within three days.
