Weekly Intel - 2026-11-15

This week Anthropic set the terms for its IPO, Jensen Huang resigned from Nvidia, and Goldman Sachs and United Airlines both put new limits on who gets to use AI tools and for what.

AI Industry Moves

Anthropic’s IPO is set for next week. Anthropic plans to list Class A shares on Nasdaq under the ticker ANT, with pricing expected November 18 and trading November 19. The last reported range was $70 to $78 a share, for a valuation of $1.3 trillion to $1.4 trillion, down from the roughly $2 trillion rumored in September (Globe and Mail). Bloomberg reported the offering was about six times oversubscribed (Reuters). Goldman Sachs and Morgan Stanley are the lead underwriters.

Nvidia employees push to bring back Jensen Huang. Nvidia disclosed in a November 12 filing that Jensen Huang resigned as president, CEO and director, and CFO Colette Kress is now interim CEO. An internal post urging the board to weigh employee sentiment has drawn more than 1,600 comments, many of them asking for Huang’s return, according to WIRED. An outside petition had about 55,000 signatures by early Saturday, about 3,600 of them from people who say they’re current employees (24/7 Wall St.). The board has opened an independent review.

AI & Software

Anthropic releases Claude Fable 5.2 and Mythos 5.2. Anthropic released both models on November 9 with gains in coding, reasoning and vision, and output up to 30% faster than the 5.1 versions. In prerelease testing, about 1,000 Fable 5.2 agents formalized the proof of Fermat’s Last Theorem in 30 hours; Anthropic’s earlier formalization took 11 days. Fable and Mythos are the same underlying model. Fable carries cybersecurity and biology safeguards and is open to all users, while Mythos goes to vetted organizations through Anthropic’s Cyber and Life Sciences Verification Programs, at $8 per million input tokens and $40 per million output tokens for both.

Goldman Sachs pulls Claude Code from its bankers and traders. Goldman Sachs has limited Anthropic’s Claude Code and Microsoft’s GitHub coding tools to its roughly 12,000 engineers and its strats, after an October update to internal guidance that followed a review of productivity and risk, Bloomberg reported. Most traders and bankers lost access, though Goldman’s in-house AI assistant stays available across the firm. The bank removed Claude for its Hong Kong bankers in April.

Amazon replaces Sonnet 4.6 with Strands Decider for ticket labeling. Two Amazon engineers swapped Claude Sonnet 4.6 for Strands Decider, a decision model launched about a month ago that returns a probability for each option instead of generating text, to sort internal tickets into 20 categories. The post says it performed as well as Sonnet at 98% lower cost, about $0.0005 per ticket against $0.0222. The engineers expect the change to save millions of dollars in token costs over a year.

Google launches Gemini 4 Hydrogen. Google released Gemini 4 Hydrogen on November 12, a fast model for everyday coding and agent work, across the Gemini API, AI Studio, Gemini for business, the Gemini app and Google Search. It scores 51 on the Artificial Analysis Intelligence Index at its default setting (54 at its highest), against 50 for Google’s frontier Gemini 4 Argon. Pricing is flat across the 1 million-token context window at $1.50 per million input tokens and $7.50 per million output tokens.

Privacy & Governance

Google says connected Gmail and Drive content can reach human reviewers. Google added an October 16 update to a 2023 post that had said Gmail, Docs and Drive content connected to its chatbot “is not seen by human reviewers.” The update says data from newer features that connect Workspace apps to other Google services “can also be subject to human review in limited circumstances.” Google’s Gemini help page says reviewed data can include summaries and excerpts of relevant emails and files, and that in some cases “the summary may just be the email or file.” A researcher’s writeup says the change covers personal Google accounts, while Workspace business, enterprise and education accounts stay under separate data terms.

Transportation

United Airlines bans AI agents from booking flights. United said on November 11 that it’s barring AI agents from making reservations for travelers, on its own platforms and through travel agencies and aggregators. The airline said sharing account credentials with third-party AI assistants can expose sensitive information and lead to unintended purchases, account changes or use of MileagePlus miles. It also said using an agent to check in could violate federal rules, including the requirement that travelers confirm in person that their luggage holds no banned items.

Infrastructure

Canada launches its Real-Time Rail payment system. Payments Canada launched the Real-Time Rail on November 9, letting financial institutions clear and settle payments instantly, 24 hours a day, every day of the year. Canada is the first G7 country to launch a real-time payment system with mandatory centralized fraud services, including confirmation of payee. Eight participants went live at launch, and all major Canadian financial institutions are set to join over the coming months.

That’s what I’m watching. What caught your attention this week?

-Eric