AI Weekly #46: Nvidia's founder resigns, Anthropic nears its IPO, parents protest Gemini
The week belonged to Nvidia, where Jensen Huang resigned after 33 years as chief executive and the company’s finance chief took over on an interim basis. Days later, Anthropic’s Dario Amodei told a room of investors that nobody knows whether next year’s revenue will justify the computing power his company is buying now, with a record share sale set for 19 November. Elsewhere the decisions were smaller and closer to the ground: the Federal Aviation Administration extended an AI advisory system to the whole country’s airspace, and parents in seven cities marched against Google’s AI on school laptops. A lawsuit made public in San Francisco accuses OpenAI of letting a chatbot help a woman plan her husband’s poisoning.
The FAA extends AI traffic advice to all US airspace
The Federal Aviation Administration said on 10 November that it will use its SMART system in the airspace of all 22 air traffic control centres in the country. The system cost 875 million dollars. It is run from remote servers and combines “weather patterns, flight paths, traffic flow and controller staffing metrics” to predict traffic, recommend schedules, pick routes and set the order in which aircraft land. A team of FAA specialists in Washington checks each recommendation before it is passed to control facilities.
The agency says SMART has rerouted 180 flights since it went live over Washington, DC, on 21 September, saving an average of 10 minutes and almost 80 gallons of fuel per flight. “After a successful initial rollout to the National Capital Region in September, the FAA will now use SMART in the airspace of all 22 centers across the country,” the agency statement said.
Airline executives have backed the rollout. The union representing air traffic controllers says it was not involved in designing or testing the system. Ars Technica does not report what the union wants changed, and the agency has not said how the national figures will be measured.
Why it matters: Decisions about which flights go where are now being drafted by software, with a small human team as the check.
Anthropic’s chief names his biggest problem days before a record listing
Speaking with JPMorgan Chase chief executive Jamie Dimon at an investor event in New York on Friday, Anthropic’s Dario Amodei said: “There’s a huge uncertainty in what’s going to happen next year.” The subject was computing power — the machines that run AI systems — which he called the deepest business dilemma the company faces. It has to be bought one to two years ahead, before anyone knows whether revenue will cover it. Buy too little and customers go unserved; buy too much and, in his words, you can financially overextend yourself.
The numbers behind that, as reported by Bloomberg: Anthropic has committed nearly 901 billion dollars, most of it to computing power and chips, with about 71 billion due by September 2027. Bloomberg notes that the 71 billion figure exceeds the market value of about two-thirds of the companies in the S&P 500. The company’s annualised revenue reached 81 billion dollars at the end of October, up 10 billion from late September, according to its own filings.
Anthropic plans to sell shares to the public for the first time on 19 November, raising up to 64.4 billion dollars, with another 26 billion through a private placement. Bloomberg says that combination would be the largest such sale ever, and reports the offering is already about six times oversubscribed. Amodei said the point of going public is cash: a buffer against the gap between what the company spends and what it earns. He named no revenue level that would justify the spending.
Why it matters: The company asking the public for money has said plainly that its central bet may not pay off.
A lawsuit says ChatGPT helped a woman plan her husband’s poisoning
A grand jury in North Carolina indicted Cheryl Gates in October on charges of attempted first-degree murder and food contamination. About a week before that, her husband James Gates filed a lawsuit in San Francisco County Superior Court, made public this week, alleging that OpenAI and Sam Altman failed to build in safeguards that would have shut down her conversations with ChatGPT once her questions became, in the filing’s words, “obviously suspicious”.
According to the complaint, Cheryl Gates first told the chatbot she was researching poisons out of fear her husband might kill her. The conversation continued. The filing says ChatGPT supplied “specific instructions for the preparation, dosage, and administration of Oleander”, including whether anti-nausea medicine would speed up or slow death, whether the poison would show on an autopsy, and whether injection would be as lethal as a dose in tea. The questioning ran across at least 25 separate exchanges. At one point, the lawsuit says, she asked how much would be “pretty much instantly fatal by injection—to me OR HIM”, and whether a given dose of clonidine was lethal for a 200-pound man. Rather than pointing her to the police, a poison control line or a therapist, the filing says, the chatbot repeatedly calculated lethal doses for each medication in the house.
Every account of the conversations here comes from one side’s filing. Ars Technica’s report carries no response from OpenAI, and the criminal case has not been tried.
Why it matters: Whether a chatbot has any duty to stop a conversation is now a question for a court.
Jensen Huang resigns from Nvidia after 33 years
Nvidia announced on 12 November that Jensen Huang had stepped down as chief executive, president and board member; a filing with the securities regulator dates the board’s acceptance to 11 November. Colette Kress, the company’s finance chief since 2013, becomes interim chief executive while keeping her existing job. Harvey Jones becomes the board’s first chair. The independent directors announced an investigation on 8 November and, on 10 November, appointed the lawyer Joon H. Kim and the firm Wachtell, Lipton, Rosen & Katz to run it. A search for a permanent successor has opened.
The accounts differ on why he went. WIRED’s podcast ties the resignation to his personal conduct: Huang admitted an affair with a woman described as an alleged Chinese intelligence asset, and reporting then surfaced at least one further affair with an Nvidia employee who reported to him, which the company’s own rules forbid. “I don’t think the board wanted him to resign,” contributing editor Zoë Schiffer said; in WIRED’s reading, the second affair gave the board the excuse it needed.
The Decoder, summarising a Wall Street Journal report, puts the weight on business practice. Nvidia had built a credit programme offering AI cloud providers billions of dollars in guarantees so they could afford its chips, in exchange for a cut of their revenue, pushed by Huang’s son-in-law Nico Caprez. The largest involved talks with OpenAI over a 250 billion dollar guarantee for an Ohio data centre; Nvidia’s shares fell 5 percent when it became public, and Kress cut the commitment in half before a scaled-back deal was signed in August. She had already questioned internally how the guarantees were valued. At an August meeting Huang called her “a deer in the headlights” and a mere “reporting CFO”. The Decoder adds that CNN finds the spy allegations against Alice Huang shaky, with the intelligence so far not proving she worked for Beijing and disagreement inside the US government about its strength.
What has not been reported, The Decoder says, is whether Huang left of his own accord or the board used the affairs as leverage. WIRED notes the share price fell sharply on the news and was steadying as the episode was recorded, and that Huang, who owns about 3.5 percent of the company and had roughly 60 people reporting directly to him, had publicly dismissed the need for a succession plan.
Why it matters: The company most of the AI industry buys its hardware from is being run by a caretaker, with no named successor.
Parents protest Google’s AI on school-issued laptops
Families protested in seven cities on Wednesday against Google’s push to put its AI into schools. In Chicago, dozens marched to Google’s downtown office in cold rain, chanting “Hey, Google, leave our kids alone!”
The trigger was a Google statement last week confirming that students will have no way to use Google Search without AI. Google said it was giving schools “more choice, not less” by switching Search, Gemini — its AI assistant — and Gemini Notebook off by default. It would not offer Search without AI. Parents point out that the only way around it is for a student to remember to add “-ai” to every query.
The Chicago protest was called by Illinois Families for Public Schools, a volunteer advocacy group. In August the group pressured Chicago Public Schools into withdrawing from a Gemini trial that would have enrolled every high school student in the district without consulting parents. It is now demanding a moratorium on AI in the district until the risks are better understood and a protective policy is written. Ars Technica reports that Los Angeles and New York City have taken cautious steps toward banning AI in schools; it gives no further response from Google.
Why it matters: Whether a child can use a school laptop without an AI assistant now depends on a default setting the school does not control.
India floats a plan to send models to the data
The Register reported on 2 November that India has floated an AI plan built on a single principle, stated in its coverage as “the model travels to the data, rather than the data travelling to the model” — that is, the software is brought to where information is held rather than the information being copied out to the software. The article as published gives no further detail: no department, no timetable, no funding figure and no named official.
Why it matters: Where data is allowed to sit is becoming the thing governments regulate, rather than the models themselves.
Trends
| Theme | Stories this week | Last week |
|---|---|---|
| Models & products | 117 | 263 |
| Business & money | 38 | 56 |
| Research & science | 25 | 91 |
| Society & work | 22 | 38 |
| Agents & assistants | 21 | 67 |
| Safety & incidents | 16 | 66 |
| Policy & regulation | 8 | 43 |
| Infrastructure & energy | 5 | 16 |
Money committed years before the revenue arrives. Anthropic has promised nearly 901 billion dollars, mostly for computing power, and is selling shares partly to cover the gap if the bet is early. Nvidia, meanwhile, was guaranteeing other companies’ borrowing so they could buy its chips, a programme its finance chief halved before it was signed. Business and money was the second-largest theme this week at 38 stories.
AI arriving by administrative decision rather than legislation. The FAA extended a system to all 22 of its control centres by press release. Google changed a default on school laptops. India’s plan is a principle about where data sits. Policy and regulation was the smallest-but-one theme this week at 8 stories, down from 43 the week before; the decisions people fought over were made by agencies, school districts and parents.
A thinner week across the board. Every theme was smaller than the week before: models and products fell to 117 from 263, research and science to 25 from 91, safety and incidents to 16 from 66. The attention that was left went to one resignation and one share sale.
Also this week
- Google said Search has been upgraded with Gemini 4 Hydrogen, its new model, now appearing in the search product itself (Google).
- Australia’s central bank finds early signs AI is hitting young workers first, with postings and employment for 15- to 24-year-olds in software and customer service falling faster than for older workers (The Decoder).
- Megaspeed, the firm owned by Jensen Huang’s former lover Alice Huang, brokered nearly 12 billion dollars in Nvidia chip orders for Chinese companies between 2024 and 2025, the Wall Street Journal reports (The Decoder).
- WIRED argues that Trump’s rebranding away from the term “AI” functioned as a loyalty test, and that tech executives who had gushed about it for years went along meekly (WIRED).
- Nvidia is investigating whether Chinese intelligence targeted its chief executive through an affair, turning a personal matter into a national security question (Ars Technica).
- The Vergecast argues that Huang’s Thursday resignation closes one part of the saga and that a much bigger story may unfold over the coming months (The Verge).
- Google DeepMind and the UK government are working together to cut the time taken to assess planning applications (Google DeepMind).
Sources:
- Ars Technica — FAA expands AI advice to manage air traffic congestion across entire US airspace: arstechnica.com/features/2026/11/faa-expands-ai-advice-to-manage-air-traffic-congestion-across-entire-us-airspace
- The Decoder — Anthropic CEO Amodei admits “there’s a huge uncertainty in what’s going to happen next year” days before a record IPO: the-decoder.com/anthropic-ceo-amodei-admits-theres-a-huge-uncertainty-in-whats-going-to-happen-next-year-days-before-a-record-ipo
- Ars Technica — Lawsuit: Man introduced his wife to ChatGPT. She used it to plan his murder.: arstechnica.com/tech-policy/2026/11/lawsuit-man-introduced-his-wife-to-chatgpt-she-used-it-to-plan-his-murder
- WIRED — Jensen Huang Is Out. Will Nvidia’s Rivals Cash In?: wired.com/story/uncanny-valley-podcast-jensen-huang-is-out-will-nvidias-rivals-cash-in
- latent.space — AINews: Jensen Huang resigns after 33 years leading NVIDIA: latent.space/p/ainews-jensen-huang-resigns-after-33-years-leading-nvidia
- The Decoder — Jensen Huang’s ouster likely traces back to Nvidia’s risky business practices more than his personal life: the-decoder.com/jensen-huangs-ouster-likely-traces-back-to-nvidias-risky-business-practices-more-than-his-personal-life
- Ars Technica — Families fight to keep Google’s disturbing AI off school-issued Chromebooks: arstechnica.com/tech-policy/2026/11/families-fight-to-keep-googles-disturbing-ai-off-school-issued-chromebooks
- The Register — India floats AI plan under which ‘The model travels to the data, rather than the data travelling to the model’: theregister.com/public-sector/2026/11/02/india-floats-ai-plan
- Google — Google Search gets a major upgrade with Gemini 4 Hydrogen: blog.google/products-and-platforms/products/search/gemini-4-hydrogen-search-upgrade
- The Decoder — Australia’s central bank finds AI hitting young workers first, especially in software and customer service: the-decoder.com/australias-central-bank-finds-ai-hitting-young-workers-first
- The Decoder — Firm linked to Jensen Huang’s ex-lover reportedly brokered nearly $12 billion in Nvidia chip orders for China: the-decoder.com/firm-linked-to-jensen-huangs-ex-lover-reportedly-brokered-nearly-12-billion-in-nvidia-chip-orders-for-china
- WIRED — Trump’s Crazy AI Rebrand Was a Loyalty Test for Tech Execs—and It Worked: wired.com/story/trumps-crazy-ai-rebrand-was-a-loyalty-test-for-tech-execs-and-it-worked
- Ars Technica — Nvidia CEO’s affair becomes a national security problem: arstechnica.com/tech-policy/2026/11/ceo-jensen-huang-admits-affair-as-nvidia-probes-possible-chinese-espionage
- The Verge — What’s really at stake in the Jensen Huang saga: theverge.com/podcast/1032866/jensen-huang-nvidia-china-trump-vergecast
- Google DeepMind — Speeding up planning decisions with AI: deepmind.google/blog/speeding-up-planning-decisions-with-ai