Moonshots Recap: Nvidia After Jensen Huang, an AI Takeover Warning, and Demis Hassabis on Self-Improving AI

The latest full episode of the Moonshots podcast on Peter H. Diamandis's official YouTube channel, "Jensen's Departure, NVIDIA's succession plan, AI doomer response, and Demis talks RSI | #311 (Live)", was published on November 13, 2026 and runs about two hours. It was streamed live and recorded the same day. Peter Diamandis hosts with Salim Ismail and Dr. Alexander Wissner-Gross, joined again by Emad Mostaque, founder of Intelligent Internet; Dave Blundin was not on this episode. The panel discusses Jensen Huang's resignation as Nvidia's CEO and what comes next for the company, a stark warning about AI from a member of the OpenAI Foundation board, Google DeepMind CEO Demis Hassabis's comments on recursive self-improvement, Anthropic's new workplace tools, AI and chronic disease, and a run of AI-assisted discoveries in archaeology and astronomy, before closing with audience questions.

This is an independent summary written by GNOUP. It is not affiliated with, sponsored by or endorsed by Moonshots, Peter Diamandis or anyone who appears in the episode. The views, claims and predictions described below belong to the speakers, not to GNOUP. Nothing here is investment, financial, medical or legal advice, and nothing here is a recommendation to buy or sell anything.

Background: On November 12, 2026, Nvidia announced that Jensen Huang had resigned as president and CEO, effective immediately, and that CFO Colette Kress had been named interim CEO. Source: NVIDIA Newsroom. The news reports discussed in this episode about the circumstances of his departure are allegations; as the panel itself notes, no charges had been filed at the time of recording, and we do not repeat personal details here.

Key takeaways

• Why did it happen so fast? Peter frames the opening question as whether Huang self-destructed or was pushed out amid political battles over chip sales to China. Alex calls his view preliminary, but suggests the bigger story may be US efforts to keep advanced chips inside an allied "Pax Silica" sphere and proposed legislation, the Remote Access Security Act, that would make it harder for China to reach GPUs through cloud services in other countries. Salim and Emad both question why Nvidia's board moved so quickly rather than running a longer investigation, and Emad stresses that nothing had been proven.
• Who can replace him? The panel agrees Huang will be very hard to replace. Emad says Huang had around 60 direct reports and deep ties to TSMC, and suggests his checklist for any new CEO: understand Nvidia's CUDA software platform and be able to negotiate with TSMC. Salim calls a CFO running a tech company a bad fit for the long run. Alex says Nvidia needs "founder mode" and expects it may rotate through several CEOs; he and Peter joke about Huang one day running Elon Musk's planned Terafab chip plant. Peter mentions a prediction market giving roughly a 30% chance he returns as CEO by the end of next year.
• Glass half full, glass half empty. Alex's optimistic case is that a post-Huang Nvidia might pull back from "circular financing" of its own customers and so reduce credit risk across the AI market. His pessimistic case is that Nvidia becomes like Apple after Steve Jobs and the "animal spirits" Huang drove into AI investing fade. Emad argues Nvidia is now at a dangerous moment because rivals such as AMD are closer technically, and customers who feared losing Nvidia supply may now diversify.
• An AI takeover warning, and pushback. The panel plays a clip of Paul Christiano, a member of the OpenAI Foundation board and a co-developer of reinforcement learning from human feedback, saying at a Toronto lecture on November 9 that he thinks there is "a high risk of an AI takeover within the next three years." Emad disagrees, saying the real near-term threats are cyberattacks on infrastructure and biological risks, and that he has lowered his own estimate of catastrophe from 50% to 20%. Alex says he does not share those fears and expects humanity to "merge" with superintelligence rather than be replaced. Salim argues serious risk assessments must also count the cost of not using AI, such as diseases left uncured.
• Demis Hassabis on "closed-loop" self-improvement. In a clip, Hassabis distinguishes today's AI-assisted coding, with humans still reviewing the work, from a future "closed loop" where AI improves AI without human review, which he calls the riskiest path and says could produce an "alien" kind of intelligence. Salim welcomes the idea of AI as a different, complementary intelligence. Alex disagrees that it will be alien, citing research that links language-model activity to brain scans. Emad argues human review cannot keep pace with today's model speeds, so closed-loop improvement is effectively already possible.
• Anthropic takes on office software. Peter says that on November 10 Anthropic put Claude Whiteboard and Claude Sheets into beta, made Claude Data generally available and launched Claude Meetings inside Google Meet, all on paid plans, and that Microsoft shares fell about 3.5% on the news. Alex sees office suites as table stakes for every big AI company. Salim points to Microsoft's distribution, compliance and installed base as lasting advantages, and calls Anthropic's new connectors for third-party apps the bigger story.
• AI and chronic disease. In a clip, Anthropic CEO Dario Amodei says diseases such as cancer, Alzheimer's and heart disease could become "relics of the past" in less than a decade. Alex calls a promise to solve disease the "grand bargain" frontier labs can offer the public in return for data centers and approvals. Emad thinks the core mechanisms could be cracked in two to three years, with actual cures taking five to ten years given trials and regulation.

Newsworthy insights

These are claims, predictions and opinions from the episode. GNOUP has not verified them.

• Nvidia's stock move. Peter says Nvidia shares fell about 13% at the open after the resignation news and closed down about 4.3%.
• Should models be "patriotic"? Emad raises whether American AI models should be trained to be patriotic and what values labs are teaching their models. Alex warns against any national, government-approved ideological test for AI, comparing it to rules China applies to its public models, and prefers a diverse, competitive field of models.
• How well do drugs work? Peter claims that an FDA-approved drug may work in only around 20% of the people who take it, as part of an argument that "virtual cells" could make medicine more personal. That figure is his and varies widely by drug.
• Talking to the planet. Peter describes Planet's new Planet AI app, which lets users search years of daily satellite imagery with everyday questions. He says it is free for eligible users, with usage limits.
• AI archaeology. Peter describes an AI researcher who, he says, used 48 AI agents to scan about 1,200 square kilometers of public NASA lidar data overnight and flagged a possible unmapped Maya site in Mexico, which Mexican officials reportedly said appears to be an original finding. Alex predicts a coming era of "computational archaeology."
• Ancient Babylon in eight hours. The panel shows a user-made 3D reconstruction of ancient Babylon around 600 BC that Peter says was built with an Anthropic model in about eight hours. Emad expects people to be able to "visit" historical places in near-photorealism within a couple of years.
• A new dwarf planet candidate. Peter describes a person who, using an AI assistant on Vera Rubin Observatory data, reported what appears to be a dwarf planet roughly 590 kilometers across, about twice as far from the Sun as Pluto. Salim's takeaway is that the telescope had already seen it; the bottleneck was our ability to notice it.
• Audience questions. Salim says UBI experiments have shown promising results but warns that serious unrest before support is in place would be very bad; Peter expects governments to respond with direct payments if job losses mount. Emad expects Nvidia's role as a major investor in AI startups to shrink under new leadership, says Anthropic's expected IPO the following week could bring new money into the sector, and predicts robotics will draw the biggest investment wave of all.

Watch the episode

Jensen's Departure, NVIDIA's succession plan, AI doomer response, and Demis talks RSI | #311 (Live) on YouTube (official channel "Peter H. Diamandis," published November 13, 2026, about 2 hours)

Episode guide

Chapter times are taken from the episode's own YouTube description (some titles lightly reworded). Select a time to jump to that point on YouTube.

• 0:00: Welcome to Moonshots: a Microsoft Outlook outage and Starship launch day
• 4:55: Jensen Huang's departure shakes Nvidia
• 11:08: National security, China and the politics around his exit
• 20:11: Open-source AI, chip restrictions and political pressure
• 27:41: Nvidia's future: who can replace Jensen?
• 39:09: An OpenAI Foundation board member warns of an AI takeover
• 47:08: AI alignment, safety and the risks of inaction
• 57:30: Demis Hassabis on recursive self-improvement
• 1:12:00: Anthropic takes aim at Microsoft with new Claude tools
• 1:21:40: Can AI cure chronic disease within a decade?
• 1:30:00: Planet AI brings satellite intelligence to everyone
• 1:32:10: AI agents flag a possible lost Maya site
• 1:35:50: Recreating ancient Babylon with AI
• 1:39:30: AI-powered astronomy and new discoveries
• 1:44:00: Audience Q&A: UBI, Nvidia and the future of AI

About this summary

This recap was prepared from the episode's title, description, chapter list and the publicly available transcript text. The transcript we could access stopped partway through the audience Q&A, so only the first two questions are summarized. The transcript did not label who was speaking, so some attributions are our best reading, and automatic transcripts can contain errors in names, numbers and phrases. We did not check the figures and claims quoted by the speakers. The opening segment concerns allegations that had not been proven at the time of recording; we left out personal details, the names of other people mentioned in connection with the reports, a partisan TV news clip the panel played, and other political asides. We also did not promote sponsors or paid offers in the episode or its description (metatrends, 6 Forks Report, Google for Startups, Blitzy, Fountain Life and others) or the guests' books and websites. GNOUP has no connection to those sponsors or offers. The show's own note says the views expressed are personal opinions and not financial, medical or legal advice, and the same goes for this summary. For the exact words, watch the episode.