Moonshots Recap: Anthropic's IPO, the Midterms and AI, a 5% Nasdaq Drop, and Faster Self-Improving AI

The latest full episode of the Moonshots podcast on Peter H. Diamandis's official YouTube channel, "Anthropic's IPO in Question, Midterms Hit AI, Nasdaq Drops 5%, and RSI Accelerates | #309", was published on November 7, 2026 and runs about one hour and 46 minutes. The description says it was recorded on November 6. Peter Diamandis was away for this one, so Salim Ismail stepped in as host alongside Dr. Alexander Wissner-Gross and Dave Blundin. The three talk through whether Anthropic's IPO could slip, how this week's election results put data centers on the political map, the sharp drop in tech stocks, a new outside study suggesting AI is speeding up AI research at Anthropic, a paper that scores AI by economic value per unit of energy, a new White House policy on DNA synthesis screening, and a long-exposure photo full of satellite trails.

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.

Key takeaways

• Will Anthropic's IPO be delayed? The cold open asks whether the election fallout pushes back Anthropic's planned listing. Dave doubts it, arguing the company launched its roadshow when prediction-market odds already pointed to this result. Alex says he genuinely does not know: he cites reports that bankers want a lower valuation, but also a possible urge to list before new rules arrive in January. Dave compares it to Facebook's 2012 IPO, which was priced and traded down before recovering strongly over the following years.
• Data centers became an election issue. Salim reads figures presented on the show: that 56% of voters in one measure opposed Big Tech and data centers, that a Goldman Sachs estimate says about 45% of planned US data-center buildout could face restrictions, and that candidates favoring pauses won or led in 12 governor's races. All three hosts say other issues (the war, inflation and tariffs) mattered more, but Dave says he has never seen a technical topic like AI play this big a role in an election. The panel kept this mostly about policy rather than parties.
• Where the buildout goes next. Dave expects construction to shift toward states that welcome it, pointing to Tennessee, and sees Canada's large power supply as an obvious opportunity. Alex argues that local and state resistance strengthens the case for putting compute in orbit, and suggests towns that do host data centers should negotiate real community benefits in return.
• AI still needs a "killer app" for consumers. Alex says the most persuasive thing the AI labs could do is help cure major diseases. Dave thinks a beloved consumer device or assistant will arrive sooner and change public opinion faster. Salim asks viewers to suggest what that app should be.
• The market sell-off. Salim reads the week's moves as reported on the show: the Nasdaq down about 5.5%, the S&P 500 down more than 4%, big chip stocks down about 12% and Nvidia down about 9.3%. Dave's view is that much of this was priced in during the weeks before the election and reflects expected delays to power and data-center projects that were planned but not yet approved, rather than a halt to projects already under construction.
• Is AI already improving itself? Salim describes an 11-week study from METR (an independent AI evaluation group) that, as read on the show, estimates Anthropic's AI development is running at about 1.9 times its historical pace, with roughly 80% of that speed-up attributed to AI helping with AI research. Alex argues recursive self-improvement is already well under way and that software gains alone could keep progress going even if new data-center construction stalled. Dave adds that labs pointing more of their compute at internal research is itself a sign of it.
• Scoring AI by value per watt. A new paper discussed on the show tested 220 professional tasks across 44 occupations and measured economic value per unit of energy. As described, GLM 5.2 produced the highest average value (about $70 per task), while the much smaller Gemma 4 31B model delivered about 2.5 times more value per joule. Alex likes efficiency benchmarks but says dollars in versus dollars out may be the better long-term yardstick. Salim offers a rough back-of-the-envelope comparison between an AI company and New York City; Dave and Alex push back on the math, and the hosts agree the comparison is apples to oranges.
• Screening DNA synthesis. Salim describes a new White House policy that requires federally funded researchers to use approved DNA and RNA synthesis providers, with screening that generally covers sequences of 50 or more building blocks, customer checks and long record keeping. Alex and Dave say controlling physical "printing" steps is a far better approach than restricting what people or AIs can think about or compute.
• A visible ring of satellites. The show ends with a long-exposure image by a New Zealand photographer that, as described, combines hundreds of photos and captures thousands of satellite trails. Alex calls it an early glimpse of a "Dyson swarm" and encourages people to find a dark sky and look up.

Newsworthy insights

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

• A shortage of compute for smaller players. Dave says the largest buyers have reserved most available AI chips and data-center capacity for years, and that one of his firm's pilot clients was turned away by a cloud provider for being too small. He argues a "right to compute" could become a political issue. Alex worries most about a scenario where AI computing becomes regulated like uranium.
• "Continuous release" by early 2027. Asked for a date, Alex predicts that by around early 2027 frontier AI models will update continuously instead of in big named releases, comparing it to how Google stopped announcing search-index updates.
• Measuring AI's real value. Dave says the value his team gets from AI agents is far higher than what it pays for the tokens, which is why revenue-per-gigawatt comparisons can mislead.
• Robots could be next for licensing rules. Alex predicts that if DNA synthesis is regulated this way, general-purpose robots may eventually face similar requirements, such as approved manufacturers and command logs.
• Disclosures. During the episode Alex and Dave mention financial interests in some companies in the AI and space compute areas they discuss.

Watch the episode

Anthropic's IPO in Question, Midterms Hit AI, Nasdaq Drops 5%, and RSI Accelerates | #309 on YouTube (official channel "Peter H. Diamandis," published November 7, 2026, about 1 hour 46 minutes)

Episode guide

Chapter times are taken from the episode's own YouTube description. Select a time to jump to that point on YouTube.

• 0:00: Anthropic's IPO: Will It Be Delayed?
• 3:26: Welcome to Moonshots
• 6:02: Midterm Elections and the Political Backlash Against AI
• 10:20: How Politics Could Reshape AI Infrastructure
• 14:08: The Search for AI's Killer Consumer App
• 18:27: Big Tech's Response to Data Center Restrictions
• 22:14: Moving AI Compute Into Space
• 25:21: Nasdaq Drops 5%: AI's Market Correction
• 31:50: AI, Politics, and the Future of Decision-Making
• 38:00: Anthropic's Accelerating Recursive Self-Improvement
• 47:52: What Happens When AI Reaches Hard Takeoff?
• 59:27: The Future of Continuous AI Breakthroughs
• 1:04:39: Measuring AI's Economic Value Per Watt
• 1:27:00: White House Policy on AI and Biosecurity
• 1:34:15: The Dyson Swarm: Is It Already Becoming Visible?

About this summary

This recap was prepared from the episode's title, description, chapter list and the publicly available transcript text. The transcript did not label who was speaking, so some attributions are our best reading. Automatic transcripts can also contain errors, so some names, numbers and phrases may differ from what was said. We did not check the figures and claims quoted by the speakers. We kept the election discussion to its AI-policy angle, left out partisan jokes and asides, and did not promote sponsors or paid offers in the episode or its description (Abundance Summit, Link Ventures, Google for Startups, Blitzy, Fountain Life, metatrends, 6 Forks Report and others). 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.