Peter Diamandis · Oct 28, 2026 · 136 min
Anthropic’s $2T IPO, Elon Predicts Neocloud Death, and China’s Kimi Catches Up | #306
The panel discusses the AI sell-off in neoclouds, the Anthropic IPO and the open-vs-closed model gap. They see the neocloud pullback as a shakeout of badly financed players rather than the end of the AI boom. Anthropic's financials are treated as extraordinary, though its governance structure draws concern. Kimi's near-parity with US frontier models is seen as pressuring US labs. A long physics and science section covers Alex's chain theory and OpenAI's math discoveries.
Where they stand
AI demand and capability are accelerating, and any pain in the neocloud and credit layer is seen as a financing shakeout, not an end to the boom.
High conviction · years
The neocloud sell-off is expected to clear weak, over-levered players while cash-flow-rich firms like SpaceX and Anthropic get stronger, with the Anthropic IPO viewed as fairly priced.
Medium conviction · months
Salim expects Bitcoin, via Lightning, to become the currency of choice for AI agents, with a fuller Bitcoin debate promised.
Low conviction · years
Key takeaways
- Neoclouds like CoreWeave and Nebius sold off 30%+ on debt concerns; Chamath and Elon predict stranded compute in 2027 and 'guppy level death' among the ~200 neoclouds, while SpaceX borrowed $40B at 8% as an investment-grade borrower. 16:13
- The panel's consensus is that this is a financing shakeout and not a technology bubble pop: compute is sold out for years, though Alex flags a possible demand shock from RSI efficiency gains or missing killer apps. 17:52
- Anthropic's IPO (week of Nov 16, raising ~$100B at $1.5-2T) is called strong: revenue of $16.3B in H1 2026, a first operating profit, and $28 revenue per compute dollar; risks are $500B+ compute commitments and customer concentration. 32:21
- Anthropic's Long-Term Benefit Trust class T shares are debated; Alex and Imad see ideological-control and fiduciary risk, and Imad argues Anthropic should have stayed private. 37:21
- Kimi K3.1 Pro from Moonshot AI is said to be about five weeks behind the US frontier, and open weights will arrive next week; the panel sees this as forcing US labs to compete. 55:15
- GPT 6.1 Terra launched at under half the prior price with a 50-point drop in hallucinations, and the panel says price-performance is doubling roughly every 7 weeks. 1:04:40
- The panel argues Europe is falling behind in power and compute, while the US faces a power gap with China. 7:49
- Alex's chain theory derives the fine structure constant, and OpenAI announced three math, computer science and physics results; the panel says verification is now the bottleneck. 1:10:50
Calls and positions
- Elon Musk says debt markets have shut off funding for companies without substantial non-data-center revenue, and the roughly 200 neoclouds will see guppy-level death and consolidation.
“the 200 neoclouds are going to see guppy level death and consolidation”
- A bubble is likelier to come from a demand surprise, with killer applications not arriving in time to justify the compute spend.
“I would bet on a demand surprise where all of this new comput is coming online”
- Stranded compute is expected in 2027, making it a turbulent year that clears out bad bets.
“2027 will be a turbulent year because it will clear out a bunch of bad bets.”
- After the Anthropic, OpenAI and SpaceX IPOs, San Francisco becomes the overwhelming financial capital of the world, with the money reinvested into AI.
“After these IPOs, San Francisco is overwhelmingly the financial capital of the entire world.”
- Salim expects Bitcoin, now scalable through Lightning, to become the currency of choice for AI agents.
“I expect Satoshi's to be the uh uh currency of choice for AI agents over time.”
- The US-China gap in power generation is expected to widen to almost 10 to 1.
“the gap in power between the US and China is going to expand almost 10 to1.”
Views expressed
- Compute is sold out for years, so anyone with a viable data center will do well unless they signed bad debt deals.
“all all compute is sold out for years. So anyone with a viable data center is going to do incredibly well.”
- Imad advises staying away from RAM stocks, saying the market is frothy and new architectures will cut memory needs.
“I would be nowhere near RAM stocks right now.”
- Anthropic at a $2T valuation is correctly priced and a cash machine, and Dave would buy into the IPO for exposure to the category.
“it's correctly priced. Uh it's growing faster than anything we've ever seen.”
- Imad argues Anthropic should not be a public company, since it has ample cash and will not care about its share price.
“Anthropic should not be a public company.”
- Alex finds the long-term benefit trust governance concerning and compares it to ideological control, warning it could degenerate into ideological control.
“I actually think is potentially concerning.”
- Open-weight Chinese models like Kimi K3.1 Pro may have passed the US public frontier on performance per cost, which keeps US labs competing.
“it's actually past the optimal frontier surface of Opus 5.5.”
- Energy is an incredible investment theme with no end in sight as AI value rises.
“energy is an incredible investment theme uh with no end in sight.”
Companies, projects and topics mentioned
- Neoclouds
- CoreWeave
- SpaceX
- AI capex
- Nebius
- Micron Technology
- Anthropic
- OpenAI
- Moonshot AI
- Bitcoin
- Power and energy for AI
- China
More from Peter Diamandis
- SpaceX’s AI Business Up 830%, OpenAI Agents Lie, and Startup Turns Mercury Into Gold | EP #305Oct 23, 2026
- Chamath: AGI Is Here, Anthropic’s $100B Giveaway, Elon’s SpaceX Shares for 72M Kids | EP #304Oct 21, 2026
- Flying Tesla Car, Robot Brains Halving in Cost Every 3 Weeks, and Why Incentives Decide AI AlignmentOct 17, 2026
- AI Solves 722 Math Problems at Once, $1.8B Race to Build Virtual Cell, Elon Sees Quadrillion SpaceXOct 15, 2026
Summary generated by AI from the video's captionsand checked against it; quotes are verbatim. Processed Oct 29, 2026.
Read on Risk summarises what other people said in public videos and podcasts. Summaries are generated by AI and checked by code against the transcript, but can still be wrong; follow the source links before relying on anything. This is not investment advice and nothing here is a recommendation to buy or sell. Disclaimer ·How this works