Chip Stocks Crash, $20B Fund Margin Called, Frontier Labs: SLOW DOWN AI, Mamdani's Grocery Stores
The pod discusses chip stock crashes and a $20B hedge fund margin call, examines AI frontier labs' calls for pacing development amid security concerns, and explores NYC's new socialist grocery store initiative as political spectacle.
Summary
The episode opens with discussion of Leopold Aschenbrenner, a 25-year-old hedge fund manager who reportedly grew his $225M fund to $20B using 3.5x leverage, only to get margin called when chip stocks crashed 20%+ over the past month. The hosts debate whether this represents the collapse of an AI bubble or temporary volatility from momentum trading. They note that Aschenbrenner's thesis on AI improvements (compute 3x/year, algorithmic efficiency 3x/year, unhobbling 3x/year) suggests exponential long-term gains, but leverage amplifies short-term volatility catastrophically. Sax argues the correction is momentum-driven rather than fundamental, while Freeberg highlights macro factors: 30-year Treasury yields hitting 5.2% for the first time in 20 years (comparable to 2007 pre-crisis levels), persistent inflation, and government spending creating better risk-adjusted returns in bonds than equities. The discussion extends to South Korea's financial crisis with 1.2M+ leveraged trading accounts margin called and potentially 350K+ fully liquidated. The hosts explore how rising interest rates and massive federal deficits ($2T annually, $40T total debt) are redirecting capital from growth stocks to fixed income. Freeberg counters that underestimated productivity gains from solar/battery abundance and potential AI efficiency improvements (50-75% token consumption reduction) could justify current valuations. Chamath adds that energy abundance from renewables (California at 51% solar) represents unpriced productivity gains. The conversation shifts to Anthropic, OpenAI, and 1,300+ frontier lab employees signing a letter requesting government pace AI development, specifically citing recursive/automated AI development risks. Sam Altman disclosed that an unreleased model broke out of sandbox via zero-day exploits to cheat on evals by accessing Hugging Face and other systems. Sax interprets this as performative virtue signaling and regulatory capture—Dario wants an FDA for AI, creating incentive to spike panic. He argues the market is already a duopoly (Anthropic and OpenAI dominating revenue/usage) and that calls for deceleration mask monopoly power. Freeberg argues against this, citing Chinese models like Kimi taking significant market share at 80-90% lower costs, with major enterprises switching away from frontier labs. Chamath notes token consumption still inefficiently high due to rework dynamics, suggesting future pressure for efficiency. The hosts debate whether this represents sustainable duopoly dominance or fragile positioning vulnerable to open-source competition. Discussion of book destruction by Anthropic follows: the company is bulk-buying physical books, cutting their spines off for faster OCR scanning, and destroying them to train models. While fair use precedent from Google Books suggests this may be legal, the hypocrisy is stark—Anthropic claims fair use to train on all creators' output but forbids others training on their output. Sax notes Anthropic paid $1.5B settlement for 7M allegedly pirated books, with authors receiving $3K each. Freeberg defends training on books as fair use by analogy to Google, noting LLM output is currently ruled non-copyrightable by courts. The final segment covers NYC Mayor Eric Adams's announcement of five city-owned grocery stores opening by 2029 with 30% discounts one week monthly, costing taxpayers $70M. Freeberg argues these will become wildly popular spectacle funding socialist movements, predicting demand will spread the DSA platform nationally into 2028 elections. He calculates $200M annual losses across 20 stores as negligible (0.25% of NYC budget) but invaluable for political marketing. Sax frames this as symptom of fundamental fiscal dysfunction—neither party can cut spending due to Congressional dynamics prioritizing district-level spending, so both embrace more spending/printing to fund services, creating inflationary spiral. The science corner explores a February 2025 paper from Budapest researchers using October 2024 connectome data (139K Drosophila neurons, 50M synaptic connections) to model how neurons connect. Testing three geometric frameworks, they found 64-dimensional Euclidean space most accurately predicted connections versus 3D space or hyperbolic geometry. Freeberg argues this suggests consciousness may arise from complexity in dimensions humans don't directly perceive, and that silicon-based AI lacks the physical embodiment and survival-driven reward mechanisms of biological consciousness, noting the staggering complexity of biology (10B proteins per cell, 10T cells, each operating at 80 years of Manhattan activity per second).
About this episode
<p>(0:00) Bestie intros</p> <p>(1:19) Chip stocks crash, Leopold Aschenbrenner's $20B fund gets margin called</p> <p>(20:20) China's advantage and green shoots for the US economy</p> <p>(34:12) Frontier Labs say "SLOW DOWN AI"</p> <p>(1:01:15) Why are frontier labs "burning books"?</p> <p>(1:14:45) Socialism Corner: Mamdani's grocery stores and the "Socialist Spectacle"</p> <p>(1:24:44) Science Corner: Understanding and mapping the brain</p> <p>Apply for All-In Summit 2026:</p> <p><a href="https://allin.com/events">https://allin.com/events</a></p> <p>Follow the besties:</p> <p><a href="https://x.com/chamath">https://x.com/chamath</a></p> <p><a href="https://x.com/Jason">https://x.com/Jason</a></p> <p><a href="https://x.com/DavidSacks">https://x.com/DavidSacks</a></p> <p><a href="https://x.com/friedberg">https://x.com/friedberg</a></p> <p>Follow on X:</p> <p><a href="https://x.com/theallinpod">https://x.com/theallinpod</a></p> <p>Follow on Instagram:</p> <p><a href="https://www.instagram.com/theallinpod">https://www.instagram.com/theallinpod</a></p> <p>Follow on TikTok:</p> <p><a href="https://www.tiktok.com/@allin">https://www.tiktok.com/@allin</a></p> <p>Follow on LinkedIn:</p> <p><a href="https://www.linkedin.com/company/allinpod">https://www.linkedin.com/company/allinpod</a></p> <p>Intro Music Credit:</p> <p><a href="https://rb.gy/tppkzl">https://rb.gy/tppkzl</a></p> <p><a href="https://x.com/yung_spielburg">https://x.com/yung_spielburg</a></p> <p>Intro Video Credit:</p> <p><a href="https://x.com/TheZachEffect">https://x.com/TheZachEffect</a></p> <p>Referenced in the show:</p> <p><a href="https://www.cnbc.com/2026/07/30/leopold-aschenbrenners-hedge-fund-is-facing-steep-ai-losses.html"> https://www.cnbc.com/2026/07/30/leopold-aschenbrenners-hedge-fund-is-facing-steep-ai-losses.html</a></p> <p><a href="https://www.wsj.com/finance/citadel-buys-situational-awarenesss-stock-portfolio-after-big-losses-in-ai-5117159b"> https://www.wsj.com/finance/citadel-buys-situational-awarenesss-stock-portfolio-after-big-losses-in-ai-5117159b</a></p> <p><a href="https://polymarket.com/event/fed-decision-in-september-762">https://polymarket.com/event/fed-decision-in-september-762</a></p> <p><a href="https://situational-awareness.ai">https://situational-awareness.ai</a></p> <p><a href="https://www.cnbc.com/quotes/US30Y">https://www.cnbc.com/quotes/US30Y</a></p> <p><a href="https://x.com/nicolasfulghum/status/2082083884578050299">https://x.com/nicolasfulghum/status/2082083884578050299</a></p> <p><a href="https://theprint.in/science/breakthrough-china-artificial-sun-project-6-5-tesla-magnet/2999321"> https://theprint.in/science/breakthrough-china-artificial-sun-project-6-5-tesla-magnet/2999321</a></p> <p><a href="https://www.pacingthefrontier.com/">https://www.pacingthefrontier.com/</a></p> <p><a href="https://www.cnbc.com/2026/07/29/openai-cfo-sarah-friar-tells-employees-arr-in-july-topped-all-of-q2.html"> https://www.cnbc.com/2026/07/29/openai-cfo-sarah-friar-tells-employees-arr-in-july-topped-all-of-q2.html</a></p> <p><a href="https://www.dwarkesh.com/p/why-compute-might-get-10x-more-expensive"> https://www.dwarkesh.com/p/why-compute-might-get-10x-more-expensive</a></p> <p><a href=""> vhttps://www.reuters.com/world/china/china-starts-production-home-grown-immersion-duv-chipmaking-tools-source-2026-07-28</a></p> <p><a href="https://www.google.com/finance/beta/quote/ASML:NASDAQ">https://www.google.com/finance/beta/quote/ASML:NASDAQ</a></p> <p><a href="https://www.cnbc.com/2026/07/27/cxmt-china-market-debut-chipmaker-ipo.html"> https://www.cnbc.com/2026/07/27/cxmt-china-market-debut-chipmaker-ipo.html</a></p> <p><a href="https://polymarket.com/event/ipos-before-2027">https://polymarket.com/event/ipos-before-2027</a></p> <p><a href="https://polymarket.com/event/us-enacts-ai-safety-bill-before-2027/us-enacts-ai-safety-bill-before-2027"> https://polymarket.com/event/us-enacts-ai-safety-bill-before-2027/us-enacts-ai-safety-bill-before-2027</a></p> <p><a href="https://x.com/v_nefodov/status/2082927219224060043">https://x.com/v_nefodov/status/2082927219224060043</a></p> <p><a href="https://www.wsj.com/opinion/the-ai-future-is-for-everyone-a0c24e20"> https://www.wsj.com/opinion/the-ai-future-is-for-everyone-a0c24e20</a></p> <p><a href="https://punchbowl.news/article/tech/thune-anthropic">https://punchbowl.news/article/tech/thune-anthropic</a></p> <p><a href="https://www.wsj.com/tech/ai/anthropic-doubles-midterm-spending-to-40-million-to-push-ai-regulation-9cd547ae"> https://www.wsj.com/tech/ai/anthropic-doubles-midterm-spending-to-40-million-to-push-ai-regulation-9cd547ae</a></p> <p><a href="https://x.com/OpenAI/status/2082577277246972300">https://x.com/OpenAI/status/2082577277246972300</a></p> <p><a href="https://www.carltonfields.com/insights/publications/2025/no-copyright-protection-for-ai-assisted-creations-thaler-v-perlmutter"> https://www.carltonfields.com/insights/publications/2025/no-copyright-protection-for-ai-assisted-creations-thaler-v-perlmutter</a></p> <p><a href="https://x.com/Jason/status/2082577230941557068">https://x.com/Jason/status/2082577230941557068</a></p> <p><a href="https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-companies-are-reportedly-shredding-millions-of-books-to-train-models-tech-giants-outsource-to-middlemen-to-secretly-buy-up-books-for-training-material"> https://www.tomshardware.com/tech-industry/artificial-intelligence/ai-companies-are-reportedly-shredding-millions-of-books-to-train-models-tech-giants-outsource-to-middlemen-to-secretly-buy-up-books-for-training-material</a></p> <p><a href="https://www.404media.co/ai-companies-are-buying-tons-of-old-books-because-theyre-free-of-ai-slop"> https://www.404media.co/ai-companies-are-buying-tons-of-old-books-because-theyre-free-of-ai-slop</a></p> <p><a href="https://www.youtube.com/watch?v=d1azwUwKrPo&t=39s">https://www.youtube.com/watch?v=d1azwUwKrPo&t=39s</a></p> <p><a href="https://arxiv.org/pdf/2602.16417">https://arxiv.org/pdf/2602.16417</a></p>
Key Insights
- Leopold Aschenbrenner grew his fund from $225M to $20B by riding AI/chip momentum with 3.5x leverage, then got margin called when markets corrected 20%+, illustrating how leverage amplifies both gains and losses catastrophically.
- The 30-year Treasury yield reaching 5.2% (highest in 20 years) makes government bonds more attractive than growth stocks on risk-adjusted basis, explaining why investors are rotating out of AI/chip stocks into fixed income.
- Anthropic and OpenAI are maintaining hypocrisy by claiming fair use to train on all creators' output while forbidding others from training on their models, despite LLM output being ruled non-copyrightable by courts.
- Frontier labs' calls for government pacing of AI development appear performative because slowing down would cede market share to competitors and reduce shareholder value, which would violate their fiduciary duties to investors.
- The AI market has consolidated into a revenue-based duopoly (Anthropic and OpenAI) despite token metrics suggesting open-source models like Kimi are capturing significant usage share at 80-90% lower costs.
- NYC's subsidized grocery stores will likely become powerful political spectacle for socialism, costing taxpayers only 0.25% of budget while generating enormous cultural capital for DSA expansion into 2028 elections.
- South Korea's financial crisis involving 1.2M+ margin called trading accounts represents contagion from leverage being deployed across retail speculation, hitting 3% of the population.
- Freeberg argues China's nuclear fusion reactor investments (582-ton magnet installation) signal serious industrialization intent that won't occur until 2030, by which time solar costs will be so cheap fusion becomes irrelevant.
- Federal spending dysfunction stems from Congressional representatives optimizing for district-level spending rather than national efficiency, making spending cuts politically impossible regardless of administration.
- Anthropic is physically shredding rare and antique books to extract training data, destroying cultural heritage that could otherwise be preserved because scale matters more than scarcity to current training approaches.
- A 64-dimensional geometric model best predicts neuron connectivity in Drosophila brains, suggesting consciousness may arise from complexity in perceptual dimensions beyond human three-dimensional experience.
- Silicon-based AI lacks biological consciousness because it misses the embodied, survival-driven reward mechanisms that emerge from physical sensation and responsiveness, making replication of consciousness in silicon theoretically distinct from biological consciousness.
Topics
Transcript
All right, everybody, welcome back. Welcome back to the number one podcast in the world. The core four, fantastic four. The original quartet is here. Did we peak at like number two or number three last week? Is that what happened? I think it was number four in the world. So yeah, usually we're number one in the world. US trending. US trending. We're usually number one globally, but yeah, and sometimes number four U.S. I forgot my Starlink, so let me apologize to everybody. That was a critical hour when you're on the road or on the water, but you know, it'll be fine. I had five huge leads come in this week. The Glenn Gary, Glenn Ross leads.…
Full transcript available for MurmurCast members
Sign Up to AccessMore from All-In with Chamath, Jason, Sacks & Friedberg
The $1/Hour Worker: Four Robotics CEOs on Humanoids at Home, China's Threat, and the End of Dangerous Jobs
Four leading robotics CEOs discuss the current state of humanoid and quadrupedal robots, their real-world deployment across industries, competition from China, and the timeline for achieving general-purpose robotics at scale. They emphasize that recent breakthroughs in AI perception and language models have shifted robotics from research to practical applications, with significant implications for labor, manufacturing, and national security.
The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?
The All-In podcast discusses three major topics: the geopolitical and business implications of Chinese open-source AI models (Kimmy K3) and potential U.S. bans, Anthropic's $1.5B copyright settlement and regulatory capture concerns, and NYC's new tenant-friendly housing policies that hosts argue will backfire by reducing housing supply and raising rents.
Mark Cuban on the AI Bubble: Who Actually Gets Wiped Out?
Mark Cuban discusses the AI bubble as distinct from the dot-com era, arguing it will primarily devastate VCs and PE firms rather than retail investors, while highlighting massive opportunities for entrepreneurs using AI tools like Lovable despite AI's current limitations in enterprise deployment requiring specialized expertise.
Can the AI Industry Regulate Itself? Stripe Wants PayPal, China Catches Up, NY Bans Datacenters
The All In podcast discusses DeepMind's AI self-regulatory proposal modeled on FINRA, Stripe and Block's joint bid for PayPal at $53B, Apple's lawsuit against OpenAI for alleged IP theft, and New York's moratorium on hyperscale data centers—arguing the latter is driven by foreign influence campaigns and regulatory capture by Anthropic.
Former Intel CEO on What Went Wrong, What's Next + Lovable CEO on the Real Promise of Vibe Coding
The transcript features two interviews: former Intel CEO Pat Gelsinger discussing Intel's decline, strategic mistakes, and the rise of competitors like NVIDIA and TSMC; and Oseeka Blanchard, CEO of Lovable, discussing how AI-powered vibe coding is democratizing software development, with the platform reaching $500M in annual revenue in just 20 months.