MurmurCast
DiscussionOpinion

The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?

All-In with Chamath, Jason, Sacks & Friedberg1h 33m

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.

Summary

The episode opens with extensive discussion of Kimmy K3, a Chinese open-source AI model performing on par with leading American models like Claude Opus while costing significantly less. The White House is reportedly considering banning Chinese open-source models, though administration officials state no final decision has been made. Sachs argues that banning open-source would be devastating to American competitiveness and that Anthropic should enforce its own terms of service rather than seek government protection. Chamath emphasizes that models are being commoditized faster than expected, with 50% of tokens now coming from open-source alternatives. The hosts debate whether AI model value will remain concentrated in frontier labs or diffuse to infrastructure providers (cloud, chips) and application layers, with Chamath predicting significant margin compression for closed model providers. Freeberg draws parallels to the internet era, arguing open-source AI will proliferate value across the economy like Apache and Firefox did for the web.

The discussion of distillation—training on model outputs rather than stealing weights—becomes central. The hosts clarify this is standard industry practice used by Google, OpenAI, Anthropic, and Chinese labs alike, not IP theft. They criticize Anthropic for applying terms selectively: claiming distillation by Chinese companies is IP theft while maintaining their own right to train on all global outputs. Sachs notes Anthropic hasn't even purchased KYC (Know Your Customer) controls that could prevent unauthorized account creation, suggesting they haven't prioritized stopping distillation despite claiming it's an industrial-scale threat.

Anthropnic's $1.5B copyright settlement is analyzed as potentially self-defeating. While the settlement involved pirated books from LibGen, Freeberg notes Anthropic could have defended fair use if they'd purchased one copy of each book. The hosts highlight hypocrisy: Anthropic claims fair use allows training on all content creators' works, yet argues Chinese distillation of their outputs violates IP law. Sachs warns this positions Anthropic poorly for future lawsuits and suggests they should have framed the issue as preventing deceptive business practices (fake accounts) rather than claiming IP theft.

Google and Tesla's earnings reports show massive CapEx investments ($195-205B for Google, 25B for Tesla), though both posted negative free cash flow—a first for Google since IPO. The hosts are largely bullish on Google's infrastructure strategy, noting a 32% average ROIC over 25 years and that fragmented AI models benefit cloud providers and semiconductor suppliers. Chamath argues the real value capture will shift from frontier models to infrastructure and applications as models commoditize.

The final major segment critiques NYC Mayor Eric Adams' anti-eviction and rent-control policies. Freeberg delivers a principled argument about private property rights as foundational to American liberty, citing John Quincy Adams' warnings that loss of property rights leads to tyranny. Sachs points out these policies harm other tenants (through building neglect) and working-class residents who depend on public spaces and housing, not just landlords. Both hosts note that permitting reform and housing supply increases would better address affordability than rent controls, pointing to Austin, Tokyo, and Argentina's Milei reforms as evidence. They argue bans on credit checks and income verification will force landlords to raise rents further and leave apartments vacant rather than accept unvetted tenants unable to pay.

About this episode

<p>(0:00) Bestie intros!</p> <p>(0:18) The fight to save open source AI: Kimi K3 panic, Anthropic/OpenAI regulatory capture</p> <p>(27:38) Anthropic/OpenAI historic growth rates, China's long game</p> <p>(48:29) Anthropic's $1.5B piracy settlement and the great IP theft hypocrisy</p> <p>(1:07:12) Google and Tesla stocks tank on surging AI capex</p> <p>(1:17:19) Socialism Corner: "Evictions = Violence" and the threat to private property rights</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://artificialanalysis.ai/models/kimi-k3?intelligence=artificial-analysis-intelligence-index#intelligence-comparisons"> https://artificialanalysis.ai/models/kimi-k3?intelligence=artificial-analysis-intelligence-index#intelligence-comparisons</a></p> <p><a href="https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi">https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi</a></p> <p><a href="https://www.wired.com/story/the-white-house-is-trying-to-figure-out-what-to-do-about-chinese-ai"> https://www.wired.com/story/the-white-house-is-trying-to-figure-out-what-to-do-about-chinese-ai</a></p> <p><a href="https://x.com/mkratsios47/status/2079933645888880708">https://x.com/mkratsios47/status/2079933645888880708</a></p> <p><a href="https://polymarket.com/event/us-government-bans-an-open-source-ai-model-in-2026-20260703221501747"> https://polymarket.com/event/us-government-bans-an-open-source-ai-model-in-2026-20260703221501747</a></p> <p><a href="https://fortune.com/2025/01/29/deepseek-openais-what-is-distillation-david-sacks"> https://fortune.com/2025/01/29/deepseek-openais-what-is-distillation-david-sacks</a></p> <p><a href="https://stratechery.com/2026/whos-afraid-of-chinese-models">https://stratechery.com/2026/whos-afraid-of-chinese-models</a></p> <p><a href="https://x.com/CommerceGov/status/2080341953086886387">https://x.com/CommerceGov/status/2080341953086886387</a></p> <p><a href="https://x.com/tickerplus/status/2080123562560504240">https://x.com/tickerplus/status/2080123562560504240</a></p> <p><a href="https://blog.tickertrends.io/p/anthropic-vs-openai-arr-tracking">https://blog.tickertrends.io/p/anthropic-vs-openai-arr-tracking</a></p> <p><a href="https://x.com/sama/status/2077817060068057493">https://x.com/sama/status/2077817060068057493</a></p> <p><a href="https://www.reuters.com/world/us-judge-approves-anthropics-15-billion-settlement-copyright-lawsuit-2026-07-20"> https://www.reuters.com/world/us-judge-approves-anthropics-15-billion-settlement-copyright-lawsuit-2026-07-20</a></p> <p><a href="https://www.anthropic.com/news/detecting-and-preventing-distillation-attacks"> https://www.anthropic.com/news/detecting-and-preventing-distillation-attacks</a></p> <p><a href="https://techstartups.com/2026/07/22/nearly-200-silicon-valley-startups-urge-trump-not-to-ban-chinese-ai-models-warn-it-could-kill-innovation"> https://techstartups.com/2026/07/22/nearly-200-silicon-valley-startups-urge-trump-not-to-ban-chinese-ai-models-warn-it-could-kill-innovation</a></p> <p><a href="https://x.com/typesfast/status/2080339052398891244">https://x.com/typesfast/status/2080339052398891244</a></p> <p><a href="https://www.thefp.com/p/why-new-york-city-has-50000-ghost">https://www.thefp.com/p/why-new-york-city-has-50000-ghost</a></p>

Key Insights

  • Sachs argues that despite Anthropic's claims of national security threats from Chinese distillation, the company's internal forecasts show accelerating growth from $10B to $70B+ ARR, suggesting competitive harm is overstated and regulatory capture is the actual motivation.
  • Chamath claims that open-source models reaching 95% capability parity for most tasks means frontier model providers face inevitable margin compression, similar to how commodity products historically lose pricing power regardless of regulatory intervention.
  • Freeberg distinguishes between stealing model weights (actual copyright infringement) and training on model outputs (benchmarking), arguing the latter is standard practice across industries and cannot be legally restricted without precedent.
  • Sachs contends Anthropic's distillation allegations are hypocritical because Anthropic itself maintains it can legally train on all global content (including the New York Times) without creator consent under fair use, making it a double standard to claim Chinese companies doing the same commits IP theft.
  • The hosts observe that Anthropic could easily implement KYC (Know Your Customer) authentication to prevent the account creation used for distillation, but chooses not to because it would slow revenue growth, undermining claims that distillation is an existential threat requiring government intervention.
  • Freeberg argues that if the U.S. bans open-source AI access while China remains unrestricted, American enterprises will face cost disadvantages that tank stock valuations and competitiveness, as they'd be forced to pay 50-100x premiums versus global competitors.
  • Sachs emphasizes that banning American access to Chinese open-source while allowing Chinese access to American models doesn't solve distillation; it only ensures China continues extracting knowledge while America loses its open ecosystem advantage.
  • The hosts note that American startups like Cursor and Thinking Machines have successfully bootstrapped derivative models using Chinese open-source as a base, then trained on proprietary data, demonstrating that open-source enables American innovation that could be destroyed by government restrictions.
  • Freeberg frames NYC's no-credit-check, no-income-verification tenant policies as economically irrational, since landlords will respond by requiring full-year prepayment at 3-4x higher rents, making housing less accessible rather than more accessible.
  • Sachs argues that preventing evictions without diligencing tenant ability-to-pay creates moral hazard where tenants have zero incentive to pay rent, causing landlords to compensate through higher rents on all tenants, harming working-class residents most.
  • Chamath observes that Google's 32-year average return on invested capital (ROIC) of 32% gives management credibility to deploy $195-205B in CapEx without market panic, unlike startups that would face skepticism for the same capital intensity.
  • The hosts contend that increasing housing supply through permitting reform, not rent control or eviction restrictions, is the proven mechanism to reduce housing costs, pointing to Austin, Texas and Tokyo as examples where deregulation reduced prices.

Topics

Chinese AI models and U.S. regulatory responseOpen-source vs. proprietary AI market dynamicsAI model distillation and IP lawAnthropic's regulatory capture strategyCopyright settlements and fair use doctrineTech company capital expenditure and infrastructure investmentGoogle's competitive positioning in AI eraNYC housing policy and rent controlPrivate property rights and economic incentivesMarket commoditization of AI models

Transcript

All right, everybody, welcome back. Episode 282 of the world's greatest podcast. It's your podcasters favorite podcast. It's your mom's favorite podcast. It's the all in podcast. With me again, David Sachs, my pilot here, David Freberg. It was a big week was a big week. The continuing number one story in the world is Kimmy K three, it sparked a debate about banning Chinese open source models here in the world is Kimmy K3. It sparked a debate about banning Chinese open source models here in the United States and it's gone all the way to the White House last Friday. We talked about it here China's moonshot AI released Kimmy K3 open source model obviously performance on par…

Full transcript available for MurmurCast members

Sign Up to Access
View original source →

More from All-In with Chamath, Jason, Sacks & Friedberg

Get AI summaries like this delivered to your inbox daily

Get AI summaries delivered to your inbox

MurmurCast summarizes your YouTube channels, podcasts, and newsletters into one daily email digest.