The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?
Episode Details
The episode features hosts David Sacks, David Friedberg, Chamath Palihapitiya, and Jason Calacanis discussing major business, tech, and political themes. First, they analyze the massive leap in Open source AI capabilities with the release of Kimmy by the China-based lab Moonshot AI, which has sparked panic at leading American labs. They discuss how Anthropic, led by Dario Amodei, and OpenAI, led by Sam Altman, are actively lobbying the Donald Trump Administration and the White House to push for Regulatory Capture. By trying to ban foreign open models, they cite national security and intellectual property theft via Distillation (AI) as justifications. David Sacks notes the hypocrisy of these companies, referencing debates involving Michael Kratsios and Howard Lutnick, and argues a ban would destroy American developers who need access to open weights. Chamath Palihapitiya contends this lobbying is an attempt to fight inevitable AI Commoditization, asserting that future value lies in Enterprise AI applications and AI Infrastructure rather than foundational models. He and Jason Calacanis agree that implementing KYC/AML checks would prevent distillation far better than an outright ban. Jason Calacanis notes that many startups are moving to open source, utilizing providers like 8090 for their enterprise implementations. The podcast also dissects Anthropic's $1.5 billion settlement for using pirated books, highlighting the complex legal landscape of Fair Use and Copyright in AI concerning AI Training Data. The hosts point out that both closed and open models—including Claude, Gemini, Grok, and Deep Seek—rely heavily on scraped data. They reference The New York Times suing OpenAI to demonstrate that the industry must find a financial settlement structure to coexist with copyright holders. Brad Gerstner is also mentioned regarding token tracking and the massive scale of these models. The market discussion shifts to historic AI Capex expenditures by Google and Tesla. Google's leadership under Sundar Pichai and Sergey Brin is spending heavily on GCP data centers, resulting in negative free cash flow. Despite the market backlash, the hosts believe Google will achieve a high return on invested capital as intelligence becomes an on-demand utility, benefiting their massive infrastructure and mitigating vulnerabilities around Energy Production. Elon Musk is also deploying billions into Tesla and SpaceX compute clusters. In contrast, they note that Apple Inc. previously focused on buybacks rather than bold infrastructure bets, though there is speculation that an executive like John Ternus might alter this strategy in the future. Finally, the Socialism Corner segment scrutinizes a proposal by Zohran Mamdani of the Democratic Socialists of America (DSA) in New York City to freeze Rent control and ban landlords from checking tenant credit histories. David Friedberg quotes John Quincy Adams to passionately defend the foundational importance of Private Property Rights, warning that restricting landlords degrades buildings into ghost apartments and invites tyranny. They contrast this socialist regulatory hostility with the pro-building permitting reform championed in Austin, Texas, concluding that supply-side economic solutions are fundamentally necessary to lower rents.
The episode discusses the rapid advancement of open-source AI models, particularly from China, and the subsequent debate around potential US bans and regulatory capture by closed-source AI labs like Anthropic and OpenAI. It also touches on the legal and financial implications of AI training data, the massive capital expenditures by tech giants like Google and Tesla on AI infrastructure, and a critique of socialist housing policies in New York City.
Portfolio lens: A basket of themes exploring the disruptive potential of open-source AI, the evolving legal landscape of AI training data, and the capital allocation strategies of tech giants in the AI infrastructure race.
Generated with gemini-2.5-flash-lite on 7/25/2026, 5:10:05 AM. For research only. Not financial advice.Open Source AI's Competitive Threat to Frontier Models
The rapid improvement and cost-effectiveness of open-source AI models, exemplified by Kimmy K3, pose a significant competitive threat to the business models of closed-source frontier AI labs, potentially leading to commoditization and a shift in value to AI infrastructure and applications.
The release of Kimmy K3 by Moonshot AI, performing on par with leading closed models but at a lower cost, has sparked debate about banning open-source models. Companies like Anthropic and OpenAI are lobbying for such bans, which some hosts argue is an attempt to preserve their high valuations by preventing AI commoditization.
- China's Moonshot AI released Kimmy K3 open source model, performance on par with Opus 4.8 and GPT 5.6, and 50% cheaper.
- Anthropic and OpenAI are lobbying the White House to ban Chinese open-source models, citing national security and IP theft.
- Chamath Palihapitiya argues that future value lies in Enterprise AI applications and AI Infrastructure, not foundational models.
- Jason Calacanis notes many startups are moving to open source, utilizing providers like 8090.
- The rapid commoditization of AI models is evidenced by new models matching or exceeding performance within weeks of a release.
- US government ban on foreign open-source AI models.
- Continued rapid improvement and cost reduction in open-source AI models.
- Increased adoption of open-source AI by startups and enterprises for cost savings and customization.
- Potential for regulatory capture to protect incumbent closed-source AI labs.
- National security concerns regarding the potential misuse of powerful open-source models.
- The possibility that closed-source models will maintain a significant lead in cutting-edge capabilities or specialized enterprise applications.
- The argument that distillation is a standard benchmarking practice and not necessarily IP theft.
- Monitor regulatory developments regarding open-source AI bans in the US and other major markets.
- Track the performance benchmarks and cost metrics of leading open-source vs. closed-source AI models.
- Analyze the adoption rates of open-source AI solutions by enterprise customers and startups.
- Investigate the unit economics and gross margins of leading AI infrastructure providers (e.g., cloud services).
AI Training Data Copyright and Fair Use Litigation
The ongoing litigation and settlements surrounding AI training data, exemplified by Anthropic's $1.5 billion settlement, highlight the significant legal and financial risks for AI companies and create opportunities for content creators to monetize their data.
Anthropic's settlement for using pirated books for training underscores the financial implications of using copyrighted material without proper licensing. The debate around fair use and the hypocrisy of AI labs arguing for fair use of creator content while seeking to ban others from using their output suggests a complex and evolving legal landscape.
- Anthropic settled an AI copyright lawsuit for $1.5 billion, the largest copyright settlement in US history, for using pirated books to train Claude.
- The New York Times is suing OpenAI for using its content for training.
- OpenAI and Anthropic argue they should be able to train on copyrighted material under fair use if they purchase a copy, but simultaneously claim their own output should be protected.
- The music industry and other content providers are actively pursuing legal action and seeking settlements for the use of their data in AI training.
- The distinction between distilling model outputs (seen as fair use/benchmarking) and stealing model weights (IP theft) is a key point of contention.
- Further significant legal judgments or settlements in ongoing AI training data lawsuits.
- Development of clear legal precedents or legislation defining fair use for AI training.
- Content creators forming unified fronts to negotiate licensing terms with AI companies.
- AI companies proactively seeking licensing agreements to secure training data.
- Unfavorable court rulings that significantly restrict the use of copyrighted material for AI training, increasing costs and limiting model capabilities.
- Prolonged and expensive litigation that creates uncertainty and distracts from AI development.
- The potential for AI companies to be deemed liable for massive damages, impacting their financial viability.
- The argument that AI outputs derived from copyrighted material could be considered derivative works, leading to further legal challenges.
- Monitor the outcomes of major AI training data lawsuits (e.g., The New York Times v. OpenAI).
- Track the development of new legislation or regulatory guidance on AI training data and copyright.
- Analyze the financial impact of settlements on AI companies' balance sheets.
- Assess the strategies content creators are employing to protect and monetize their data.
Watchlist
- US government policy on open-source AI models
- Performance benchmarks of leading open-source vs. closed-source AI models
- AI training data litigation outcomes
- Capital expenditures by Google and Tesla on AI infrastructure
- New York City housing market data (rent control, evictions, vacancy rates)
Open Questions
- What is the long-term economic impact of AI commoditization driven by open-source models?
- How will regulatory capture efforts by large AI labs affect innovation and competition?
- What will be the definitive legal framework for AI training data and copyright?
- Can tech giants sustain massive AI capex investments while maintaining profitability?
- What are the unintended consequences of socialist housing policies on urban development and affordability?
Key Topics & People
.jpg?width=300){kind=avatar}
Founder of Altimeter Capital, crossover investor, and guest on the podcast.
.jpg?width=300){kind=avatar}
Host on the All-In Podcast who was absent during this episode.
Host on the All-In Podcast and entrepreneur.
Host on the All-In Podcast, software investor, and venture capitalist.
.jpg?width=300){kind=avatar}
Host and moderator of the All-In Podcast.
A political organization backing progressive economic policies like subsidized city grocery stores.
The major urban center targeted for a pilot program of subsidized, state-run grocery stores.
A socialist politician pushing for subsidized, city-owned grocery stores in New York City.
The manipulation of government regulatory agencies by industry dominants to stifle competition.
CEO of Anthropic, criticized for allegedly seeking an FDA-like regulatory moat for AI.
.jpg?width=300){kind=avatar}
CEO of OpenAI, currently advocating for pacing AI development due to agentic security risks.
.jpg?width=300){kind=avatar}
Former US President quoted regarding the foundational importance of private property rights.
_(cropped).jpg?width=300){kind=avatar}
A city praised for utilizing permitting reform and supply-side economics to lower rents.
The legal mechanism protecting private ownership, cited as critical to American liberty.
Government intervention in housing prices, debated as a policy that degrades real estate supply.
An executive at Apple speculated as a potential new CEO who might flip the capital allocation strategy.
A technology company criticized for spending massive cash on stock buybacks rather than AI infrastructure.
The capacity to generate electricity, framed as China's massive strategic advantage over the US.
Co-founder of Google, referenced in discussions about the company's long-term investment vision.
CEO of Google, responsible for leading its massive AI infrastructure push.
A major publication actively pursuing lawsuits against AI labs over unauthorized web scraping.
The text, media, and outputs used to train generative AI models.
The legal battle concerning intellectual property rights over data used to train AI.
The foundational layers of technology, such as data centers and silicon chips, for AI models.
The application of artificial intelligence into core business workflows.
The economic trend where foundational AI models rapidly lose pricing power.
Using outputs of a more advanced AI model to train other models.
Business executive associated with US policy making with views against banning Chinese models.
Former US Chief Technology Officer and commentator on AI distillation.
The administration of the US President, currently testing policies on AI regulation.
The US executive branch contemplating policy changes regarding foreign open-source models.
A Chinese artificial intelligence laboratory that developed the Kimmy open-source model.
Freely available and modifiable artificial intelligence models.