Phil Rosen
July 28, 2026
AI Abundance, Bitcoin Scarcity, and the Future of Intelligence Markets
TL;DR
American open-source AI models will outcompete Chinese models due to superior innovation, while Bitcoin serves as a scarcity hedge against AI-driven deflationary abundance, forming a unified investment supercycle.
Key Takeaways
- 1The 'fat protocol thesis' and 'fat model thesis' are both incorrect; value accrues across fragmented stack layers including consumer software, specialized workflows, and models from multiple providers
- 2Chinese AI companies are 'drafting' off American foundational models through inference, similar to how pacemakers help world-record runners by breaking wind; this has triggered desperation from U.S. labs seeking government bans
- 3American open-source models will win because the U.S. created foundation models, Jensen Huang's Nvidia is backing open source, and Americans drastically underestimate their own innovation capacity
- 4Bitcoin and AI form a paired trade: AI creates abundance of intelligence (deflationary), while Bitcoin provides scarcity of value; the Fed has expanded its balance sheet by $200 billion since December despite 3.5% inflation because they anticipate deflationary AI forces
- 5Model routers are becoming essential because users care about total task cost (tokens × efficiency), not token price alone, driving adoption of multiple heterogeneous models in production systems
- 6Silvia's bet on specialized applied AI for personal finance outperforms the fat model thesis; heavy users grew net worth 16% in 6 months regardless of income level by using personalized portfolio insights instead of generic advice
- 7The company's stock down 80-90% from December IPO high creates asymmetric upside for believers; the small team of 'misfits' who understand AI's financial applications can create enormous value in an overlooked category
Notable Quotes
“The AI people should just be quiet and go listen to the crypto people. The fat protocol thesis is wrong and the fat AI model thesis is wrong as well.”
— Anthony Pompliano
“The Chinese models are doing to the American companies... they say, 'That's so cute that you're spending so much time and money to go train these foundational models. Wouldn't it be bad if we simply drafted off of you and we were able to then go create these open weight open source models that are just as competitive and lower cost?'”
— Anthony Pompliano
“In a world of abundance, what becomes valuable is scarcity. When you get abundance of intelligence, you then want scarcity of value. And I think that's where Bitcoin becomes really interesting.”
— Anthony Pompliano
“I love every second of it, right? If you go and you look, everyone's sitting there and they're laughing at you and they're like, 'You're an idiot.' I love those odds. Because guess what that means? It means there's tons of asymmetry.”
— Anthony Pompliano
Chapters
1. The Fat Protocol Thesis Revisited in AI
Pompliano explains that the 'fat protocol thesis' from crypto proved incorrect because value accrued across the entire stack rather than concentrating in the base protocol; the same fragmentation is now happening in AI with competition between open and closed source, American and Chinese models.
2. Chinese AI Drafting and American Desperation
Chinese companies are achieving competitive open-weight models at lower cost by 'drafting' off American foundational models through inference, similar to pacemakers helping Josh Kerr break the 1-mile world record; American labs are requesting government bans rather than competing on market performance.
3. Why American Open Source Will Win
Despite Chinese cost advantages, American open-source models will prevail because the U.S. created foundation models, Jensen Huang is backing open source, and Americans have underestimated their innovation capacity; fragmentation means users will route queries to specialized models rather than relying on single providers.
4. Model Routers and Total Task Cost Economics
Users increasingly care about total cost to complete a task, not token price alone; an expensive token-efficient model may cost less overall than a cheap but inefficient model, driving adoption of model routers that direct queries to the optimal provider for each job.
5. Bitcoin as AI Scarcity Hedge
The paired trade is abundance of intelligence (AI-driven deflation) versus scarcity of value (Bitcoin); since 2019 inflation has averaged 4% annually, the Fed expanded its balance sheet by $200 billion in recent months because they understand deflationary forces from AI, robotics, and tariffs will require dollar devaluation.
6. Silvia: Specialized AI Beats Fat Models
Silvia applies proprietary technology (file systems, memory, multi-agent orchestration, model routing) to personalized finance rather than generic models; heavy users grew net worth 16% in 6 months by receiving portfolio-specific tax and investment advice instead of generic financial guidance.
7. The Impossible Comeback and Asymmetric Odds
The stock fell 80-90% within 3 months of December IPO after Bitcoin holdings lost 50% in value; Pompliano frames this as an 'impossible comeback' with asymmetric upside because few outsiders believe success is possible, while a small team of 'misfits' holds conviction.
Key People & Entities
- Anthony Pompliano
- Crypto investor and CEO of Silvia, a personalized AI financial advisory platform; formerly involved in crypto since before Bitcoin's mainstream adoption
- Jensen Huang
- CEO of Nvidia; Pompliano credits him as the 'Godfather of AI' who is 'going all in on open source' and understands market dynamics
- Sam Altman
- CEO of OpenAI; credited with providing foundational AI model training service to the world
- Dario Amodei
- CEO of Anthropic; credited with providing foundational AI model training service to the world
- Josh Kerr
- Runner who broke the 1-mile world record with pacemakers 'drafting' for him; used as metaphor for Chinese AI companies leveraging American foundational work
- Jordi Visser
- Analyst who discusses Bitcoin as an AI play and supports thesis of pairing abundance of intelligence with scarcity of value
- Shane Nohr
- Lead engineer who helped Pompliano build Silvia product; described as the 'best engineer' across all their companies
- Silvia
- Personalized AI financial advisory product that combines proprietary technology (file systems, memory, multi-agent orchestration, model routing) with portfolio-specific context
Glossary
- Fat Protocol Thesis
- Early crypto theory that a single base protocol would accumulate most value by capturing the entire technology stack; disproven in practice as value distributed across layers including consumer software and middleware
- Drafting
- In AI context, the strategy of building competitive models by leveraging insights from others' expensive foundational model training through inference and model replication
- Model Router
- Software that directs queries or tasks to the most appropriate AI model based on cost-efficiency, task type, or performance characteristics rather than using a single provider
- Token Cost
- The price per unit of text (token) that an API charges; one component of total task cost but not the only factor determining actual expenses
- Open Weight
- AI models where the neural network weights are publicly released, allowing anyone to download, modify, and deploy the model without relying on a company's API
- Cultural Weight
- Ideological biases embedded in AI model training data and weights that reflect the values and assumptions of the training dataset creator
- Monte Carlo Simulation
- Statistical technique that runs thousands of probabilistic scenarios to estimate potential outcomes, used in Silvia to forecast portfolio growth over a time horizon like 10 years
- Specialized Applied AI
- Domain-specific AI systems optimized for particular workflows (like personal finance) with context about an individual user, as opposed to general-purpose models
- SOC 2 Compliant
- Security and data protection standard certifying that a company's systems meet industry requirements for safeguarding customer information