Interview, Fireside Chat
The Fight Over Open Source AI, Anthropic's $1.5B Payout, NYC Socialists: Evictions = Violence?
- US government intervention to ban open-source models is viewed as a probable future event, with Polymarket traders pricing a 45% probability of such a ban occurring in 2026 following the establishment of public record facts.
- If a ban is enacted, forecasts predict US AI costs will become 50 to 100 times higher than competitors via a "token tax," the US will lose the AI race to China, and the stock market will immediately tank.
- Anthropic and OpenAI valuations are expected to crater under a US ban due to reliance on regulatory capture, face massive margin compression, and risk being derailed by non-zero chances of IPO failure if open-source proliferation continues.
- Conversely, in the absence of a ban, Anthropic is forecast to reach over $70 billion in ARR by mid-year, potentially hitting $100 billion, while OpenAI is predicted to raise year-end ARR forecasts from $60 billion to $75 billion or higher.
- Market dynamics are expected to shift toward foundational models becoming commodities within five to ten years, compressing transition cycles, while value capture will increasingly concentrate in the cloud and chip infrastructure layers rather than model layers.
- Google is projected to be significantly rewarded for capital expenditures over the next 20 years, leveraging its ability to support fragmented models through silicon and cloud infrastructure, whereas Apple's new CEO is expected to shift capital allocation toward cloud services and acquisitions.
- In the New York housing market, strict regulations regarding rent control, eviction bans, and credit checks are predicted to cause 50,000 apartments to become "ghost apartments," reduce the housing supply, and force landlords to exit the market or charge multi-month prepayments.
- Economic consequences of New York's housing policies are forecasted to include increased rents, dilapidated housing stock, and a systemic erosion of private property rights, with the author comparing the trajectory to failed experiments in other jurisdictions and anticipating a "great American politburo."
- Legal precedents regarding IP theft and AI training data are expected to result in settlements where AI companies pay royalties similar to the music industry, though the "IP theft" argument is predicted to potentially backfire on companies like Anthropic.
- Future publishing schedules for 2027 are expected to feature new titles from podcast hosts and high-profile investors, while content creator lobbies may assert claims to 100% of revenue if IP theft arguments are accepted by courts.