Brian Armstrong on Bitcoin, Anthropic releases Fable 5 and Mythos 5, and NewLimit’s $435M age-reversal raise
A wide-ranging discussion connecting Bitcoin, AI agents, space-based compute and NewLimit’s epigenetic reprogramming work.
This episode connects three infrastructure races: money for AI agents, compute for frontier models, and biology for healthier longevity. Brian Armstrong frames Bitcoin as a long-term digital-gold asset, but the more immediate shift is crypto as payment infrastructure for autonomous agents through wallets, stablecoins, APIs and on-chain reputation.
Key takeaways
- Coinbase wants to become the financial account layer for AI agents, enabling payments, portfolio actions and on-chain services that traditional banking rails cannot easily support.
- Quantum risk is not treated as imminent, but Armstrong argues the Bitcoin ecosystem will eventually need a post-quantum migration plan, including a difficult decision about old unmigrated coins.
- The AI race is increasingly a compute and infrastructure race: IPO timing, GPU contracts, hyperscaler capacity, SpaceX AI satellites and vertically integrated manufacturing all appear in the discussion.
- NewLimit is applying AI-driven screening to epigenetic reprogramming, with the goal of making cells functionally younger without changing their cell type.
Why it matters
The recurring theme is industrialization. AI agents need payments, capital formation and trust. Frontier models need energy, chips, factories and perhaps orbital data centers. Longevity companies need data, assays, biomarkers and expensive clinical trials. The winners may be those that turn scientific or software breakthroughs into scalable operating systems.
Signals to watch
The episode highlights several tensions: Bitcoin as digital gold versus risk asset, open-source AI as a much cheaper alternative for many workloads, public ownership stakes in AI labs as a political idea, and Apple’s reported reliance on Gemini to rebuild Siri. Each points to a market where infrastructure and governance may matter as much as raw model capability.
Source
- Chaîne: Peter H. Diamandis
- Vidéo source: https://www.youtube.com/watch?v=isd2y37j8v4