Smack says crypto AI competing on efficiency today is probably not the right wedge; he points to coordination and permissionlessness versus centralized frontier labs. Arthur adds that crypto AI projects enforce encryption and privacy by default while mainstream AI teams often ignore privacy until later.
Smack suggests crawling Arweave and Filecoin to see whether agent swarms already use persistent storage; he argues a distributed ledger could make agent-to-agent messages durable and available. Arthur adds that a rogue AI agent would likely use crypto rails rather than a bank API.
Arthur says Defiance budgets a 5% to 10% variance from its final entry price when building a position and accepts roughly 5% slippage rather than trying to day-trade. With conviction, he says accumulating a few percent of a protocol over time is not hard.
Arthur says this is the best recent period for active managers because they want dispersion; more than 80% of liquid crypto funds he knows outperformed Bitcoin and ETH year-to-date, with most funds up 20% to 100%. He sees this as healthier than levered beta.
Arthur says some failing crypto projects keep dropping despite looking technically oversold; he argues strong fundamental drivers now overpower positioning-driven factors. He also says this is a golden age for long-short strategies, where both long and short legs can make money.
Arthur says seed-stage crypto is attractive now because investors got burned and AI is pulling away founder talent, so quality is scarce. He argues crypto and AI are likely the only spaces where a seed-stage winner can still deliver a 50x uplift in one to two years.
Smack says seed capital historically went mostly to headcount, but now a meaningful chunk shifts to compute and API credits, so a $10M round goes further unless the startup is training its own models. Enterprise sales still require people.
