AI Agents Will Need Crypto to Work (Here’s Why) w/ Tom Schmidt
Speakers
Summary
- AI-Crypto Convergence: The real intersection isn't decentralized training (too slow/expensive), but "Agentic Payments." AI agents cannot open bank accounts; they will use stablecoins for micro-transactions and API calls.
- Edge Compute Thesis: Open-source models are becoming efficient enough to run locally on consumer hardware (e.g., Apple Silicon), reducing reliance on centralized GPU clusters.
- Institutional Pivot: Major players like BlackRock, Fidelity, and Morgan Stanley have moved from "exploring" to pivoting entire roadmaps into crypto, signaling the end of the "tourist" phase.
- Prediction Markets: The sector is expanding beyond elections into sports and pop culture, validating "event contracts" as a superior financial primitive for information discovery.
- Regulatory Optimism: The current administration and potential CFTC leadership changes suggest a move toward regulated, onshore crypto derivatives (perpetuals) in the US.
Trade Ideas
Schmidt highlights a portfolio company (Exo) that enables distributed inference by sharding models across local devices. He explicitly mentions using "Apple Mac Studios, Mac Minis, and iPhones" because new open-source models can run locally on high-end consumer chips. As AI models become efficient enough to run on "edge" devices (to avoid censorship or cloud costs), demand for high-performance consumer hardware with strong neural engines (like Apple Silicon) increases. Apple becomes the hardware layer for decentralized AI. LONG. Apple is the "pick and shovel" play for local/edge AI inference. Open-source models failing to catch up to closed-source giant models (OpenAI/Google), rendering local compute irrelevant.
Schmidt discusses the explosion of prediction markets (like Poly Market), noting that volume is sticky even post-election because users are trading sports and pop culture events. While Poly Market is offshore/crypto-native, the demand for "event contracts" is proving to be massive. Robinhood (HOOD) recently launched prediction markets for US customers, and DraftKings (DKNG) operates in the adjacent sports betting vertical. These regulated US entities will capture the onshore demand for this market structure. LONG. Betting on the "gamification" of financial events and the legalization of prediction markets in the US. The CFTC could ban event contracts in the US, forcing this volume back to offshore/crypto-native platforms.
Schmidt argues that AI agents (autonomous software) need to make payments but cannot get credit cards or bank accounts. He states stablecoins are the "perfect fit" for these micro-transactions and cross-border API calls. If AI agents proliferate, transaction volume on stablecoin rails will explode. Coinbase (via USDC partnership), PayPal (PYUSD), and Visa (integrating stablecoin settlement) are the primary regulated infrastructure providers that will capture fees from this new "machine economy" volume. LONG. These companies own the "rails" for the next generation of automated B2B/B2C payments. Regulatory crackdowns on stablecoin issuance or the development of a Central Bank Digital Currency (CBDC) that bypasses private issuers.
Schmidt notes that the industry is "growing up," citing that "BlackRock is here, Fidelity is here, and Morgan Stanley just came out pivoting their whole roadmap into crypto." This is no longer a retail speculation game; it is an institutional asset class. BlackRock (ETFs/Tokenization) and Morgan Stanley (Wealth Management/Custody) are positioning themselves to earn fees on the securitization and custody of digital assets for the wealthy. LONG. These incumbents will capture the "safe" yield and management fees as crypto becomes a standard portfolio allocation. continued regulatory hostility or a catastrophic failure of a major custodian that scares institutions away.
Up Next
This Milk Road Daily video, published March 04, 2026, features Tom Schmidt discussing AAPL, HOOD, DKNG, COIN, V, PYPL, MS, BLK. 4 trade ideas extracted by AI with direction and confidence scoring.
Speakers: Tom Schmidt · Tickers: AAPL, HOOD, DKNG, COIN, V, PYPL, MS, BLK