David Krais — AI Lead at the Ethereum Foundation (4 trade ideas)

← All Speakers
Date Ticker Direction Thesis Source
Feb 08, 2026
ENS
LONG In the "Identity Registry" (the agent's passport), David notes: "There can also be like ENS name attached to it." Just as humans need readable names (DNS/ENS) to transact, agents will need readable identities to build brand and reputation. If ERC-8004 becomes the standard, ENS becomes the de facto naming convention for millions of AI agents, not just humans. LONG. A second-order derivative of the "Agent Identity" thesis. Agents are machines; they can read hex addresses easily. The need for human-readable names might be lower for agent-to-agent commerce than human-to-human commerce. Unchained (Chopping Block)
Want to Hire an AI Agent? Check Their Reputat...
Feb 08, 2026 LONG David explicitly highlights "Virtuals Protocol" and "Olas" as the two teams that were "seeing that the future of AI is multi-agent" and creating agent registries before ERC-8004 launched. ERC-8004 is an interoperability standard, not a competitor. Its launch validates the thesis of these specific protocols. As the "Agentic Internet" grows, the first-mover protocols that organize and coordinate these agents (Virtuals/Olas) will capture the initial liquidity and developer mindshare. LONG. These are the pure-play assets for the "Agent Economy" narrative. The standard (ERC-8004) is open-source; value might accrue to the base layer (ETH) rather than the application layer protocols if they don't maintain a competitive moat. Unchained (Chopping Block)
Want to Hire an AI Agent? Check Their Reputat...
Feb 08, 2026
ARB /OP /ETH
LONG "We are launching like two or three chains a week... on all Ethereum layer 2s." He also states, "The demand for blockchains and trust is essentially only going to grow" as agents conduct commerce. AI agents perform high-frequency, low-value transactions (micro-services). This volume is impossible on L1 due to gas costs, making L2s (Arbitrum, Optimism, Base) the primary execution environment. However, the reputation and final settlement anchor to Ethereum. This creates a flywheel: L2s get the transaction fees/volume, ETH gets the collateral/security demand. LONG. Group trade: ETH as the store of value/trust, L2s (ARB/OP) as the transaction rails. If agents move to high-throughput monolithic chains (like Solana) due to latency requirements, Ethereum L2s could lose the "Agent Economy" war. Unchained (Chopping Block)
Want to Hire an AI Agent? Check Their Reputat...
Feb 08, 2026
GRT
LONG When discussing how to verify agent quality and prevent fraud (Sybil attacks), David mentions "people are using subgraphs" and "a project in the Graph ecosystem" to build "watchtowers" that measure agent latency and quality. AI Agents generate too much data for humans to audit manually. "Watchtowers" (automated auditors) are required infrastructure. These watchtowers rely on indexing protocols to query on-chain data efficiently. The Graph (GRT) is the fundamental indexing layer for this data. More agents = more data queries = higher demand for GRT. LONG. This is a "pick and shovel" play on AI data infrastructure. Competitors in the data availability or indexing space could erode market share; agents might optimize for off-chain data storage to save costs. Unchained (Chopping Block)
Want to Hire an AI Agent? Check Their Reputat...