The economics of AI are shifting dramatically.​ Inference costs now exceed training expenses by 15-20x, fundamentally changing infrastructure priorities.​

Key metrics:

• Train a model for $1M → expect $15-20M in inference costs over its lifetime
• By 2030, inference will consume 75% of all AI compute spending
• Infrastructure optimized for sustained workloads now matters more than one-time training runs

This shift has major implications for AI companies and their cost structures, particularly as inference becomes the dominant operational expense.​

ENSIP-25 introduces a standardized method to verify that AI agents registered onchain are genuinely associated with their claimed ENS names.​

Key features:

• Defines a standardized ENS text record linking registry entries (like ERC-8004) to specific ENS names
• Enables wallets to display verified ENS names for AI agents instead of raw addresses
• Allows independent confirmation that agents like researchbot.​eth are authorized by their ENS name

Why it matters: As AI agents increasingly register identities, participate in governance, execute strategies, and sign transactions autonomously, verifiable identity becomes essential infrastructure.​ ENS extends its identity layer beyond humans, DAOs, and organizations to include software agents.​

Read the full technical details

New Liquidity Opportunity on Monad

The cbBTC/USDC trading pair is now receiving incentives on Monad's Liquidity Book DLMM (Discrete Liquidity Market Maker).​

What This Means:

• Liquidity providers can earn both trading fees and additional rewards
• This follows the recent AUSD/USDT0 stablecoin pair incentivization from February 25th
• The pool uses Liquidity Book's DLMM technology for efficient capital deployment

For Liquidity Providers:

Users can add liquidity to the cbBTC/USDC pool to start earning immediately.​ The incentive program aims to deepen liquidity for this Bitcoin-stablecoin pair on the Monad network.​