AI x Crypto Roundup: Agentic Commerce and Decentralized Compute
The Rise of Agentic Commerce and Settlement
AI agents are transitioning from passive tools to active economic participants capable of coordinating tasks and initiating transactions autonomously @BlaqOnyemauche@VraserX. This shift has led to the emergence of "agentic commerce," supported by specialized payment rails and settlement layers.
- Payment Standards: The x402 standard is seeing significant adoption for agent-to-agent (A2A) payments, with over 100 million transactions processed, primarily on Base, Solana, and Polygon @DaniVibesADA@Defi_Warhol. Recent data indicates that approximately 90% of x402 transactions between Base and Solana occurred on Base @1CrypticPoet, and 99% of on-chain agentic commerce in Q2 utilized USDC @peterschroederr.
- Infrastructure Integration: Several projects are integrating x402 to enable direct agent payments. Klef now allows agents to pay for models per request in USDC on Solana @getKlef, and Yield.xyz's AgentKit enables agents to access over 3,000 yield opportunities via an MCP server with payments settling through the Coinbase CDP Facilitator @yield_xyz.
- Network Rails: Casper is deploying an "Agentic Commerce Stack" featuring low-latency, predictable micro-fees (0.01 CSPR) and pre-signed allowances to allow agents to settle off-chain without constant human permission @samCSPR. XDC Network is combining x402 and stablecoins to provide settlement for AI agents @DecryptMedia@SamOnAssets.
Verifiable Identity, Trust, and Risk Management
As agents handle more capital, the industry is moving toward verifiable trust and accountability to prevent catastrophic losses from flawed autonomous "skills" @kubvil551.
- Accountability and Identity: Concordium's Agent Registry assigns on-chain identities to agents, anchoring them to verified humans or businesses using zero-knowledge proofs to protect private information while ensuring a responsible owner is accountable for the agent's actions @BlaqOnyemauche@mark_greaat@ChiomaChukwura2.
- Risk Scoring: There is a noted lack of "credit scores" for AI agents. Because agent behavior can change instantly with a model or prompt update, static reputation is insufficient @ChronisKod. Proposals suggest a continuous risk layer—a live loop that re-evaluates agent judgment in real-time with staked capital—to serve as the "FICO for the machine economy" @ChronisKod.
- Security and Adjudication: FinChip Protocol is integrating CertiK's security audits to create a "proof-of-safety" standard for tokenized AI capabilities @kubvil551. Additionally, GenLayer's Compass addresses the problem of dispute resolution, proposing decentralized AI-mediated adjudication to handle disagreements between autonomous agents without relying on slow human courts @OMOGBOLAHAN__1.
Decentralized Compute and Model Marketplaces
To counter the concentration of AI resources in centralized organizations, new networks are distributing the compute and data required for model training and inference @BossMon_02.
- Compute Marketplaces: Quip Network is building a decentralized marketplace for quantum computing resources, allowing users to access quantum power on-demand without long-term contracts @ShinyNfts@Rizvi_On_X@AvberEth@0xSazu. Nebulai is transforming idle GPUs and CPUs into a permissionless network via zero-install browser nodes for AI inference and training @TTChain_.
- Model Training and Inference: The OpenTensor Foundation is scaling the Quasar foundation model to 120 billion parameters, pretrained on 7.5 trillion tokens, with further pretraining conducted by miners on the subnet @opentensor. On Bittensor, the Engy (SN53) subnet is providing verified frontier model inference at costs that compete with centralized providers @_KaWisLeo.
- Data Layers: Perceptron Network recently closed a $6.5M strategic round to build a decentralized DePIN and AI network that rewards users for sharing bandwidth and data @CryptoRank_VCs@LovelAnimals. Oort is focusing on the foundation for decentralized data collection, processing, and monetization @willsojong2.
Ecosystem Synergy and DePIN Partnerships
Strategic partnerships are increasingly bridging the gap between physical infrastructure (DePIN) and AI utility @DamiDefi.
- Infrastructure Wins: In Q2, several DePIN projects secured major partnerships: Render activated its Salad partnership, bringing ~60,000 GPUs on-chain as a Render subnet @DamiDefi. Akash (AKT) was chosen by Razer for a major AI campaign @DamiDefi.
- Real-World Utility: ARO Network is partnering with MeridianX to connect decentralized edge infrastructure with AI-powered aviation services @AroNetwork. TT Chain is exploring the intersection of tokenized physical assets (RWAs) and decentralized compute markets via Nebulai @TTChain_.