- BlackRock says AI and blockchains both create machine-readable tokens, linking AI growth with broader digital asset use.
- Stablecoin market capitalization surpassed $300B, while adjusted transaction volume exceeded $11T in 2025.
- AI agents could use protocols such as MCP, A2A and x402 to coordinate tasks, access, compute and settle automated payments.
According to BlackRock, large language models and blockchains use different forms of tokenization. LLMs turn language into numerical tokens, while blockchains represent value and economic claims as digital tokens. However, both create machine-readable formats.
That link becomes more relevant as agentic AI handles multistep tasks. MCP connects AI applications with external data, while Google’s A2A supports communication between agents.
Payment Rails Adapt To Agentic Commerce
BlackRock identifies x402, developed by Coinbase, as an emerging protocol for machine-initiated payments. It uses the HTTP 402 status and can settle transactions with digital currencies, including stablecoins. Notably, x402 supports frequent, low-value payments without human intervention.
Stripe and Tempo developed the Machine Payments Protocol, while Stripe and OpenAI developed the Agentic Commerce Protocol. Google’s Agents Payment Protocol and Visa’s Trusted Agents Protocol focus on authorization.
Stablecoins also feature prominently in the research. Their circulating market capitalization exceeded $300 billion in September 2026. Adjusted stablecoin transaction volume topped $11 trillion in 2025, while growth from 2020 to 2025 reached 80% annually.
Compute Becomes Another Digital Asset Market
The paper also examines compute as an economic resource for AI systems. Estimates for AWS, Microsoft Intelligent Cloud and Google Cloud indicate combined revenue could reach about $1.1 trillion by 2030.
BlackRock notes that compute markets could develop standardized claims linked to capacity. Such contracts could represent compute rights, support collateral and enable programmable settlement.
Meanwhile, AI agents could compare compute prices, performance, latency and location before selecting resources. MCP and A2A could handle coordination, while x402 could settle payments per use.
The research also cites Stripe’s August 2026 agreement to acquire OpenRouter. OpenRouter routes workloads across more than 400 models from over 80 providers based on cost and performance.
