McKinsey: $1T will flow through AI agents

McKinsey’s agentic-commerce estimate places up to $1 trillion in orchestrated U.S. retail revenue through AI agents by 2030, and a global range of $3 trillion–$5 trillion by 2030 in its agentic-commerce analysis. (digitalcommerce360.com) McKinsey’s June 2025 brief “Seizing the Agentic AI Advantage” frames the shift as requiring an “agentic AI mesh” and says the majority of firms have adopted generative AI but remain immature—the report cites widespread adoption with low measurable impact and notes many high‑value vertical use cases are still stuck in pilot stages. (searchyour.ai) Open standards and vendor protocols are already emerging to make merchants agent‑reachable: OpenAI and Stripe published the Agentic Commerce Protocol (ACP) to standardize agent→merchant discovery and checkout (the ACP powers Instant Checkout in ChatGPT). (developers.openai.com) Anthropic’s Model Context Protocol (MCP), released in November 2024, aims to standardize how models access external data and tools so agents can read and act on catalog, policy and inventory state, while Google Cloud’s Agent Payments Protocol (AP2) (announced Sept. 16, 2025) and related industry repos define auditable agent payment flows. (anthropic.com) (cloud.google.com) McKinsey and follow‑on industry analyses warn businesses lacking machine‑readable product schemas, pricing rules and API endpoints will be “invisible” to agentic flows unless they expose structured data and transaction hooks for discovery, negotiation and fulfillment. (gend.co) (calleosolutions.com) Protocol specs and payment‑rail initiatives explicitly require merchant‑side order acceptance semantics, shared tokens, and transaction audit trails—AP2 and ACP docs call out signed handshakes, merchant acceptance/decline signals, and webhook/receipt traceability for agent transactions. (ap2-protocol.org) (agentic-commerce-protocol/README.md) McKinsey’s playbook emphasizes four pillars—people, governance, technology and data—and its analysis quantifies the adoption gap and the need to redesign workflows; that gap is driving platform efforts to ship SDKs, example servers, and reference implementations (ACP/MCP SDKs and GitHub samples) to move merchants from pilot to production. (searchyour.ai) Payments incumbents and fintechs are concurrently racing to operationalize agent rails—Stripe, Visa, Mastercard and newer entrants announced agent‑payments products and integrations in 2025–2026, signaling commercial paths for merchants who implement ACP/AP2 endpoints and machine‑readable catalogs. (forbes.com)