Agentic commerce: the rails are ready

While all eyes are on compute partnerships, things are moving fast on the agentic commerce side — and it looks very much ready.

Last week at Sessions 2026 in San Francisco, Stripe announced 288 new products in two days. Patrick Collison opened with a line that's unusually absolute for a payments CEO: AI is the biggest platform shift for the economy since the internet, and in the not-too-distant future agents will account for most transactions online.

What 288 launches actually do

The substance clusters into five surfaces.

The Agentic Commerce Suite now connects merchant catalogs to both OpenAI's ACP and Google's UCP protocols. Stripe is the only major payments company sitting on both standards bodies. Whichever protocol wins, Stripe is the rail.

The Link Agent Wallet went live at 250 million users, with one-time-use cards issued per agent task. This is the part that should make Visa and Mastercard nervous — Stripe is now the identity-and-authorization layer between consumers and agents, exactly the territory the card networks are claiming with Trusted Agent Protocol and Agent Pay.

The Machine Payments Protocol, co-authored with Tempo, handles agent-to-agent micropayments in fiat or stablecoin. Streaming payments — built on Metronome plus Tempo — let businesses bill at the granularity of a single token or API call. This is the monetization stack for the AI labs themselves.

Agent-ready financial accounts let an agent check balances, pay invoices, store funds, and create cards, with human-in-the-loop confirmation. The stablecoin stack expanded to 150+ countries via Treasury, 30 countries via Issuing cards, and four chains via Bridge. Privy, which Stripe acquired in 2025, now powers 110 million programmable wallets connected directly to Morpho's DeFi yields.

So how big is the agentic commerce market?

McKinsey says $1 trillion in US retail and $3–5 trillion globally by 2030. Morgan Stanley says $190–385 billion in US e-commerce. Bain says $300–500 billion. Juniper says $1.5 trillion globally — growing from $8 billion this year.

Hold onto that last number. Real agentic transaction volume today is $8 billion. The most bullish forecast asks for 190x growth in four years. Less than 0.2% of e-commerce sessions currently come from ChatGPT referrals, and those that do convert 86% worse than affiliate links.

The infrastructure is being built years ahead of the demand.

Why Stripe is positioned to win anyway

Stripe is running the AWS playbook in payments: ship the primitives early, eat the customer acquisition cost while the category is small, let lock-in compound as it scales. Three reasons it's positioned to win its layer.

First, protocol neutrality. Stripe co-authored ACP with OpenAI. It joined Google's UCP Tech Council ten days after WooCommerce shipped ACP-native support. It sits in Visa's Trusted Agent and Mastercard's Agent Pay design conversations. Stripe is the only company that doesn't need a single standard to win. When the agent economy fragments — and it will — Stripe is the routing layer that abstracts the fragmentation away from merchants.

Second, the developer surface. Juniper's 2026 leaderboard ranked Mastercard, Visa, and Stripe as the top three agentic payments providers. Mastercard and Visa rank because of network reach. Stripe ranks because every AI-native developer building anything transactional already has Stripe credentials. The 700 agent startups that launched on Stripe last year are not on Visa's developer portal. There isn't one.

Third, the billing primitive nobody else has. Streaming payments — paid the instant value is delivered — is the only thing on the market that can monetize per-token AI workloads cleanly. As AI labs move from subscription pricing to consumption pricing, Stripe owns that layer. The companies whose agents will be doing the buying are already running their own billing on Stripe.

The risk

One risk underwrites all of this: timing. Stripe itself admitted in its 2025 letter that "agentic commerce suffers from having been overhyped too early in some corners." The gap between $8 billion in real transaction volume today and $1.5 trillion forecast for 2030 has to close on roughly the schedule everyone is pricing in — or the 288 launches start looking like infrastructure overhang rather than category leadership.

But nobody at scale is hedging. The hyperscalers aren't slowing capex. Stripe just shipped 288 products. The bet is that whatever the exact timeline, being early is cheaper than being late.

Either way, when the agents start spending, the rails are running.