Digital generated image of robot’s hand holding multicoloured credit card against blue background.
getty
Google is building the infrastructure layer for AI-driven shopping, and major retailers including Walmart, Target, and Shopify are backing it. That’s because Google’s Universal Commerce Protocol (UCP) isn’t just another integration point. It’s not another API to connect. Google’s announcement of UCP is the beginning of new infrastructure.
Agentic commerce is a fundamental shift in e-commerce and Google is building the rails for AI-native commerce, positioning itself as the protocol layer between consumers, retailers, and payment systems. Meanwhile Amazon is conspicuously building walls and other AI giants are making their own attempts at agentic commerce features.
This is the end of search-based shopping as we know it. The infrastructure wars for agentic commerce are just beginning.
The Protocol Layer Play
UCP standardizes the entire transaction stack—discovery, authentication, payment, fulfillment—as a protocol layer instead of just another API endpoint. According to Google’s technical documentation, UCP establishes “a common language and functional primitives” that enable seamless commerce journeys between consumer surfaces, businesses, and payment providers.
Think about it like TCP/IP for commerce.
Nobody owns the internet protocol, but the companies that defined those standards shaped how the entire internet works. Google is making the same play here.
When ChatGPT, Claude, Gemini, or any AI agent needs to execute a purchase, they don’t build custom integrations with every retailer. They just speak UCP. The protocol defines how AI agents talk to payment providers, inventory systems, and fulfillment networks using a standard language.
Google open-sourced the protocol to accelerate adoption. UCP is built on industry standards—REST and JSON-RPC transports, with Agent Payments Protocol (AP2), Agent2Agent (A2A), and Model Context Protocol (MCP) support built in. Google’s documentation confirms that UCP “provides businesses flexible ways to integrate via APIs, Agent2Agent (A2A), and the Model Context Protocol (MCP),” giving retailers multiple integration paths depending on their technical infrastructure.
But here’s what matters: Google maintains the core specification. That means Google decides what gets added, what gets deprecated, how versioning works. Once retailers build their systems to speak UCP and AI agents train on UCP patterns, moving to a different protocol becomes expensive even though you could technically fork it.
It’s the same play AWS made with cloud infrastructure. So while the protocol is explicitly “built to work with existing retail infrastructure” and is “compatible with Agent Payments Protocol (AP2) to provide secure agentic payments support,” there’s more to it than that. As an infrastructure play, users can leave but why would they rebuild everything?
Why Major Retailers Are Backing Google
Google explicitly states that UCP is “co-developed with industry leaders” and positions it as “an evolving open-source standard designed to be community-driven.” That list includes the likes of Walmart, Target, Shopify, Etsy, Wayfair, and more than 20 other global corporations. The big news is that they didn’t just sign on as partners, they’re co-developing and endorsing UCP as a standard. One name notably absent from the list: Amazon.
Amazon captures roughly 40% of U.S. e-commerce spending and runs an advertising business generating around $70 billion annually—revenue that depends on humans browsing and clicking. In November, Amazon sued Perplexity to stop the startup from using its AI browser agent to make purchases on its marketplace.
While everyone else is forming partnerships and building integration protocols, Amazon is building walls.
This signals a strategic divide. Walmart’s incoming CEO John Furner stated that the transition from traditional web search to agent-led commerce represents “the next great evolution in retail” and that Walmart isn’t just watching the shift, they’re driving it.
Here’s why retailers are choosing Google’s infrastructure over proprietary systems: they retain full ownership of customer relationships, data, and the post-purchase experience. As Google’s documentation explicitly states, “you own your business logic, and you remain the Merchant of Record.” While retailers surrender some control at the protocol layer, they’re gaining a much larger upside of agent-mediated distribution without needing to build and maintain the infrastructure.
PayPal’s SVP and Head of AI Prakhar Mehrotra put it plainly: “Protocols like UCP turn agentic commerce into something merchants can actually adopt at scale.”
The Advantage Of Shopping Graph Data
Google’s Shopping Graph isn’t just a product database. It’s a structured knowledge graph of commerce relationships—and UCP is designed to integrate directly with it.
It maps products to attributes, attributes to user intent, intent to purchase patterns, retailers to inventory, prices to competitive positioning. Google’s Shopping Graph contains over 50 billion product listings, with more than 2 billion of those listings refreshed every single hour. According to Google’s technical blog, UCP allows agents to “discover your business capabilities, invoke checkout and even apply a discount at the checkout request,” while supporting “capabilities such as identity linking and order management.”
Most AI shopping implementations bolt LLMs onto traditional search indexes or product catalogs. They do retrieval-augmented generation where the LLM queries a database, gets results back, then tries to make sense of it.
That’s slow. And the LLM doesn’t actually understand commerce relationships—it’s just pattern matching on text.
Google connects Gemini directly to Shopping Graph’s entity relationships. That means Gemini can reason about complex queries by traversing the graph’s actual relationship nodes, not just keyword matching.
When you ask for “wireless headphones under $200 with good battery life that work with my existing devices,” Gemini understands that “good battery life” for headphones means 20+ hours, that “my existing devices” connects to your Google account’s device history, that price comparisons need to account for current promotions.
The moat is that Shopping Graph has years of commerce relationship data validated through actual transactions. You can’t replicate that by scraping product pages.
And because UCP feeds transaction data back into Shopping Graph, it gets smarter with every agent-mediated purchase. The protocol feeds the graph, the graph trains the model, the model executes better transactions, which generates more protocol data. Google has built “the first reference implementation of UCP, to power a new buying experience that allows consumers to purchase directly from eligible businesses across Google’s conversational experiences like AI Mode in Search and Gemini.”
It’s a flywheel.
How UCP Actually Works
Most coverage of UCP focuses on the partnership announcements. But the technical architecture reveals why this protocol matters.
UCP uses a modular, capability-based design. According to Google’s developer documentation, the protocol supports discovery and negotiation mechanisms that allow agents and merchants to communicate what they can handle before executing transactions.
Here’s how it works: Merchants declare what capabilities they support—checkout, identity linking, order management, discounts, loyalty programs. Agents discover these capabilities, negotiate what they can execute, and proceed with the transaction using only the features both sides support.
This is fundamentally different from traditional API integrations where every retailer needs to support the entire specification. With UCP, a small business can implement basic checkout while an enterprise retailer can expose complex loyalty integrations—and both work within the same protocol.
The payment architecture is equally clever. UCP separates payment instruments from payment handlers, which means the protocol doesn’t need to define every payment method. Instead, each payment provider—Google Pay, Shop Pay, PayPal, regional payment processors—publishes their own handler specification. Merchants advertise which handlers they accept, agents pick one that works for the user, and the transaction proceeds.
New payment methods can join the ecosystem without requiring protocol updates or committee approvals. That’s how you achieve scale.
The roadmap reveals where this is heading. Google’s documentation indicates upcoming features include multi-item carts, account linking for loyalty programs, and post-purchase support for tracking and returns. The protocol is designed to be “modular and extensible to support rich commerce experiences using capabilities and extensions,” meaning it can evolve to handle increasingly complex shopping scenarios without breaking existing integrations.
The Market Opportunity Is Massive
Bain estimates the U.S. agentic commerce market could reach $300 billion to $500 billion by 2030, representing 15% to 25% of total online retail sales. Gartner forecasts that AI agents will intermediate more than $15 trillion in B2B spending by 2028.
Seem radical? Half of all consumers already use AI when searching the internet. ChatGPT alone has more than 800 million weekly users. The shift from research to transaction is happening now.
AI and autonomous agents influenced about $3 billion in U.S. Black Friday sales according to Salesforce. What begins as AI-mediated discovery increasingly carries through to execution, as AI agents compare options, assemble baskets, and complete checkout.
What This Means For Your Business
If you’re still optimizing your e-commerce business for keyword rankings, you’re solving yesterday’s problem. Here’s what you need to do now:
- Structure your product data for agent discoverability. AI agents don’t search like humans. They traverse relationship graphs and evaluate structured attributes. Your product information needs to speak the language of entity relationships, not just keyword optimization. Google is rolling out “dozens of new data attributes in Merchant Center designed for easy discovery in the conversational commerce era,” including “answers to common product questions, compatible accessories or substitutes” that go beyond traditional keywords.
- Build for conversational commerce. AI Mode in Search now pairs Gemini models with Google’s Shopping Graph to handle conversational shopping queries. Users describe products in natural language instead of relying on keywords and filters. According to Google’s announcement, UCP will “soon enable direct checkout from Google’s AI Mode and Gemini App, allowing users to complete purchases using Google Wallet without ever leaving their conversation with AI.” Your optimization strategy needs to account for how agents interpret intent, not just how they match terms.
- Prepare for new attribution models. When AI agents intermediate purchases, traditional attribution breaks. You need systems that track agent-mediated transactions differently than click-through conversions.
- Understand the security requirements. Nearly 80% of financial institution leaders surveyed by Accenture expect fraud will increase due to agentic commerce. Visa announced it has successfully completed hundreds of secure, agent-initiated transactions using the Trusted Agent Protocol, an open framework designed to help merchants distinguish between malicious bots and legitimate AI agents acting with consumer authorization.
The AI Infrastructure War Is Just Beginning
Amazon is betting on proprietary systems and marketplace control. OpenAI is building shopping features into ChatGPT. Perplexity is experimenting with commerce integration. Meanwhile, Google is building the infrastructure layer everyone else will need to build on top of.
Google isn’t trying to own the transaction. They’re trying to own the protocol layer that makes agent-mediated commerce possible at scale.
For businesses ready to integrate, Google has published a full integration guide and merchant interest form through Merchant Center. The protocol specification is available on GitHub, where developers can participate in discussions and contribute through pull requests.
The businesses that recognize this shift early—that understand they need to optimize for agent discoverability rather than search rankings alone—will capture the next wave of commerce growth.
The businesses that don’t will wonder why their traffic disappeared.
The rails are being built right now. The question is whether you’re building on them or getting left behind.