Agentic commerce is here - but the margins aren't
Who is positioned to win AI shopping profits?
At the 2026 Upfront Summit, Harley Finkelstein called agentic commerce the biggest transformation in Shopify’s history - and described the TAM expansion it will generate, far beyond today’s digital commerce.
It makes sense - if we free shopping from the click-through checkout process on a retailer’s website, then commerce suddenly appears in far more places where there is intent. Commerce shifts from destination behavior to ambient behavior.
But just a week later, OpenAI appeared to be pulling back from the idea that ChatGPT would directly own in-chat commerce, according to reporting from The Information.
This contradiction just demonstrates the reality of retail economics. While transactions may expand dramatically, the capture of profits will not.
Retail is not a high-margin industry. According to sector data compiled by Aswath Damodaran at New York University Stern School of Business, the median gross margin for general and specialty retail is 33-35% - not terrible, but half of what margins are for classic software businesses.
And those margins already support a crowded stack of intermediaries - payment processors, marketplaces, logistics providers, advertising platforms. To support a new agentic commerce player, someone else has to get bumped - there’s not more margin to go around.
If ChatGPT or Perplexity or Daydream is getting a cut, someone else probably isn’t - a Shopmy influencer? A marketplace referral? Even a retailer? To own the transaction and take that margin will require that it come from somewhere.
So who is positioned to be the big winner of all this TAM expansion? Shopify and the other infrastructure players of course. NOT the new demand surfaces.
Shopify’s view is that AI agents simply become another demand surface plugging into their backend. The company has been promoting a concept called the Universal Commerce Protocol, which allows agents to connect directly to merchant infrastructure for inventory, pricing, checkout, and fulfillment. In the same way Shopify supports social commerce, influencer commerce, live commerce - it can also support agentic commerce. So from Shopify’s perspective, the new world of agents isn’t really disruptive - it’s just a growth engine.
That may explain why the companies building the most powerful AI agents are not rushing to become commerce platforms. There are many other opportunities for those companies to capture much bigger margins - and operating a full commerce stack is more like running a global financial network than building a software interface.
So who else besides Shopify is positioned to win agentic commerce? Companies that already operate the messy, operational layer of commerce - inventory, payments, checkout, fulfillment. Companies like Max Retail (I’m an investor), founded by Melodie van der Baan - who has built an uparalleled supply network of in season and near season fashion inventory - and made it available to any sales channel, including your AI agent. Or Constructor - a company founded a decade ago but well positioned to support agentic commerce. Or a company like Rebuy - focused on repeat purchases, where an AI agent has a natural role to play. I also think the established marketplaces with unique supply will be winners - Etsy, eBay and of course our friends at Amazon.
In this environment, the companies most likely to win are not necessarily the ones creating the agents. They are the ones already running the rails underneath them. Side note - this is is part of our investment thesis at Everywhere Ventures - where we invest in second-order effects businesses. If the big, game-changing innovation is agentic commerce, we look for the second order effect - the infrastructure companies that will capture margin created by that new market. If you are early building in agentic commerce infrastructure - real time inventory management and data aggregation, order routing, fraud detection, agent payments - we’d love to chat.
Anna Barber is former Managing Director of TechStars, former partner at M13, and a General Partner at Everywhere Ventures.