The payments industry has a time-worn pattern: it works itself into full-blown hysteria over a technological change, then waits, and waits, and waits for that catalyst to show up in the numbers.
The most recent example was mobile wallets. The carriers, the retailers, the phone operating-system providers, and hundreds of startups all raced to build one. Mobile payments were going to fundamentally change how consumers paid, and in doing so, reshape payments market share and economics.
The timeline tells a humbler story. Apple Pay launched in October 2014. ISIS, the carrier-backed wallet later renamed Softcard, launched nationwide in November 2013. Google Wallet had already launched in 2011, and the banks eventually rolled out their own “Pays.” Yet from the first launch, it took years for mobile-wallet payments to break just 1% of volume.
Start with the macro view. Of the $17 trillion in digital U.S. payments, 67% are card-present: a customer paid for a good or service in person, in the real world. E-commerce is growing more than 3x faster than card-present, and it still accounts for only around 15% of payment volume, more than 25 years after it was introduced.
So which large bucket of payment volume is agentic likely to touch? Unless you forecast an explosion of robots carrying payment credentials moving feverishly through the physical world to buy things, the agentic impact will be felt first in e-commerce.
So what should a platform actually do?
If you own or operate a vertical software or AI platform, here is how to think about it.
If you monetize the payment volume flowing through your platform, getting card-present right has a far greater return than chasing agentic, now and for a long while. If your customers carry a traditional mix of roughly 67% card-present / 33% card-not-present, growing your attach rate and wallet share by 50% will almost certainly generate an eight-figure increase in enterprise value. Playing at the bleeding edge of agentic standards, working out who owns the liability for a disputed agent-initiated transaction, is a waste of your focus.
If you’re a pure-play online retailer, the answer may differ. That’s the core customer of Stripe and Adyen, not platforms. For those merchants, agentic commerce could mean lost share if the models are originating purchases on a customer’s behalf and choosing one brand over another.
And if you’re a business operating system for small and medium businesses? Grab some popcorn. Let the global e-commerce giants work out how to migrate a three-party payments universe to accommodate a non-human fourth party. You have a more valuable problem to solve in the meantime.
