Mastercard CEO: AI Shopping Agents, Machine-to-Machine Payments, and the New Infrastructure of Commerce

**Michael Miebach** (0:03)
Who sits in the middle and drives interoperability, makes sure all of this connects, and is not a spaghetti, a plate full of spaghetti? Astercard. So, we're closing this quarter and acquisition of a company by the name of PVNK, which is the largest stablecoin platform out there to connect all of this for the world.

**Bart Shannon** (0:27)
That was Michael Mebach, CEO of MasterCard on why the fragmented world of stablecoins and digital payments need someone to hold it all together, and why MasterCard intends to be that someone.
I'm Motley Fool producer Bart Shannon. Last week in part one, Tom Gardner and Michael covered how MasterCard's payment network works, the $15.6 trillion cybersecurity threat, and why stablecoins are more opportunity than threat. This week in part two, they get into the future, AI shopping agents, machine-to-machine payments, how MasterCard thinks about capital allocation, and what the AI revolution really means for employment and consumer spending. Hope you enjoy.

**Tom Gardner** (1:10)
Let's talk about agented commerce, where AI becomes the customer's main relationship, and that that transaction then starts to look for the cheapest alternative. Maybe we can talk about AP4M as well.
I don't know if GPT should have allowed this, but I created an image here of AP4M. It actually put the MasterCard logo in the center there, which makes it questionable as whether GPT should be able to place logos. But anyway, I'd like to hear a little bit about AP4M and I will stop my third screener there.

**Michael Miebach** (1:47)
Let's talk about agentic commerce first and hang it up a little bit higher before we come to AP4M. I can also decipher what that actually means when we get there.
Agentic commerce, what's actually happening is, in straightforward online commerce, what happens today is you might go and go to your favorite search website and just put in whatever you're looking for. I will give you a bunch of sponsored links, and then it gives you a set of more links, and then you pick where you go. Or you might be just irritated by all of that and just go straight to your favorite marketplace. Those are all things that we probably have experienced as consumers. Now, you just referred to an LLM where you created this lovely image just now, Chetjibiti is the example that you used. So, what you might want to do today is get a potentially better and more holistic answer for whatever you're trying to do. Let's say you want to go on a camping trip and say, what do I need for a camping trip? And it gives you 15 things. And because it knows your history, it could have given you 25 things, but it already knows you have a tent. So it's not offering the tent to you, it's giving me all the things you don't have, that for that particular destination actually makes sense.
So if you have that result, then imagine you still have to go to every single website and say, all right, I'm going to find that tent now, or I'm going to find that little cooker or whatever the ingredient is, and then you're all over the place and you have just wasted half a day planning your trip. Wouldn't it be much easier if you could check out right in that moment on whatever that recommendation is from the favorite LLM of your choice, and say, I'm going to check out right there. I'm going to delegate the checkout to the agent, to an agent, which in this case would be the LLM that does all the checkout for you, and it uses a mastercard behind it and everything works.
Would be so much easier. Now, for that to happen, we need to recognize that suddenly there's an entity in between that never existed before, that's the agent. Then back to cyber security, how do we know that this agent is actually an agent that is known and it's not a fraudulent agent? That the agent is actually what you're trying to do, buy stuff for your camping trip and not ordering something else.
Or that the agent actually makes mistakes or not. It orders two grills instead of one. How do you prove that? In the end, when you can't, it would be debited. So there's all things that we thought about that today, in the world of MasterCard payments, straightforward payments work very easily. You have a charge back. You're always protected. You say, that never happened. I never ordered that. And you just undo the whole thing. So we created AgentPay. AgentPay is basically a protocol that ensures that an agent is recognized, is accredited as an additional party in the ecosystem. So there could be an LLM or it could be a very large retailer that has an agent for all their brands, et cetera. So that's registered. The next thing is that this transaction is fully tokenized. What that means is every bit of data associated with this transaction is captured. So there can be as a proof point to say this is exactly what the consumer wanted to do. Then the various parties in the ecosystem, the bank of the merchant and the bank of the consumer, everybody knows and it flows the same way as it does today at a MasterCard transaction. So it's very, very technical. So I'm going to keep it at that level. So all of this is what is happening and the ecosystem is ready. And these transactions are starting to flow now. So for us, is this a growth opportunity? It is. It is because tokenization is a service that we sell. All these transactions are fully tokenized, which is very different than the real world today. But not every transaction is tokenized just yet. So that's a growth opportunity. The related cybersecurity solutions for these transactions is a growth opportunity for us, etc. Will people buy a lot more?