Well, Andy, thank you for joining The Information here in Davos. It's great to see you.
Great being here. Thanks for having me.
So I think we last spoke around two years ago and my, have things changed. At the time you were talking about Amazon's unique position in AI being all parts of the stack from the chips, the data centers to the models to the applications. So we'll get into all of that and I'm excited for the updates. I wanted to just start with shopping. You know, in recent weeks, we've seen OpenAI, Anthropic, many others try and make big pushes or big talk about the power to use their chatbots for commerce. I'm wondering, is that a threat or opportunity for Amazon?
Well, first of all, it's great to be here with you. I appreciate the time. You know, I think that we're excited about agentic commerce. I think that it has the chance to make it easier for customers to find what they want. You know, if you know what you want, it's pretty hard to find a better experience than popping onto Amazon and searching and finding it. But the one place still where physical retail has some advantages in my opinion is the ability to go in, not know what you want, ask questions, refine those questions, have somebody point you to different things. And I think agents are going to help customers with that type of discovery. And it's part of why we've invested so much in Rufus, which is our shopping assistant, which has really gotten quite good. And I think that over time we will work with other third-party agents as well. I think today the experience hasn't been great yet. You know, I think that a lot of these third-party agents, they don't have your buying history. They don't have what you like. A lot of the information about pricing and the product is off. But over time I do believe that we'll get better. I also think there needs to be the right value exchange between the agents and between the retailers themselves, but I am optimistic that those will work out. We're having conversations with lots of people and I'm very bullish on agentic commerce.
And so those conversations you mentioned, some of the economic terms someone likened it almost to like we need the tariff system set up. I don't know if that resonates at all, but I do wonder why do a deal with OpenAI and commerce? You mean, you're Amazon. Like what would be the benefit?
Well, I think that when you get down to what really matters for customers, which is what we spend all our time thinking about. I think in retail, it's going to be really broad selection, which agents are pretty good at aggregating, but then it's really low prices, very fast shipping, taking care of the customer the right way, having your personalized history so we put the right things in front of you and can make recommendations. And those things you don't really get in agents typically. You know, people who are regularly shopping on Amazon are going to find the low prices they want on Amazon, the really same day or next day shipping on Amazon, the personalized history they know will take care of them. So, you know, I think what'll happen over time is that you'll find most retailers will have their own agents and then you'll also have third-party agents and we'll see over time how many people want to shop their retail on kind of a generalized agent versus where they go every day and seeing the product and having, if it's a good agent like what we've been building with Rufus, which will keep improving, you're right there and you can one-click shop. And so I think a lot of people are going to choose to use our agent and then when we can find opportunities to make the experience right for customers with more generalized agents, we'll have those conversations too.
Stay on OpenAI for a second. They've said they're going into the advertising business. So I mean you can envision a world or they're probably envisioning a world where I'm looking for new shoes on OpenAI and they're giving me an ad for those shoes. How does that affect Amazon's advertising business?
Well, I mean, look, there have always been lots of ways that you can find products. I mean, as meaningful a business as our retail businesses, it's still just about 1% of the worldwide global retail market segment. And so, there's lots of places that you can do research on products, that you can choose to buy products. And as such when you're displaying lots of different pages to people you have the opportunity to introduce them to new products. But people still largely start with Amazon who shop on Amazon. I think in part because we have such broad selection and low prices and really fast shipping. And I think one of the things that matters with advertising is consumers aren't that tolerant of you just putting an advertisement in front of them that has nothing to do with what they're looking for. And so our advertising team spends most their time and is mostly people who are really AI and machine learning experts to make sure that we're putting items in front of people that are relevant to what they care about both what they've searched for and what they've bought before. And so we continue to work hard in that. I'm sure others will as well but it's not simple to do that well.
And one more on the topic, you know, The Information reported that you've been in talks to invest in OpenAI. Anything to announce?
Well, you know, we, as you probably can imagine, there are a lot of rumors about us and about lots of other companies, and we don't comment on those. But I will say that we have a lot of respect for OpenAI. We just did a very significant agreement with them.
You didn't, which I think I yelped when I saw it because that was one of the things. Yelp might be an overstatement, but...
Yeah. But two years ago, they weren't a current partner. Obviously, Anthropic, which still is, is a main partner. So you did the cloud deal and...
We, you know, I think we had many conversations with OpenAI and we have a lot of respect for them and I think both of the teams really wanted to find a way to work together and we were both happy that we were able to build that agreement and I hope that we have a chance to deepen the relationship over time.
Okay, let's talk chips. I know they're the hot topic in the industry of many. You've had a lot of news around Trainium, your chip, and talked about the much lower cost of what's out there. What are you seeing in terms of new customer adoption and are there any big new customers coming online for it?
Well, you know, as you know, the chips are such an important part of the performance and the cost structure for people running technology infrastructure. And we learned in the CPU side of the business, we had this deep relationship with Intel, which we still do, but when you have a significant leader, it's not always their priority to take price performance down for customers. And one thing we learn about customers over and over again is they want better price performance. And so we built Graviton, our own custom CPU silicon, which is about 40% more price performant than the leading other x86 processors. And that has been really great for our customers and business. And about 90% of our top thousand customers now use Graviton in a very significant way. And we just saw this same movie happening in the AI space. And we have a very deep partnership with Nvidia and we will for as long as I can foresee. But customers badly want better price performance. And so that's why we built Trainium. Our Trainium 2 chip has been fully subscribed. Anthropic runs hundreds of thousands of Trainium 2 chips as they're training their next model of Claude on top of it. And it's a multi-billion dollar business. And we just released Trainium 3 which is our next version of chip which is 40% more price performant than Trainium 2. And Trainium 2 was about 30 to 40% more price performant than the other leading GPUs out there. And so it's just, if you want to allow customers to be able to use AI as expansively as they want, you must take the cost of inference down and the chip is a big piece of it. And if you're building a big inference business like we are, if you want to have a business that not just lets customers pay less, but also where you have sustainable margins over time, you're strategically disadvantaged if you don't have your own custom silicon.
Do you see a world in which AWS is primarily a Trainium cloud?
I suspect that for as far as I can foresee, we're going to always have customers who want to run other chips, whether it's Nvidia or AMD.
That seems to be the one they want to run still.
Yeah. I mean, you're going to still see that. But I think you're seeing a pretty big shift in how people are using what chips they're using. I mean, if you just look at Bedrock, which is the significant part of our inference, Bedrock already is predominantly Trainium running underneath it. And so I suspect over time that you will see us have quite a bit of our usage as Trainium. And then I expect we'll have other chips that run very successfully for customers as well.
On the data center side, there's been so much news around energy and questions about that. President Trump said several days ago that he wants big tech companies to bring their own power, right, to be behind the meter. What's your reaction to that? And is that really going to increase the cost of building a data center?
