Welcome Fort Knox. I am John Ford here with AWS CEO Matt Garman. We were just on CNBC a few minutes ago, but thanks for a little bit more time so we can go a little bit deeper and understand some of the mechanics that you guys are working with. So, the street was expecting something in the neighborhood of, I don't know, 140, 150 billion dollars in spend from Amazon capex for the year. Closer to 200 billion is the number that you guys put out. I can't even wrap my head around how you spend that much money. I mean, if you're buying whole towns to build data centers, it's hard for me to understand it. So in the mix of where that investment goes, give me a sense of when you're looking at we could spend this much or we could now let's go for the 200 billion number because we're going to do what in '26.
Yeah. Well, and here's one thing that's important I think to understand too, that this is investment for the next couple of years, right? And so it is as we're starting to think about the companies that are relying on us to provide compute infrastructure so that they can grow, so that they can modernize, so that they can implement their AI capabilities, our now planning horizon is longer. And so we're thinking about how do we have enough capacity for '26, but also much of this investment is building capacity for '27, even '28. And so as we're thinking about that, you know, a lot of where people are saying is how can you spend all that money in '26, it's not for compute in '26. When you have to build data centers, when you have to build power generation, when you have to build thinking about a bunch of those things, it's some of those are many year out investments. And so many of these are start, you won't see revenue for a couple years. But our job is to think about how we can ensure that as, look, today by all estimates less than 20% of workloads are on the cloud. And when a workload is not on the cloud, it can't take advantage of the AI revolution that's happening right now. And customers have a really hard time of building agentic workflows with their data when it's not in a cloud world. And so we're seeing this tailwind effect where because people want to take advantage of AI, they're rapidly, even faster, moving their workloads to the cloud. And so we see this kind of flywheel effect. And that's really leading us to go and invest. And that's part of why you see our business accelerating, right? On a really, there's, it's not that often that on a business the size of $140 billion, you see revenue growth acceleration. And that's what you're seeing with AWS. And you see that growth coming. And so we're quite bullish about where we see the opportunity for us. It's also not, you know, we worry a lot about concentration risk. How do we make sure that we're not, you know, our revenue is not dominated by just one customer. It's spread across a broad swath of customers. And that also gives us confidence that what we see is real demand for now and over the foreseeable future.
A lot of what made modern Amazon possible, particularly on the e-commerce side, was an early investment in logistics. I remember the beginning of my career as a tech reporter covering when Amazon bought this Fruit of the Loom factory in Kentucky to build out a distribution center. It was a big deal because the town had been sort of devastated by that exodus of manufacturing jobs and Amazon was still a really new book seller. When you at the highest leadership levels of the company talk about this capex investment, how much does that quarter century ago investment cycle play in? To what degree do you think you're investing differently than the rest of the market because you're Amazon?
Yeah, I think it's interesting. I think it informs for us actually one of the things that we've learned from that history, and I think it's illustrative as to the benefits that your company can grow and benefit from that over a long period of time. I think the other thing that we've learned as a company is how do you make these large capital investments in a fiscally responsible way so that you eliminate and you limit risk as much as possible while giving yourself options to be able to grow and take advantage of that upside. And I think one of the things that we saw in the early e-commerce days, and this was before my time at Amazon, but when we really saw that was when you see a trend where lots of shopping is moving online, you can bet on that trend that more of it is going to come to you if you deliver a better experience for customers. And for us, we see most compute workloads are going to want to move to the cloud over time. And if we can have the very best infrastructure, which we have today, the very best set of services, which we have today, and the broadest footprint, we think we can continue to capture the lion's share of that business, which is what we do today. And so when we think about investment, actually if you think about investment compared to our peers, it really is on a relative basis. And you kind of think about, you know, how you can, and everybody is investing a lot because they see this opportunity out there. And even with all of this investment, my best estimation is we will be capacity constrained for the next couple of years. We will sell every single server and every single bit, and we will wish that we had more. And that I think is where the state of the world is going to be for at least the next couple of years. And so we are on this, you know, look, AI is a step change in technology. It's a brand new infrastructure. It's a buildout. It's driving a bunch of these pieces. And so we are investing to make sure that AWS maintains our leadership position in that. And we feel quite good about the prospects there.
One might have said something similar about capacity constraint in e-commerce as COVID was kicking off and indeed there was capacity constraint but then there was also an overbuild. What potentially happens once that capacity constraint opens up? What kinds of gears are you prepared to shift in order to make sure that you continue to have that fiscal responsibility?
Yeah, there's a couple things I would say to that. One, number one is there was that overbuild and now Amazon has grown into it. So, it still is a short-term issue as opposed to a longer-term issue on the retail side. And the AWS business is different, right? In a retail scenario, if you buy a book, you don't necessarily buy another book tomorrow. You may or may not if you need another book, but you don't necessarily. For AWS, it is largely a recurring business. And so when we win a workload from a bank, you know, it can fluctuate up and down based on their volume, but largely, or we could eventually lose that business if we don't do a good job of providing service. But for the most part, these workloads build on each other. And so you don't have, you're not going to have the same drop in the same way that you may have consumer spending that is, you know, it is a single transaction you have to come back for. It is a little bit more smooth in that way. And so for that, and we also have much more kind of visibility into what our customers are doing from an Amazon perspective. We have hundreds of millions of customers that we have no forecast into what their future purchases are going to be other than, you know, general, you know, we talk to customers, we know that they're happy, etc. But I have no idea as a consumer what you're going to buy on Amazon for the next week or for the next month or for the next year versus our enterprise customers. We talk to them. We know what their workloads are. We know what their migrations are going to be. We have long-term contracts with them where they're going to be spending for the next couple of years. And so there's some mechanics of the business that are quite different that actually limit risk for AWS in a way that's not true of our retail business.
How has your go-to-market changed as the AI era unfolds? We have all these tools that we talk about around sales, around identifying where customers are satisfied or where there might be opportunities that's out there in the marketplace, but we don't necessarily hear how some of the biggest players that are trying to drive this capability are changing the way their people get out there and provide this service. How are you doing it?
Yeah, there's a couple things I would say. Number one is we use AI to eliminate as much of our field's time spent on non-customer-facing activities as possible. And so, you know, if I go back a couple of years, we were lucky if 25% of our field time was spent in front of a customer actually working with them. They were doing things like, you know, filling out Salesforce forms, building account plans, building presentations, doing all those kind of work and then they would actually go and talk to a customer 20, 25% of the day if we were having like a really high functioning sales team. Our view is if you can even eliminate parts of that 75%. And this is what we've done is basically say okay now AI can do this where you used to build an account plan it would take you 3 days now you can do it in 10 minutes. It used to be you would put a plan together for talking points of what you're going to go talk to the customer. You'd have to do a bunch of research, pull a bunch of stuff that might take a couple of days. Now you can do it in hours. And now you can support more customers. You can drive more sales that way. And we're seeing meaningful results there. And now, and actually we find that our customers need more time with us because in addition to just talking about technology, right? We can talk about the latest compute we have and the latest storage we have and that's important, but we're also talking about transformation with customers now and we're understanding how their industry is going to transform, how what they do is going to change. And so there's more time that we spend to them actually frankly in almost like a consultative basis just working with them to say how can technology really change the fundamentals of what your business is doing. And so AI has allowed us to not, it doesn't like we get rid of salespeople. It has allowed us to shift what they do to really helping customers take advantage of more of this technology by spending more time with them and walking through and showing them what's possible.
Is that mostly AI that you are building and adapting and layering on top of the software that you're buying out there to use for your Salesforce and dealing with customers or is it mostly solutions that third parties are providing to you?
