AI Will Replace Some Jobs, But Create New Ones Too, Says Amazon CEO Andy Jassy | Exclusive

It's great to be here. Thank you for having me. I've been to the country many times and it's exciting to be here.

We are. So, as you said, we've invested about 40 billion dollars since 2010 in India. And we announced another 35 billion by 2030, and we're upping that by another 13 billion. So, an additional 48 billion total by the year 2030.

I'd say this additional 13 billion will be largely AWS and cloud and AI and just the pace at which AI is growing and the pace at which AI is growing here in India is really remarkable.

Well, I think AI is the most transformative technology of our lifetime which is saying a lot because we've lived through the internet and through mobile and the cloud. But it's going to change every single customer experience that we know and it's going to enable all sorts of new ones that we only imagined before. And I think you can just look at the rate of growth. When we started AWS and I was part of the team that started AWS 20 years ago, the first 3 years of the AWS business we felt like we were growing really fast and the business was a 57 million dollar annual run rate business. If I look at the first 3 years of this wave of AI in AWS it's over 15 billion dollars annual run rate. That's 260 times larger than what we had at the beginning of AWS. So it's growing unbelievably quickly and if you look at the last quarter we grew 28% year over year in AWS and on a very large 150 billion dollar annual revenue run rate. That's very unusual. And so, the way that the cash cycle works in a business like AWS is that you have to lay out the cash for the data centers and for the hardware and for the chips and the networking gear in advance of when you can monetize it. And so, that means that in the early years, when you're investing a lot of capital expense, until your revenue growth starts to outpace your capital expense growth, you'll have headwinds in the free cash flow side. But, once your revenue growth starts to outpace the CapEx growth, which will happen in a couple years, you really like the return on invested capital and the free cash flow. And that's what I think will happen here. We've had an example of that in the first phase of AWS, where you really like the operating income and the free cash flow and the ROIC. It's following the same pattern in AI. And I think because we have this really unusual opportunity where everything's going to be reinvented with AI, if we're able to successfully get customers capacity with the right services at the right service levels, our business 5 to 10 years down the road will be much larger on a revenue side, on the operating income side, and on the free cash flow side.

Well, I think that we will end up generating much larger free cash flow for Amazon and for our shareholders in the next 5 to 10 years. I don't know exactly which year, but typically what happens is you invest a lot of this capital up front, and then you really like the free cash flow returns a couple few years down the road. And I expect the same thing to happen here.

Well, India is an area of very high growth for AWS, and for the company as a whole. On the AWS side, we have hundreds of thousands of customers here in India who use AWS. The rate of growth there both on the AI side and non-AI, which we call core, is very substantial. We have a fair number of GPUs and AI accelerators here for startups and individual developers and researchers and enterprises and government entities to use. So, I think we'll be a very big part of our growth that's a lot of what the incremental $13 billion we're investing in India is about. And then we're continuing to grow so fast on the marketplace side here. And you look at just the growth in quick commerce as an example. So, we have the broadest selection of products that get to people the fastest. So, we can deliver thousands of items to customers in minutes, tens of thousands of items to people in a few hours, over a million items we can deliver same day, over 4 million next day. It's very unusual and what we find with how fast our quick commerce is growing, where it's doubling every quarter, Prime customers who've tried quick commerce are buying three times more frequently than Prime members that don't. And so, we're thrilled with the growth here and we have a lot in front of us in India.

Well, you know, remember we've been experimenting with lots of different delivery models in India and outside of India for many years, including a form of quick commerce several years ago. And what we have to do at our scale and with the number of customers we serve, we have to find delivery models that really work for customers and an economic equation for the company. And so it took us a little bit of time to find the equation we like, but with quick commerce, which we started about a year ago, the rapid pace of adoption is pretty amazing. And if you layer on top of what we have in quick commerce, the broadest selection that's out there at the shortest delivery times that are available. And you layer on top of that prime, which is a very unusual subscription and membership program where customers get free and fast delivery on the broadest selection of items. They get Amazon pay with cash back benefits. They get all the content you get in prime video and in prime music. You layer on top of that the broad selection, the fast delivery, the quick commerce growth, the prime and then the announcement we just made today, which is we're now extending our Amazon now quick commerce offering to over 300 cities here in India. And it's a very powerful equation for customers and that's why it's growing so fast.

Yeah, I think we've done a lot of experimenting before in different fast delivery models. Everything from different quick commerce models to we're continuing to experiment with drones. And we have found an equation that both resonates with customers and is economic for us and so much so that we're expanding it very rapidly.

We, you know, what's really interesting you mentioned it earlier is that so many of the inventions that have been born here in India that our team here in India have experimenting with and found success with are being taken to other parts of the world. And I do think that quick commerce we're already now meaningfully experimenting with and expanding quick commerce in the US and in Europe, but it started here in India. And I do expect that quick commerce will be in virtually every marketplace in which we're in. And that we'll continue to find inventions in India that we're able to take to other parts of the world.

Well, of course the most important thing is it's amazing for customers to get this really broad selection in minutes. And then on the back end, we have these micro fulfillment centers. They're much smaller units. They're organized differently. I was at one this morning and I had a chance to watch it in motion. I made an order that was delivered to me in a little less than 4 minutes.

It's incredible. It's just a magical customer experience. And I think also we're doing some things that are really different. We're not only delivering quickly for customers, but we're doing it in a way that's great for our Indian employees. If you look at our safety efforts, what we do in terms of having academic scholarships for our employees' kids to be able to get education, providing health insurance, what we've done with ashrae's where we've got these mobile units for both our own Amazon now employees, but even other employees at other companies where on a hot day they can come in, it's air conditioned, they can charge their phones, they can use the restroom. And so we're trying to grow in a really responsible way for the country and I think we're having a lot of success so far.

Yeah, I think one of the amazing parts about our marketplace here in India is that we have over 1.7 million third-party selling partners who together we bring this experience to consumers. And one of the things that we wanted to do and that we committed to the government here and in the country was that we wanted to help Indian sellers be able to reach other consumers in other countries. And so we took this ambitious goal of 80 billion in exports. We're already over 20 billion and we're well on our way. I think we'll get there.

I think sovereignty in countries has been an issue since the beginning of our offering AWS out outside of the US, which goes all the way back in the year 2008. And I think we have always shared with customers and governments that the physical infrastructure we put in countries, it's secure, it's ruled by local law, the data that customers have, it's their data. It doesn't move unless they choose to move it. Our employees don't have access to it. And then for our customers who have extra concerns about data sovereignty, we give them multiple ways to encrypt the data. They can encrypt it in motion, they can encrypt it at rest. And they get to keep the keys, our customers, so that even in the very odd event that one government wants to try and get access or seek subpoena access to data, if it's encrypted, they don't hold the keys, they have no ability to do so. So, that's really been true for a long period of time. It remains true today. And I think for most companies and governments to not have to lay out all the capital for your own infrastructure, for your own data centers, for your own hardware, for your own network gear, to not have to rebuild all the software services with all the features associated with it and be able to use it as you consume it, is a very compelling proposition to provide value for your customers.

You know, I think it's interesting. At the same time that we've seen a number of companies have role reductions, those same companies are still hiring a lot of people. And so, I think what's happening is as I said earlier, AI is the most transformative technology in our lifetime and it'll change every customer experience we know, but it'll also change every job we all do. I think the way humans do their jobs will be very different every year for the coming next 5 to 10 years. And I think it'll mean that you'll have fewer humans doing some of the jobs we've had humans doing for a while and you'll have other humans doing other jobs, some of which are altogether new jobs, some of which we haven't even figured out yet. So, I think that you'll find the amount of coding that's being done today by agentic coding, the amount of customer service that's being done by agentic customers, that's all growing at a very rapid rate. But, then again, I look at how many subject matter experts are being hired to train these models. People who are experts on law or finance or medicine, those jobs didn't exist before. So, I think with every big technology transformation it shifts what people work on and it creates new jobs. I think the same thing will happen here.

Well, as fast as we are adding an incredible amount of capacity in our data centers and as fast as we're adding that capacity, it is being monetized. So, there is a lot of pent-up hunger for doing AI and, as I mentioned in AWS, we grew 28% year over year on a $150 billion revenue run rate and we're still dropping demand on the floor. We still don't have enough supply for all the demand. So, there is a very high demand for it right now. And I think in the medium to long term, what always happens will happen here, too, which is that you have to build amazing customer experiences that people want and that warrant the investment incrementally in whatever you're spending. And I have a very strong belief, and it's one of our missions at Amazon, that the cost of inference in AI needs to come down significantly. It's a big part of why we're investing in our own custom silicon and training, which is about 30% more price performance than other comparable AI solutions and chip solutions. And it's also why we're investing so significantly in our inference service bedrock, which not only has the leading array of frontier AI models, but it also does inference really cost-effectively and in a high-performance way. And the only way that you'll see AI rolled out as expansively as I believe and a lot of people believe is if the cost of inference continues to come down and companies find that they could have the right return on invested capital equation for that. And I strongly believe that's going to be the case for AI.

We have for a long time been wildly impressed by the entrepreneurial spirit here in India and just the way the population has grown up being mobile first and digital first. Very strong technology universities here, very strong technical ecosystems. A number of companies have fueled that. And as such, you see a number of very impressive startups. And we have been lucky enough to invest in a number of them. I think we'll continue to go hard at where we can add value to those startups and we'll see over time.

I'll be here. Thanks for having me. I appreciate it.