WEBVTT - Amazon CEO Andy Jassy Talks AI 0:00:02.360 --> 0:00:07.480 Bloomberg Audio Studios, podcasts, radio news. We now want to 0:00:07.520 --> 0:00:11.720 welcome to our TV and radio audiences worldwide. Amazon's president 0:00:11.760 --> 0:00:13.760 and CEO, Andy Jesse, Great to have you. 0:00:13.720 --> 0:00:15.239 On New York. Thanks for having me. It's great to 0:00:15.240 --> 0:00:15.600 be here. 0:00:15.520 --> 0:00:17.919 Back home where you grew up at and you are 0:00:18.120 --> 0:00:23.120 unveiling generative AI infused Alexa Plus. How is this going 0:00:23.160 --> 0:00:26.599 to really well excite your customer base? What are you 0:00:26.600 --> 0:00:27.520 most excited for it? 0:00:27.680 --> 0:00:31.120 Well, you know we've had Alex has been around for 0:00:31.120 --> 0:00:34.560 ten years. We have six hundred million devices in customers 0:00:34.600 --> 0:00:39.840 homes and offices, and Alexa Plus is our next generation 0:00:40.159 --> 0:00:44.480 Alexa Personal Assistant, and she's meaningfully smarter and more capable 0:00:44.479 --> 0:00:47.320 and useful than her prior self. You can do all 0:00:47.360 --> 0:00:49.240 the things that you used to do, but every single 0:00:49.320 --> 0:00:51.360 one of those functions is better. And I can give 0:00:51.400 --> 0:00:53.280 you just an example. There are so many examples, but 0:00:53.800 --> 0:00:58.800 you know, if you have smart home controls with Alexa now, 0:00:58.880 --> 0:01:02.720 you can say, Hey, Alexa, I have guests coming over 0:01:02.840 --> 0:01:08.000 at seven pm. Could you raise the drapes, could you 0:01:08.640 --> 0:01:11.360 raise the temperature by five degrees, turn on the porch 0:01:11.440 --> 0:01:14.440 lights and the driveway lights, and put on melo dinner 0:01:14.520 --> 0:01:17.640 music in the dining room and music you're not me. 0:01:17.760 --> 0:01:20.160 I choose shop Fighters, but most people will choose Melo 0:01:20.280 --> 0:01:22.920 dining music. But you can do that all verbally and 0:01:23.000 --> 0:01:26.600 simply using conversational language like that. You don't need a nap. 0:01:26.840 --> 0:01:30.360 It just happens. And so Alexa plus, with what we've 0:01:30.440 --> 0:01:33.120 just announced and what we're launching, it really is. There 0:01:33.120 --> 0:01:35.480 have been a lot of chatbots around that are good 0:01:35.520 --> 0:01:39.279 at answering questions, but they don't take actions. Alexi plus 0:01:39.360 --> 0:01:41.400 is really Alexa is going to be the first one 0:01:41.400 --> 0:01:44.840 that not only is highly intelligent to answer various questions, 0:01:45.040 --> 0:01:47.440 but she can do so many things for you. She 0:01:47.480 --> 0:01:50.080 can play music and play video, and control your smart 0:01:50.080 --> 0:01:53.960 home and make reservations for you and hire people to 0:01:54.000 --> 0:01:57.600 fix your oven. I mean, it really is the first big, 0:01:57.880 --> 0:02:01.440 large scale practical use of gen AI. The consumers are 0:02:01.440 --> 0:02:03.120 going to be able to see and use naturally. 0:02:03.400 --> 0:02:05.400 I can see how you're going to obsess over it, 0:02:05.680 --> 0:02:07.720 whether it's ordering the latest Buffalo Wings or in New 0:02:07.800 --> 0:02:09.560 York and seeing where the best space is. 0:02:09.560 --> 0:02:10.000 To do that. 0:02:10.040 --> 0:02:12.840 But how are your stakeholders of shareholder based going to 0:02:12.840 --> 0:02:16.240 obsess about this? About gender to AI in Alexi Plus. 0:02:16.240 --> 0:02:17.639 How is that adding value for them? 0:02:17.840 --> 0:02:21.919 Well, you know, I think if you think about what 0:02:22.000 --> 0:02:26.079 Alexa allows customers to do, it makes shopping more easy 0:02:26.200 --> 0:02:28.679 because it's so much more intuitive now to buy products. 0:02:29.080 --> 0:02:33.440 It makes enjoying music easier, It makes enjoying video and 0:02:33.480 --> 0:02:37.200 streaming media better. It allows you to control your smart 0:02:37.200 --> 0:02:39.240 home in a different way. So every single one of 0:02:39.280 --> 0:02:43.639 our consumer customer experiences gets better with Alexa Plus. And 0:02:43.680 --> 0:02:46.720 then you know, of course, Alexa has its own business model. 0:02:46.800 --> 0:02:49.200 We have a brand new lineup of devices that are 0:02:49.200 --> 0:02:50.959 coming in the fall. I think that are beautiful. I 0:02:51.000 --> 0:02:54.600 think people are going to really like we have opportunities 0:02:55.440 --> 0:02:58.960 to service new products and advertising in various interfaces like 0:02:59.000 --> 0:03:01.720 our mobile and our desk top interface that's coming in Alexa. 0:03:02.160 --> 0:03:04.840 And then we have subscriptions, and you know, so I 0:03:04.840 --> 0:03:07.480 think there's a sustainable business model there as well. 0:03:07.880 --> 0:03:10.760 Talk to me about the subscriptions because you're getting it free. 0:03:10.800 --> 0:03:11.400 If you've got. 0:03:11.240 --> 0:03:13.800 Prime, I get a lot with Prime. Now are you 0:03:13.880 --> 0:03:15.639 able to increase the prices there? 0:03:15.639 --> 0:03:20.200 Do you think of Prime? Well, Prime is an incredible value. 0:03:20.240 --> 0:03:23.320 If you think about getting free shipping on three hundred 0:03:23.320 --> 0:03:25.720 million plus items. You know when we launch Prime, it's 0:03:25.720 --> 0:03:27.639 free shipping on about a million items to say, it's 0:03:27.639 --> 0:03:29.639 three hundred million items. And most of the time you're 0:03:29.639 --> 0:03:31.760 getting your products now inside of a day. And so 0:03:32.360 --> 0:03:34.400 you know, between that and what you get with Prime 0:03:34.480 --> 0:03:38.360 Video and Prime Music and the grocery subscription, and you 0:03:38.400 --> 0:03:42.840 know our unique selling events, primes and incredible values increase 0:03:43.000 --> 0:03:46.800 to add on top of it Alexa plus. You know, 0:03:46.800 --> 0:03:49.760 it's just great value. So what do they plan right now? 0:03:49.800 --> 0:03:52.480 But it's I do think Prime is unusual value and 0:03:52.480 --> 0:03:54.200 it's why people use it so expansively. 0:03:54.840 --> 0:03:58.400 And to increase that use has been this invention, this 0:03:58.520 --> 0:04:02.640 innovation of generative and that costs money. And you've actually 0:04:02.680 --> 0:04:04.880 just took to the stage yesterday to say, out of 0:04:04.880 --> 0:04:07.880 all companies, you are spending the most on AI. How 0:04:07.960 --> 0:04:09.320 much are you spending on AI? 0:04:09.480 --> 0:04:12.960 Well, you know, I think we don't disclose the exact amount, 0:04:13.000 --> 0:04:15.720 but you know we you know, we're spending a pretty 0:04:15.720 --> 0:04:19.440 significant amount of cappax and the allion's share of it 0:04:19.520 --> 0:04:23.640 is on generative AI. We've said in our aws business 0:04:23.760 --> 0:04:26.800 even though general AI for US is a many billion 0:04:26.839 --> 0:04:29.800 dollars a year business and growing triple digit percentages year 0:04:29.800 --> 0:04:32.000 over year, that if we had even more capacity, we 0:04:32.040 --> 0:04:34.520 could use it to monetize it. And we have this 0:04:34.680 --> 0:04:40.920 really interesting and very fortunate flywheel and AI inside of Amazon, 0:04:40.960 --> 0:04:44.520 which is if your mission is to make customers' lives 0:04:44.560 --> 0:04:46.800 easier and better every day, which it is for us, 0:04:47.160 --> 0:04:51.039 and if you believe that all the customer experiences we 0:04:51.120 --> 0:04:53.479 know of today are going to be reinvented through generative AI, 0:04:53.560 --> 0:04:56.200 which we also believe. You believe those two things, you're 0:04:56.240 --> 0:04:58.520 going to be building a lot of genera ABAI apps. 0:04:59.080 --> 0:05:00.960 If you build a lot of gener of aif, by 0:05:00.960 --> 0:05:03.279 the way, other companies are too on top of AWS, 0:05:03.279 --> 0:05:06.360 which is the leading technology infrastructure platform. So if there 0:05:06.360 --> 0:05:08.599 are a lot of gener of AI apps being built 0:05:08.640 --> 0:05:10.599 on top of you, you can't help but get a 0:05:10.600 --> 0:05:13.520 lot of feedback from people on how they want those 0:05:13.560 --> 0:05:16.440 building blocks that create gener AI to be better. And 0:05:16.480 --> 0:05:19.360 if you're willing to invest in those building blocks, which 0:05:19.400 --> 0:05:21.440 we are, as you know, with our own chips with 0:05:21.520 --> 0:05:24.599 Tradium and with our own frontier model with Amazon Nova 0:05:24.680 --> 0:05:27.840 and model building services and stage Maker AI and betterck. 0:05:27.839 --> 0:05:30.640 If you're willing to invest in those building blocks and 0:05:30.640 --> 0:05:32.880 you're getting a lot of feedback, they get better much 0:05:32.920 --> 0:05:35.240 more quickly, which can't help them make it easier and 0:05:35.279 --> 0:05:38.200 quicker for people to build gender AI applications, which means 0:05:38.200 --> 0:05:40.360 you get more running on the platform. So that flywheel 0:05:40.880 --> 0:05:42.880 is very unique for Amazon. 0:05:43.320 --> 0:05:46.599 The line share, though, you did say basically one hundred 0:05:46.600 --> 0:05:49.960 billion dollar run rate for CAPEX expenditure. Can you give 0:05:50.040 --> 0:05:51.919 us even like a percentage ratedown of how much that 0:05:51.960 --> 0:05:54.480 goes to distribution logistics and how much goes to AI 0:05:54.760 --> 0:05:55.479 line shares? 0:05:55.760 --> 0:05:57.800 You know, you can tell you most of it. You 0:05:57.800 --> 0:05:59.800 know most of it. You know the line shares. 0:05:59.520 --> 0:06:02.000 More than fifty Yes, is it more than eighty percent? 0:06:03.720 --> 0:06:05.840 We're playing the warmer and colder game? Yes, exactly. 0:06:06.160 --> 0:06:09.080 I love a game taught us about that capacity that 0:06:09.120 --> 0:06:12.920 you talked about. Yeah, AWS could grow even faster if 0:06:12.960 --> 0:06:15.440 you had all the chips that you needed, all the 0:06:15.440 --> 0:06:18.120 power you needed, the motherboards you needed. How much faster 0:06:18.160 --> 0:06:19.120 could a WS grow? 0:06:20.000 --> 0:06:22.840 I could. It's hard to put an exact percentage, but 0:06:23.520 --> 0:06:26.359 I do think it could be growing faster. I'm confident 0:06:26.400 --> 0:06:30.200 could be growing faster. And you know there's there is 0:06:30.440 --> 0:06:32.599 you know, for a long time, there still aren't as 0:06:32.640 --> 0:06:35.839 many chips as we all want we have. We're fortunate 0:06:35.880 --> 0:06:38.560 in that we're very big partners with Video. But then 0:06:38.560 --> 0:06:41.680 we also have our own custom AI silicon in tradium too, 0:06:41.680 --> 0:06:44.320 which we just released in Reinvent, which is thirty to 0:06:44.320 --> 0:06:47.719 forty percent more price performance than the GPU powered instances, 0:06:47.720 --> 0:06:49.600 which is a big deal at scale if you're doing 0:06:49.640 --> 0:06:52.040 gender of AI on the inference side. So we have 0:06:52.200 --> 0:06:54.560 maybe more chips than some others might have access to, 0:06:54.640 --> 0:06:56.400 but we still don't have enough. And then there's just 0:06:56.560 --> 0:06:59.640 is not enough power in the world right now, and 0:07:00.320 --> 0:07:02.560 we're all working really hard on that. I expect that 0:07:02.680 --> 0:07:05.680 to relieve some the second half of this year, but 0:07:06.400 --> 0:07:08.400 right now, and you know, the world can change, but 0:07:08.560 --> 0:07:11.040 right now we have just insatiable demand. 0:07:12.240 --> 0:07:15.280 We had amazing demand coming from Jensen Wang who had 0:07:15.320 --> 0:07:19.000 his numbers out yesterday and Nvidio. Was there a limitation 0:07:19.120 --> 0:07:21.280 on the in video chips in particular that pulled back 0:07:21.280 --> 0:07:23.160 capacity And at what point do you think can depend 0:07:23.160 --> 0:07:26.520 even more on your own in house built chips offset 0:07:26.560 --> 0:07:26.960 any of that? 0:07:27.400 --> 0:07:29.960 Well, you know, I would say, I mean there's a 0:07:29.960 --> 0:07:32.400 lot of demand for gener Ai right now. People are 0:07:32.520 --> 0:07:35.760 very excited about it. I think that all the different 0:07:35.760 --> 0:07:39.680 providers of chips have been constrained to some extent. I 0:07:39.720 --> 0:07:43.200 think some of the some of the new generations maybe 0:07:43.280 --> 0:07:45.560 have gone through different evolutions and when they're going to 0:07:45.600 --> 0:07:48.440 be released maybe a little later than people thought. There 0:07:48.480 --> 0:07:51.240 are some components like motherboards and things like that that 0:07:51.600 --> 0:07:55.119 we all use that particular ones that are in shorter 0:07:55.200 --> 0:07:58.040 supply than others. So you know, I do think there's 0:07:58.040 --> 0:08:00.920 some there are some supply chainges choose which I expect 0:08:00.960 --> 0:08:04.080 to get better. I do you know people are very 0:08:04.080 --> 0:08:06.920 excited about trainingum Too to your question and whether we 0:08:06.920 --> 0:08:09.880 could see more demand there, and we have gone back 0:08:10.400 --> 0:08:13.240 at least a couple occasions to make more trainingum Too 0:08:13.240 --> 0:08:15.440 than we'd intended because we have so much demand. So 0:08:15.920 --> 0:08:18.560 I expect the will of customers for as long as 0:08:18.600 --> 0:08:21.920 I can foresee wanting to run compute on instances that 0:08:21.960 --> 0:08:24.160 have in vidio chips. But I think a lot of 0:08:24.160 --> 0:08:27.240 the demand will also be served by training the. 0:08:27.200 --> 0:08:32.320 Customers that you couldn't serve because of the limited capacity. 0:08:32.920 --> 0:08:34.160 Is it that you just have to put it back 0:08:34.160 --> 0:08:36.319 from everyone a little bit more generally or who lost 0:08:36.320 --> 0:08:36.679 out here? 0:08:36.760 --> 0:08:42.400 Do you think it's always a combination. I mean, there's 0:08:42.440 --> 0:08:46.280 for people that just have a very small amount of 0:08:46.400 --> 0:08:49.160 accelerators that they need. They don't usually have a problem. 0:08:49.200 --> 0:08:51.719 We have something called capacity blocks. It's kind of like 0:08:51.760 --> 0:08:57.760 an on demand way to use accelerators and and generate 0:08:57.880 --> 0:09:00.480 e chips and that continues to go. It's really the 0:09:00.480 --> 0:09:04.800 folks who have built, you know, have an idea for 0:09:04.840 --> 0:09:08.079 a new application, but they need a lot of chips. 0:09:08.160 --> 0:09:10.840 Where if we don't have the capacity, we have to 0:09:11.280 --> 0:09:12.800 you know, we have to give them what we can 0:09:12.840 --> 0:09:14.360 give them and give it to them as fast as 0:09:14.360 --> 0:09:16.880 we can and push our partners to get it in sooner. 0:09:16.960 --> 0:09:19.720 And they can't get their initiatives done as quickly as 0:09:19.720 --> 0:09:21.679 they want to if the capacity isn't there. 0:09:21.760 --> 0:09:23.800 Andthropics had all the capacity it needs, you've got on 0:09:23.800 --> 0:09:24.640 a close relationship. 0:09:24.679 --> 0:09:27.880 Yeah, we have a very close partnership with Anthropic. We 0:09:27.920 --> 0:09:30.440 have this project called Project right near with them. They're 0:09:30.440 --> 0:09:33.760 building their next model, their next version of their frontier 0:09:33.800 --> 0:09:37.000 model on top of trainum to and our custom AI silicon. 0:09:37.360 --> 0:09:39.959 They're going to use over four hundred thousand trainum two 0:09:40.040 --> 0:09:46.160 chips and so yes, they have capacity. They're they're ramping 0:09:46.240 --> 0:09:48.520 up and we're excited about that partnership and what they're building. 0:09:48.840 --> 0:09:51.960 You mentioned the power side. How much is that something 0:09:51.960 --> 0:09:53.920 you're talking to the administration about. How much is the 0:09:53.960 --> 0:09:56.680 administration supportive of the buildout that you need to do. 0:09:57.520 --> 0:10:01.560 We have been talking to administration, you know, multiple administrations 0:10:01.600 --> 0:10:03.680 in this country over the years, as well as in 0:10:03.760 --> 0:10:06.720 other countries as well. And I think the power shortage 0:10:06.800 --> 0:10:10.000 really snuck up on people, you know, really right after 0:10:10.040 --> 0:10:13.800 the pandemic. And I would say that the current administration 0:10:13.880 --> 0:10:16.839 is very receptive to it. They understand the constraints it's 0:10:16.880 --> 0:10:19.440 having on the economy right now and are convicted about 0:10:19.480 --> 0:10:19.920 solving it. 0:10:20.600 --> 0:10:25.120 What about the restrictions around chips? And what's interesting is 0:10:25.160 --> 0:10:28.520 Microsoft just called on the administration today to say, this 0:10:28.640 --> 0:10:33.559 limitation on chip access for some of our close allies 0:10:33.840 --> 0:10:36.200 around the world is going to limit our global business. 0:10:36.240 --> 0:10:37.400 You are a global business. 0:10:37.480 --> 0:10:38.559 Is that something you're worried about? 0:10:39.040 --> 0:10:42.400 Hey, we are? You know. I think I think that 0:10:43.720 --> 0:10:46.800 you're really talking about that AI diffusion act, I think, 0:10:46.960 --> 0:10:50.560 and I'm going to be curious to see where that goes. 0:10:50.600 --> 0:10:53.800 I mean, it was it was enacted pretty quickly at 0:10:53.800 --> 0:10:56.760 the very end of the last administration. I don't know 0:10:56.800 --> 0:10:59.480 how this administration feels about it, but I would say 0:10:59.480 --> 0:11:03.080 that we share the concern that it has limitations on 0:11:03.160 --> 0:11:07.120 certain countries who are natural allies of the US who 0:11:07.320 --> 0:11:09.720 just to be able to do their business and those 0:11:09.760 --> 0:11:11.439 companies to be able to get done what they want 0:11:11.480 --> 0:11:14.120 to get done on top of these technology infrastructure platforms 0:11:14.160 --> 0:11:16.880 like AWS, they're going to need more chips, and so 0:11:17.200 --> 0:11:19.360 I think if we don't do it, we're going to 0:11:19.400 --> 0:11:24.440 basically give up that business and those relationships to other 0:11:24.480 --> 0:11:27.240 countries who can provide those chips, and I think we're 0:11:27.240 --> 0:11:28.640 better off being partners with them. 0:11:28.720 --> 0:11:30.000 Is it a risk to AWS. 0:11:30.360 --> 0:11:33.400 It's not so much risk. I mean, I mean it 0:11:33.440 --> 0:11:37.559 had it, you know, in the scheme of things. It's 0:11:37.600 --> 0:11:41.079 not a big swinger. But I also think that there 0:11:41.080 --> 0:11:43.880 are so many countries who are in the early stages 0:11:43.880 --> 0:11:48.280 of their economic development who both really need access to 0:11:48.360 --> 0:11:51.600 the most cutting edge sophisticated technology to build the right 0:11:51.600 --> 0:11:55.600 customer experiences, and that could be big geographic markets for 0:11:55.800 --> 0:11:59.000 companies like ours and lots of other technology companies, where 0:11:59.000 --> 0:12:00.640 I think it would be a shame to limit them 0:12:00.640 --> 0:12:01.720 and to limit the companies. 0:12:02.760 --> 0:12:05.440 The AI diffusion rules that was brought in very swiftly 0:12:05.480 --> 0:12:07.560 by the Biden administration all of this is in the 0:12:07.559 --> 0:12:11.040 context of US versus China and not wanting to get 0:12:11.080 --> 0:12:14.600 the most sophisticated equipment and chip and technology into China. 0:12:15.720 --> 0:12:19.440 What rate for us for a moment the administration and 0:12:19.480 --> 0:12:22.600 whether it's been positive or negative for your business when 0:12:22.600 --> 0:12:25.480 it comes to China Dominimus, for example, the fact that 0:12:26.280 --> 0:12:28.480 Chinese competitors, if we call them that, Shean and the 0:12:28.640 --> 0:12:30.240 like it can have to pay more to get goods 0:12:30.280 --> 0:12:32.319 into the country. Does that help or hinder you? 0:12:33.320 --> 0:12:38.440 Well, you know, I would say, you know, on Deminimus specifically, 0:12:38.800 --> 0:12:41.880 you know, we have a certain number of items that 0:12:41.880 --> 0:12:44.360 are shipped in that way as well, for things like Hall, 0:12:45.160 --> 0:12:48.280 which is our new low price offering. We maybe have 0:12:48.400 --> 0:12:50.360 less of it than some other companies like the ones 0:12:50.400 --> 0:12:54.600 that you mentioned, but I think it's early in this administration. 0:12:54.679 --> 0:12:58.120 But what I would say is that it is encouraging 0:12:58.200 --> 0:13:00.920 to us that we have an administry that wants to 0:13:00.960 --> 0:13:03.640 hear from business. I would say that, you know, we've 0:13:04.080 --> 0:13:07.320 been a business through six administrations. Every single one of 0:13:07.320 --> 0:13:10.079 them are primary for focuses to take care of customers. 0:13:10.080 --> 0:13:12.160 But we try to build a productive relationship with the 0:13:12.160 --> 0:13:14.600 administration because we want to help the country, and I 0:13:14.600 --> 0:13:17.360 would say that some administrations are more receptive to it 0:13:17.360 --> 0:13:22.280 than others. But this administration cares about what business thinks. 0:13:22.360 --> 0:13:28.120 And I've always been surprised that it isn't obvious that 0:13:28.200 --> 0:13:31.280 the best economic results for a country are going to 0:13:31.280 --> 0:13:34.000 be when the public and the private sector collaborate. And 0:13:34.160 --> 0:13:38.160 you know, I don't expect the government to koutout to 0:13:38.240 --> 0:13:40.960 what companies want, but they should get their feedback and 0:13:41.000 --> 0:13:44.280 their input because they're going to make policies better together 0:13:44.360 --> 0:13:47.080 if they collaborate. And I'm encouraged early on that this 0:13:47.120 --> 0:13:48.960 administration wants to talk to businesses. 0:13:49.080 --> 0:13:50.600 Are you taking calls possily? 0:13:50.920 --> 0:13:55.440 You know, I take calls. We talked to you know, 0:13:55.520 --> 0:13:57.679 the same thing with all the administrations. We talk to 0:13:57.720 --> 0:14:00.200 people in the administration. We share what's working for us, 0:14:00.240 --> 0:14:03.280 what's not working for us, concerns that we have. As 0:14:03.280 --> 0:14:06.400 I said, some administrations care more about our feedback than others. 0:14:06.520 --> 0:14:07.600 Has Trump can about yours? 0:14:07.640 --> 0:14:10.600 Have you spoken to I have spoken to the President. Look, 0:14:10.800 --> 0:14:13.360 as I said, this administration has been pretty busy the 0:14:13.400 --> 0:14:17.160 first month, and but I am encouraged that they are 0:14:17.240 --> 0:14:22.080 having conversations with businesses and they do care about our 0:14:22.080 --> 0:14:24.920 feedback and we'll see what happens. But I you know, 0:14:24.960 --> 0:14:26.840 it starts with a dialogue. You have to have a 0:14:26.880 --> 0:14:29.200 dialogue to have any kind of relationship, and. 0:14:29.160 --> 0:14:32.200 They care about AI infrastructure. Just take stargate. Are we 0:14:32.200 --> 0:14:35.120 going to hear more from you on how much you're 0:14:35.120 --> 0:14:37.520 investing here in the United States on the AI infrastructure 0:14:37.560 --> 0:14:38.280 build out as well? 0:14:38.520 --> 0:14:41.560 Well, you know, as we talked about earlier, we said 0:14:41.600 --> 0:14:45.680 it was you know, directionally right in terms of the 0:14:45.720 --> 0:14:49.120 run rate on our capex. But you know, we're spending 0:14:49.160 --> 0:14:52.240 a lot on AI infrastructure, and the lion's share of 0:14:52.280 --> 0:14:54.280 it is not just on AI but also in this 0:14:54.360 --> 0:14:57.280 country in the US. We spend elsewhere because we have 0:14:57.280 --> 0:14:59.960 a global business. We have customers everywhere. We have customers, 0:15:00.120 --> 0:15:03.760 you know, a couple hundred countries. But we have a 0:15:03.800 --> 0:15:08.800 pretty substantial investment that I don't expect to attenuate soon. 0:15:09.080 --> 0:15:13.360 Only we could get a number. I'm interested in something 0:15:13.360 --> 0:15:15.920 that perhaps is going to feel a more sensitive topic, 0:15:16.520 --> 0:15:19.920 and it comes around perhaps some words that were missing 0:15:20.000 --> 0:15:23.120 in your annual report this year, which we're diversity and inclusion. 0:15:23.160 --> 0:15:25.200 I put this in the context of the administration as 0:15:25.240 --> 0:15:28.160 it stands, because I know that Amazon strives to be 0:15:28.240 --> 0:15:32.040 the Earth's best employer. And I'm just wondering how your 0:15:32.080 --> 0:15:36.480 employees react to perhaps the lack of certain words now involved, 0:15:36.880 --> 0:15:39.480 and whether or not programs might be forced to change 0:15:39.560 --> 0:15:39.800 or not. 0:15:40.200 --> 0:15:42.320 Yeah, well, what I would tell you just at a 0:15:42.400 --> 0:15:46.160 high level, if you serve as many customers as we do, 0:15:46.280 --> 0:15:50.440 in as many diverse groups as we do, and we 0:15:50.520 --> 0:15:53.640 intend to moving forward, you have to have a diverse 0:15:53.720 --> 0:15:57.640 team to be able to build products that work for everybody. 0:15:57.720 --> 0:16:00.640 And that has always been our intention to continues to 0:16:00.640 --> 0:16:06.080 be our intention. I think that you know, there were 0:16:06.160 --> 0:16:09.440 so many programs that we launched and other companies launched 0:16:10.440 --> 0:16:13.520 in the pandemic, and as you probably have seen over 0:16:13.560 --> 0:16:17.240 the last three years, we've gone through very thoroughly every 0:16:17.360 --> 0:16:20.680 single one of our business areas, and the programs that 0:16:20.760 --> 0:16:23.720 we had conviction about we doubled down on, and the 0:16:23.760 --> 0:16:26.760 programs that we don't have conviction about, we streamlined and 0:16:26.800 --> 0:16:28.720 we stopped doing. And so we did the same thing 0:16:28.760 --> 0:16:32.520 and looking at all of our programs on diversity, and 0:16:32.920 --> 0:16:34.880 you know, we have some programs that we have doubled 0:16:34.880 --> 0:16:37.560 down on. A good example is our Career Choice inside 0:16:37.560 --> 0:16:41.400 our Fulfillment Network, our Fulfillment Center. Teammates are able to 0:16:41.400 --> 0:16:44.840 get an advanced education for free on us to advance 0:16:44.920 --> 0:16:48.360 their career and their own development. And that has been very, 0:16:48.480 --> 0:16:50.840 very successful and very meaningful, and so we doubled down 0:16:50.880 --> 0:16:53.480 our program like that. There are other programs that really 0:16:53.520 --> 0:16:56.520 just haven't been that successful and haven't moved the needle much, 0:16:56.560 --> 0:16:59.720 and those we just moved away from. But we have 0:17:00.080 --> 0:17:02.080 first group. We're trying to continue to build out a 0:17:02.120 --> 0:17:03.240 diverse group and that won't. 0:17:03.120 --> 0:17:07.040 Change and you can't. You can still use that word ultimately, 0:17:07.040 --> 0:17:08.159 you're not having to reframe it. 0:17:08.840 --> 0:17:11.920 That is. I mean, look, you can call lots of people, 0:17:11.920 --> 0:17:15.240 call lots of different things, but we have a giant 0:17:15.280 --> 0:17:18.399 customer base of every imaginable group of people where we 0:17:18.440 --> 0:17:20.440 want builders who can build for them. 0:17:20.760 --> 0:17:24.640 Culture is key. Yeah, and you just talked about how 0:17:24.680 --> 0:17:27.600 you're doing more is less. It's always a focus on frugality. 0:17:27.720 --> 0:17:30.240 That's in your very principles. That's something you've been doing 0:17:30.280 --> 0:17:32.720 at the employee base as well. You wrote out made 0:17:32.720 --> 0:17:35.200 it very clear that they were going to be reducing layers. 0:17:35.600 --> 0:17:38.280 How is that going, Yeah, it's gone. Well. You know, 0:17:38.320 --> 0:17:42.120 I look, if you have a company where the culture 0:17:42.280 --> 0:17:45.360 is an important ingredient in your success, which has absolutely 0:17:45.400 --> 0:17:49.560 been true for Amazon. It's not your birth rate to 0:17:49.640 --> 0:17:51.840 keep having a strong culture, you know, especially as you 0:17:51.840 --> 0:17:54.080 grow the number of people, the number of businesses you're in, 0:17:54.119 --> 0:17:56.840 the geographies that you're in, and so you have to 0:17:56.880 --> 0:17:59.359 work at it all the time. And for us, you know, 0:17:59.359 --> 0:18:01.560 there are two areas when we looked at it as 0:18:01.560 --> 0:18:04.760 a leadership team last year that we wanted to strengthen. 0:18:04.800 --> 0:18:11.000 One was we have always hired really strong owners, smart ambitious, 0:18:11.080 --> 0:18:14.480 strong owners who get to own the allian's share of 0:18:16.119 --> 0:18:19.000 In this case, I would say, you know ninety plus 0:18:19.000 --> 0:18:24.840 percent of the decisions, and they you know, as you 0:18:24.960 --> 0:18:28.080 add a lot of people, you end up with a 0:18:28.119 --> 0:18:31.320 lot of middle managers, and those middle managers, all well intended, 0:18:31.400 --> 0:18:33.600 want to put their fingerprint on everything. So you end 0:18:33.720 --> 0:18:36.119 up with these people being in the pre meeting for 0:18:36.160 --> 0:18:38.240 the pre meeting, for the pre meeting for the decision meeting, 0:18:38.560 --> 0:18:41.359 and not always making recommendations and owning things the way 0:18:41.400 --> 0:18:43.720 we want that type of ownership. So we took a 0:18:43.760 --> 0:18:48.439 goal collectively to increase the ratio of individual contributors to 0:18:48.480 --> 0:18:50.840 managers by over fifteen percent as a company by the 0:18:50.880 --> 0:18:53.320 end of this quarter, we've made very good progress in 0:18:53.359 --> 0:18:55.679 that we've beat that already and it's going to allow 0:18:55.840 --> 0:18:57.800 us for the people that are doing the work, they're 0:18:57.800 --> 0:18:59.320 going to have more ownership and they're going to be 0:18:59.320 --> 0:19:01.480 able to move more QUI And then I think the 0:19:01.520 --> 0:19:04.600 other thing we saw was that if you're a culture 0:19:04.600 --> 0:19:07.360 that invents a lot and collaborates a lot like we do, 0:19:08.640 --> 0:19:11.359 if you don't have people in the office together doing 0:19:11.400 --> 0:19:13.879 that invention, it's just meaningfully worse. And you know, you 0:19:14.320 --> 0:19:16.880 don't invent the same way, you don't collaborate the same way, 0:19:16.920 --> 0:19:18.520 you don't connect with each other the same way, you 0:19:18.520 --> 0:19:21.560 don't learn the culture the same way. And so having 0:19:21.560 --> 0:19:24.680 people back in the office together more frequently, we felt 0:19:24.720 --> 0:19:26.159 very strong. We bet it's going to be better for 0:19:26.200 --> 0:19:27.159 customers in the business. 0:19:27.480 --> 0:19:31.000 Let's talk about that invention and that collaboration. We've got 0:19:31.000 --> 0:19:33.160 a new quantum chip us A lot love the name. 0:19:34.119 --> 0:19:36.000 What does that mean for you to have this more 0:19:36.040 --> 0:19:39.240 efficient chip at this exact moment. Will we start seeing 0:19:39.280 --> 0:19:40.800 it be practically useful soon? 0:19:41.280 --> 0:19:45.400 Welltum compute. Quantum computing is very high potential. It has 0:19:45.480 --> 0:19:50.199 the chance to solve some very computationally intense problems, and 0:19:50.720 --> 0:19:54.800 I still think it's realistically a few years away from 0:19:54.880 --> 0:19:56.919 having a real shot at solving those problems. But you 0:19:56.920 --> 0:20:00.520 have to solve a bunch of these challenges that relate 0:20:00.520 --> 0:20:03.280 to quantum computing along the way, and one of them 0:20:03.320 --> 0:20:06.320 really is around error correction on the CUBIS, and that's 0:20:06.320 --> 0:20:09.320