EP336 - Amazon Q4 Earnings
EP336 - Amazon Q4 Earnings Chapters Welcome to the Jason and Scott Show NRF Highlights and Personal Updates Upcoming Conferences and Events Anthropic's Super Bowl Ads Claude Bot and Agentic Technology Amazon Earnings Overview Meta's...
EP336 - Amazon Q4 Earnings
Chapters
Welcome to a new installment of the Jason and Scot Show! In this special episode recorded on February 6th, 2026, hosts Jason Goldberg and Scot Wingo reconnect for an enlightening discussion following their experiences at the NRF and share key industry insights. This episode is packed with updates on Amazon's recent earnings call, the advertising landscape, and the emerging role of agentic commerce in the retail world.
Join Jason and Scot as they kick off the episode by reflecting on their time at NRF, where Jason delivered an impactful keynote that had listeners buzzing. In a light-hearted banter, they dive into personal stories, including the often-asked question about Jason's son’s sunburn during their recent Hawaiian vacation. From there, the conversation takes a deep dive into the evolving nature of retail, where both hosts express their excitement over how businesses are embracing the latest innovations in technology and AI.
The duo then brings their focus to Amazon's latest earnings report, which revealed a robust growth in retail despite an ominous forecast for capital expenditures. While they discuss the implications of Amazon's projected $200 billion in investments—including expansion into AI infrastructure, fulfillment centers, and essential goods—Jason and Scot analyze whether Wall Street's reaction of an 8% drop was justifiable considering Amazon’s overall performance in retail and advertising.
Listeners will gain valuable insights on how Amazon's grocery strategies and its dominance in everyday essentials have created a competitive landscape that mirrors historic retail expansions, like Walmart's grocery initiative. The hosts also emphasize the relevance of the Amazon Marketplace, where third-party sellers are taking up 62% of total units sold, highlighting the synergistic effect between advertising and retail performance.
The podcast also delves into the broader context of the e-commerce space, touching on AWS growth dynamics, the challenges posed by capacity constraints, and how they contribute to the narratives surrounding AI integration. With mentions of Anthropic's recent ad campaigns and competitor movements, listeners will appreciate the comprehensive look at the competitive dynamics shaping the industry.
Throughout the episode, Jason and Scot maintain an engaging dialogue on the potential of agentic commerce, discussing the opportunities for businesses to adapt and innovate in a rapidly changing environment. Their analysis of the demand for AI-driven tools and the importance of maintaining consumer experiences culminates in a call to reflect on the value of assortment and the implications of shifting consumer behaviors.
In conclusion, Episode 336 serves as both a detailed recap of Amazon’s recent performance and a broader commentary on the evolution within the e-commerce sector. With thought-provoking insights and a playful rapport, Jason and Scot ensure that listeners walk away with both the latest news and a deeper understanding of where the retail industry is headed.
Episode 336 of the Jason & Scot Show was recorded on Friday, February 6th, 2026
Join your hosts Jason "Retailgeek" Goldberg, Chief Commerce Strategy Officer at Publicis, and Scot Wingo, CEO of ReFiBuy and Co-Founder of ChannelAdvisor as they discuss the latest news and trends in the world of e-commerce and digital shopper marketing.
0:23 Welcome to the Jason and Scott Show. This is episode 336 being recorded on Friday, February 6th. That's a lot of sixes. I'm your host, Jason Retailgeek Goldberg. And as usual, I'm here with your co-host, Scot Wingo. Hey, Jason, and welcome back, Jason and Scott Show listeners. Jason, my 2026 is off to an awesome start. I got to hang out with you for the first time in a long time at the National Retailer Federation. We had a great time. We did. That is how NRF in the beginning of the year are meant to be, is you and I getting to spend some time together. I didn't realize how much I was missing the couple of years that you were too big and fancy for us.
1:04 Yeah. And if you didn't catch Jason's keynote, which we put across all of our media empire, it was an audio here on the Jason Scott show. It's on the YouTube and audio channel for Retail Gentic. It has been a very popular keynote. There was no one in the room, But man, people liked it in the replay. Yeah. But remember, Scott, we pretended in the recording like it was a full room. So don't, don't ruin the illusion. Yeah, I did this. I cut in, we spliced on the video, we spliced in some random shots of a Super Bowl stance just to kind of, you know, give you. Yeah. If you're wondering why everyone's holding a beer and a hot dog for my keynote. They're wearing like 49ers jerseys and stuff. Exactly.
1:47 Not 49ers, but okay. Yeah. Next. I was thinking it would be a previous year. It's not the one that's coming. Fair. The but in all seriousness it was a great keynote it was i haven't seen you do your thing in a while and you've you've gotten better you're always good but it was really good. Oh that's super nice of you to say thank you it's it's always fun talking about these topics obviously like a lot of the those topics were developed in conversations with you and many of our other smart friends so i'm grateful for the network of smart people i have that helped me put it together the. Number one question we've been asked in the YouTube comments. Everyone would like to know if your son got a sunburner or not in your trip to Hawaii. That's like the number one question out of the keynote. I thought it was going to be something kind of, you know, about ACP or UCP or one of these authentic commerce kind of things.
2:34 But really, everyone is very concerned about your son's skin health. The personal stories. And no, he did not. As you know, my wife is Lebanese, So we call my son a lebahonky. So he has more natural melatonin protection than I do. I'm the most likely person in our family to get a sunburn. Well, I know this will be in the comments for this episode. So did you get a sunburn? Yes. Okay, good. You went to all that trouble to use Agenda Commerce to find the perfect sunscreen and you didn't put any on you. Well, yes, that, yes. I did use sunscreen, just not diligently enough. Yeah, I found a... There's a little spot up top that sometimes gets sunburned.
3:17 I'm not really used to putting sunscreen up there. Yes, I grew up. The balding hair area. Yes, yes, I'm, yes. We had more protection at one point than we do now. How about any other trip reports? I'm going to Shop Talk coming up in March, but sadly you won't be there because you've got spring break where you'll be getting yet another sunburn. Yes, yes, I'll be updating all those pictures for next year's keynote as if they would invite me back. Yeah, unfortunately I am missing Shop Talk again, which I'm super sad about. I do have FOMO for that. I've been on the road every week, but honestly, it's all been like at the beginning of the year, every one of our clients like kicks off their year and very frequently they have all hands meetings or, you know, national sales meetings.
4:06 And so I'm kind of going from event to event, which is a grind in some ways, but it's awesome in some ways because they're all now thinking about all the same things that you and I have been thinking about and talking about for a year. And they have a bunch of, you know, new smart people looking at the same problems from different angles. And so it's been... It's exciting to see how every brand is applying all of these potentially disruptive changes to their unique circumstances and coming up with different plans and different strategies and different ways to tell the story. I feel like we were total nerds talking about something that no one else cared about a year ago, and now that topic has suddenly become cool and the cool kids want to talk about it. So that's fun.
4:51 Yeah, I like to reframe it as we're prediction geniuses, especially when we're right. And if we're wrong, we don't really talk about it again. So there you go. Yes. That's a little inside info on how you get to be a good predictor. Hi, everybody. Yes. Says the guy that claims he's too busy to record the episode to see how well, how good his predictions were last year. That was you. You're on a, you're always on a flight. You've spent 11 days on a plane. Back to back. I did the math. Yeah. Yeah. That was impressive. Impressive, man. I'm also going to the Digital Shelf Summit. And so I guess we won't see each other until retail club. So we're both going to be at retail club and that'll be fun. Yeah. But I actually think it's only semi-private, but I think we're doing a Citibank conference together. Are we not?
5:35 Yeah. Yeah, we are. Yeah. Yeah. Oh, are you keeping that secret? No, I just don't think the average person can just go to that one. Okay. You have to be a city client, but I don't know. All you city clients, come on down. I think having a credit card is enough. I'm sure they'll welcome you in. Or a dollar to invest. To be clear, none of our listeners are average people. They're all above average, of course. Yeah, but I certainly am looking forward to Retail Club. That's going to be a fun one this year. Although that's quite a ways down the road. Yeah. And then the other big news I wanted to pick your brain on is Anthropic came out this week with some savage Superbowl ads that basically attack pretty much, uh, attack Chachubuti straight on that they're going to have ads and Claude is not.
6:23 And they, they're, they're probably the most clever creative I've seen where, you know, they, they perfectly capture the, the sycophantic nature of ChatGPT. And then they also inappropriately introduce ads and it creates these really good comedy moments. The acting on it is all great. The writing and acting and everything is really good. Have you seen those guys yet? I have. It's pretty funny. I'm super looking forward to this Super Bowl. In my industry, obviously, the Super Bowl is always a big deal, but there are some years when the stuff going on in the Super Bowl overlaps with your and I's interests more than others.
7:04 And so there's some years where it's super exciting for the advertising industry, but it's kind of a nothing burger for us. There's some years when like a bunch of dot coms like show up in the Super Bowl ads for the first time or, you know, Timu showing up in the Super Bowl is a big deal for the first time. And I have a strong feeling this is going to be one of the years where it's going to be well worth it for all of our listeners to be paying attention to the Super Bowl. I think there's going to be a lot of relevant content. That is super exciting. And this first ad, A, I agree with you. It's awesome. I think it's super well done. There's an interesting thing, conventional wisdom at these top platforms on this, like super expensive advertising platforms, it generally isn't advisable to go negative.
7:52 Like the general school of thought is you don't call out competitors and you don't focus on competitors' disadvantages. And there's a fair amount of history behind those kind of ads underperforming. And so it's like, obviously, Anthropic is a first time advertiser. They put together like, what I think is a really awesome negative ad. And we've already seen evidence that it got, it's already landed, it's already gone under ChatGPT's skin. So they're living rent free inside Sam Altman's head right now. So that's going to be super interesting. But I'm also going to be interested to see all these other guys with these crazy evaluations, what they do at the Super Bowl.
8:34 And, you know, what and if will we see from Amazon and what and if will we see from Microsoft? And, you know, it's going to be fun to watch. How about have you gotten caught up in the Claude bot, Molten bot, Open bot crazy, where you can have a Mac mini that basically sits there and you text it and it does your job for you. You kick back on a beach. Yes. I feel like that's a bigger deal for people that had harder jobs than me. Like technology has been able to do my job better than me for a long time. So it's, it's slightly less of a big deal, but yeah, I'm pretty, I mean, bigger picture, maybe we'll talk about this, but like it's all just evolving so fast. the pace of change.
9:16 Like I remember when like we were all chat GBT all the time and it seemed like everyone else was behind. And then like, like Google moved up the, to everyone. And now like, you know, everyone in my circles is, is mostly talking about Anthropic this week. And so like, if anything, I feel like I'm behind in, I'm trying to dog food all these things, but there's like so many tools evolving so fast, it's hard to use them all simultaneously. So I feel like I need to take some more quality time with Anthropic. Are you, have you, have you been using the agents a lot? Are you? I have. This one's pretty neat. It's this one guy open sourced it and it has no rails. So it's a little scary, but there's ways to put it in kind of a contained environment. The breakthrough is you can text with it and it's super multimodal.
9:59 So you could have it call you. You could, you know, It basically is unbound because a guy, just a random dude, just open sourced it and put it out there to the world. So it doesn't have any of the normal constraints. So it is kind of interesting to see what does that look like and what all can it do. And it's kind of scary. There's a lot of agentic shopping things. People are using it to book trips and do all kinds of complicated stuff. I'm going to eventually do a blog post on it. But yeah, it's pretty. This is definitely the year of agentic generally, like these agents doing things. And this is where they're having a lot of breakthroughs in the coding systems. it's it's you know you're going to hear things like swarms of agents and it adds an element of parallelism to this where once you get to a task where you're like well i could do this you know over 80 different i know you do a lot of data manipulation well you could go and say well spin up an agent for every one of the last quarters of this data i want to look at and have them do it in parallel and then bring it back to data and fold it back together so they can do some really interesting things now and they've kind of cracked they're getting close cracking the orchestration on the front and back end of that so it's it's, It's going to be the next kind of leg up in effectiveness of these things.
11:08 Yeah. I don't know about you. Like, it feels like the theme in my life, like they keep improving and then they talk about all these cool use cases they do now. And it's like the examples of the things that the robots are taking over from the human are all the fun things that I enjoy doing. And it still seems like the case that the robots aren't perfect at doing all the stuff I really don't want to do. So, like, I have a test suite for these do-it-for-me agents, and it's canceling my Comcast account and switching my cell phone providers. You know what I mean? It's those calls you dread. It's like, but I will tell you, the one that they seem to have won is, like, providing tech support for my mother-in-law on Thanksgiving.
11:50 Good. And so, those are cool. You are right. Versus you. Yeah. No, exactly. And oh, my God. And for my 10-year-old, like legitimately, there are all these, you know, parental permissions. How can my son, you know, do this with his friends in Fortnite if I don't want to let him do this kind of thing that are way harder to figure out in the world than they should be? And, you know, typically are things that I could figure out, but take me 30 or 40 minutes of research to do that now I'm getting... Perfect guided responses from the bots instantly. So I do feel like there are these use cases where it's getting much, much better, quicker.
12:33 You hit one that's a sore spot for me is I do have to do a bunch of data extraction and manipulation every month. And February 10th, we're recording this on the 6th, the 10th, we get December data from the US Department of Commerce. So that's a, i literally like block a day off because i'll have all of 2025 data and there are all these interesting questions i have but there's a lot of grunt work to get to where i want to be with that and i keep testing all of the latest bots and they're they're they're they get closer and closer but they're still not reliable so so. I know it's a matter of time yeah i have the unlock for you and listeners um yeah hit me in in the anthropic ecosystem under the clod label they now have a excel and they just came out yesterday with a powerpoint add-on the excel add-on is so good i i got an email from anthropic and i was like oh no i i've used all my tokens or something they're like congratulations you're one of the top users of clod with excel we'd like to invite you to our customer something something feedback board so i've been just chewing that up because i i love modeling and stuff i hate the drudgery of you know like waterfalls cut copy pasting for hours and all that kind of stuff so you can just tell it i one shot at a whole new financial model for uh my new company and it, It was like when ChannelVisor was at a 50 million kind of run rate type model, it just like one shot at that.
14:04 And there was no errors. There's like a couple of formatting things that, you know, like percentages were not in percent. But it was just amazing. So I think you'll find it to be, it will be the one to unlock for you. I agree. And I have played with that. And fun fact, everybody's pretty excited about the Excel integration. And what's interesting is, of course, Microsoft owns Excel. And has deep integration with Copilot, but the Anthropic integration with Excel is much better than the Copilot integration. Yeah, Copilot's terrible in PowerPoint and Excel. Yeah, I would argue it's making better progress in other places, but shockingly, not that strong in their suite. So I have a very complicated spreadsheet with a lot of this data in it that is all, because I built it improperly in the beginning, year-specific.
14:53 So you know what I mean? It's got a five-year history in all these tables, but when we roll over to a new year, there's a bunch of work I have to do across 40 different spreadsheets. That's usually a day of my life that I dread. I've already done a test, and it already seems like Anthropic is going to easily adapt my whole workbook to the next year's data. That's a huge win. And what's less awesome is like, I try to track the 50 public retailers and their quarterly earnings growth, and they don't all report that uniformly. So there's a lot of logic behind the scenes about how to report all that.
15:36 And let me just say, like, the problem now is you assign any of these agents to go do that research for you, and it looks super credible, and it's 85% correct. But there's still 15% of mistakes that are harder to find, but significant. Yeah. Yeah, they'll get there. Yeah, yeah, yeah. I'm sure they will. I make mistakes too, by the way, so it doesn't have to be perfect. It just has to be better than me. I think Claude would, you could say start a new year and use the last year. I think it'll one shot it for you. Yeah, I'm going to try for sure. That's my prediction. Let us know. Next pod, you'll let us know how it goes. Okay. And I can't recommend it strong enough to everyone listening. It's really good.
16:18 I just got the PowerPoint one last night and I had to do this frequently asked question presentation for potential investors and it created 38 slides for me and it was perfect. That was great. I was like, oh my God. My life is so changed right now. This is classic Jason and Scott show material. We're doing the Amazon earnings podcast and we're 20 minutes in. We haven't mentioned Amazon. We're going to keep it short this time. We're going to keep it under three hours. But that's a good setup, Jason. Yeah. We're here to talk about Amazon earnings, but I also think it's important to take a step back and kind of look at what is going on and you kind of think about our perspective of are we in a bubble?
16:58 Now, the main reason people are asking that is there's all these deals of, you know, NVIDIA investing and then getting money back and all that. That's the one that people focus on. That's not really our wagon, our bailiwick. So, What we're going to do is really look at these through the e-commerce and retail lens, first of all, but also try to frame this in what the companies are saying. We get all the earnings calls and all the Wall Street reports. So we've picked through all that and we found some tidbits that I think help paint the picture. And I like to go through this, long-time listeners will know, I like to go through it in kind of a chronological order. The setup is usually that Amazon is the last one to report, and that's how it played out this time.
17:41 So first we had Microsoft, and they came out January 27th. And the part that's relevant to our discussion is the cloud computing, which at Microsoft is called Azure. That grew 66% year-over-year, which is pretty material acceleration, to $37.5 billion. Now Azure on its own they've only been breaking it out for a while for a long time they just called this all cloud they included Microsoft 365 so for the first time I think this is the third quarter maybe this is the fourth quarter I can't remember when they started breaking out but it hasn't been long but that's but that's pretty good but here's the shocking thing it missed Wall Street's whisper numbers we're at that point now where we have you know the the low and high estimates and if you beat the high or sometimes you beat the midpoint everyone's happy now you've got to beat the high and hit some kind of unspoken whisper number so that's kind of where we are on expectations there the the thing to understand and we'll recap this at the end but each one of these companies has started announcing a backlog because they're trying to get wall street to understand hey the reason we're doing this is we do not have enough capacity and this is what we're is backlogged right now.
18:51 We could recognize this revenue if we had enough servers. And Microsoft's backlog is $625 billion. So that's crazy. And that has grown 100% year over year. So December quarter of... And they have their quarters are off by one. So I won't. Calendar December quarter was 298 last year. And then this year it's 625. So yeah, more than 100% growth. A couple other things I thought were interesting in there. They have 250 customers that are north of a 1 trillion token a year pace. When you're using AI, the kind of currency on AI as tokens. And that's the one currency. The currency underneath that is basically power.
19:36 So we're at this point now where these build outs that were running into physical constraints with the. Power system, the ability to bring enough, get enough land and concrete and all the things. They added a one gigawatt of capacity. Satya Nadella said, agents are the new apps. And then they announced some CapEx that we'll go into in a minute. Then the day, next day, and the stock was off pretty substantially at Microsoft. It was off as much as 10%. I think it's bounced back a little bit. So Wall Street's in this thing where last year they were patient. They said, okay, you guys are investing against this and we want to see return. And this year, everyone's saying, hey, I know you're nervous, but we're going to triple down. How do you feel about that? Wall Street said, not good.
20:23 So the stock market has had, on top of that, all these new anthropic tools have got Wall Street freaked out that software companies are going to go out of business. So it's been a really rough, as we're recording this, basically 10 trading days on Wall Street because of all the AI's disruption, which was the theme of your NRF keynote, I will add. So then the next day, Meta announced the artist formerly known as Facebook and their fourth quarter advertising revenue just absolutely crushed it at 58 billion. That was a material acceleration. And the reason they had such a big acceleration is they're using AI internally and a lot of other, they have this other thing that's not really Gen AI, but it's more of a.
21:10 Some kind of machine learning algorithm it has a name you may know it may have Max in its name I can't remember, But they specifically called out a couple reasons that their ad business is accelerating. They have this AI powered creative tool. So it will create the, you know, for small businesses that don't have a fancy ad agency, like let's say Publicis, you know, they have a little tool that will create their ads. That's just the ads from that reached a $10 billion contribution on revenue. And then they also have a new way of measuring ads that are really unlocking things for advertisers. And that's driven by AI that increased conversions 24%. So I think it's when I when I parse the language there, I think they're getting much better at targeting with AI.
21:56 I think AI is able to do much better at targeting. And then they basically said, hey, here's here's what's going on. We think the future we're going to we're going to really invest heavily in AI and we use all this money we make from ads to fund it. And AI is going to be the big growth driver after ads. Famously, so they bet on this open source system called Llama that they built, an open source model. And it kind of ran out of steam. So famously, Mark Zuckerberg opened up his wallet and has been hiring people for north of, you know, $100 million kind of salaries. And some rumored as high as a billion would call it spending the Zuck bucks. And he has put together kind of a Justice League of America type team of AI superstars. And they're working on a new model.
22:41 This week, it came out in an article in the information that that is codenamed Avocado. And the other rumor is that they're very excited about its initial results from pre-training. So we could have Meta kind of come back into the game here. The thing that got me so excited is during the call, he basically said they're going to come, they're going to jump into a jump to commerce, which is really fascinating because you and I know they've gotten out of, they had all this commerce stuff and they've gotten totally out of it. So to see them kind of waking up and saying, wait a minute, we need to get back into this game is pretty intense. Let me just call out a specific quote. He said, this also has implications for commerce. Our ads today help businesses find just the right, very specific people who are interested in their products.
23:26 New agentic shopping tools will allow people to find just the right, very specific set of products from the businesses in our catalog. We're focused on making these experiences work across both our feeds and across business messaging. So they're basically saying it's going to be on Insta, Blue Facebook, and in WhatsApp as well. Then they proceeded to guide CapEx to 115 to 135, which depending on which part of that range you look is as high as a 90% year of your growth rate in capital expenditures. And Wall Street hated that. So they were off. So despite, you know, beat and raise, beat and raise, beat and raise, the fact that they're spending so much on CapEx, Wall Street was like, no bueno.
24:06 Anything to throw into that one? Yeah, well, I think we're going to see that that's the reoccurring theme of the quarter is great financial performance, but we're going to have to spend like a drunken sailor next year. And in general, Wall Street's not liking that. The meta stuff is super interesting. The meta broadly calls all of these AI-enabled functions in the advertising platform, they call them Advantage Plus, right? So there was an era where you would be a human and you'd say, I want to buy this audience on Meta to advertise my product to. And increasingly, they've said, oh, instead of you specifying the audience, you should let our Advantage Plus robot pick the audience for you.
24:50 And at first, the robots didn't perform very well compared to the best human media buyers. But increasingly, the robot is able to find a much better audience that's much more economical, more efficient to bid on. And so they are making a bunch of progress by saying, hey, just turn your budget over to our black box. Trust that it'll make a bunch of good decisions for you and get you good ad results. And it legitimately is. It's scary because you don't know exactly what it's going to do. But it's really helping their ads outperform and grow the ads business. And you hit one interesting thing. Everyone wants to imagine that all these ads on Meta are coming from huge CPGs and huge advertisers with Super Bowl ads.
25:36 It actually mostly isn't. It actually is mostly small businesses and direct-to-consumer businesses from China and folks that do have a lot less resources to be really sophisticated at managing these advertising campaigns. And so the fact that they have tools that are accessible and performant for these small businesses has really been an unlock for them. So that's, I think, what's going on at Meta. The commerce thing is interesting. I think there's a little bit of room for semantics there. I think, you know, Meta has a product they call Advantage Plus Shopping, which, you know, is all of the buy something called the action focused ads.
26:16 And so if Zuckerberg's talking about that as agentic commerce, that's something different than what I think you and I are interested in, which is like, does he actually get back into converting his product feed into transactions on his platform and who would be the seller of record and all these interesting questions that would come up there. So we'll keep watching that. There was one other tidbit. He, at one point during, during the earnings call, they were talking about the, the hardware products and the AI glasses. And he said, like, very matter of factly, I can't imagine a future when we're not all wearing agents on our face. Like that. It just seems so obvious to me that we're all, and, you know, we're, we're having a big dispute right now about where the glass is going to be for, for these agents. Is it, is it the vertical agents that run on, on traditional retailers websites?
27:07 Is it the horizontal agents that have their own websites and apps? Is it the browsers? Is it the OS? And Mark is like, like I'm a hundred percent convinced we're all going to be wearing an agent, which is to me totally viable and pretty interesting. Yeah. Yeah. We'll see so far. I haven't met an agent I want to wear on my face, but maybe at some point we'll see. Yeah. Agreed. But I, it does seem like it's coming one way or another. We'll see what what Sir Ives comes up with. Yeah, the Ray-Ban things are good, but they feel like they're a couple beats behind at this point. It's everything's going so fast. Then we had February 4th, which was Google, also known as Alphabet.
27:48 So just a couple tidbits from this one. They said Gemini models now process over 10 billion tokens a minute through API, up from 7 billion the prior quarter. They love throwing around these crazy big numbers. Side note, isn't it fun that the currency that was most important to us as 10-year-olds in the arcade is still the most important currency? It is. Yeah, we've been on this token thing for quite a while. Yeah. The Chuck E. Cheese ones were the best. Absolutely. Gemini 3 Pro has processed three times more tokens than the previous Gemini. They had 350 customers that processed more than 100 billion tokens. So everyone's trying to find these kind of standout metrics.
28:29 And they change them every time, which is annoying. For a long time, Google put out this one that was tokens in the quarter, I think it was. And then it slowed down a little bit. So now they've stopped using it. There was another one i don't think i captured it here they did announce that gemini has crossed 750 million active users i forget if that was i think that was monthly i don't think it was weekly, so just a lot of growth on on the consumer facing side they also spent a lot of time talking about you know there's all these people that and i was in this camp i will i'll admit i was wrong on this prediction that it just seemed intuitive to me they wouldn't be able to figure out how to monetize an answer engine as well as a search engine.
29:13 For the last year, they have actually gotten to the point where the monetization is better. So they've actually, it was a little worse and they were actually kind of clear on that. And now they say it's materially better. They were almost like, it's hard to get this across here, but they were almost like a little giddy about it. It seemed like they were getting excited. And I've always felt like there will be a moment in time where AI mode becomes, you know, the default. And I think this will happen to Rufus too. At some point, these companies are going to switch the old mode, which is the core search. And then you get a click on something to get the answer mode. That's going to flip. And I think Google is going to be first. And it'll be interesting because a big moment in time to kind of think about is holiday sales 2026.
29:57 Will we see one of these big companies, either Amazon or Google, make a big move before holiday? to if because the logic is if the economics are actually getting better inside of that ai thing then why have the old thing you know right so it will. Yeah that'll be the moment that. They're really serious about it. Yeah i actually think it's a pretty safe bet by holiday 2026 because it it when we first started talking about this ai overviews were showing up in three percent of google queries um and by some counts it's it's approaching 50 now so it's like we talked about this, but essentially that, you know, they really constrained it when they didn't, when they had the revenue to protect in their legacy business that they hadn't figured out how to replace in the AI overviews.
30:39 But now that it seems like they're, at least for some kinds of queries, confident in how they're going to monetize the queries, it seems like they're turning that dial pretty darn fast. Yeah, it's crazy. So they had all these amazing vanity stats and how awesome that's going, also their cost to serve everything is going down because they have their own TPUs. They always like to kind of throw it in NVIDIA's face that they're not on their hardware. And then they said, here's our CapEx guidance. We are going to double our year over your CapEx. And Wall Street took all that interesting, exciting news and basically sent the stock down, three to five percent yeah so i'm sure that was right it's almost like investors.
31:19 Like share buybacks better than they like capex investments. They do yes they are very short-term thinkers it turns out some of them so we'll see and then there's a couple they did talk a little bit about agentic commerce and i don't know if you get into this but i get i get in conversations. The naysayers have a more nuanced attack on agentic commerce and so it used to be definitions And they're still out there. And now I'm getting people that say, how serious is Facebook, Google, and all these people about this? Like, is this, are they going to just rug pull us on this or is this really serious? So, you know, so basically in the fourth quarter earnings transcript, here's three little snippets. So the chief business officer said, as Sundar mentioned in his candid remarks, we're building an era of agentic commerce and working with our partners to introduce the universal commerce protocol in our consumer products and across the entire web.
32:07 What they mean there is they also did a they're adding Gemini to Chrome and it will go buy things for you and check out through UCP. So UCP is actually kind of. You know, breaking out of the site experience and becoming untethered. So none of this is live. I'm having a little hard time imagining how well that's going to work until I see it. But there, my overall overarching point here is Google is really serious about this stuff. This is not a 20% time thing from one developer. This is from the CEO down. We've received tremendous feedback from the industry. Soon people can use a new checkout experience to buy directly in AI mode in Gemini. for select merchants.
32:49 So this is not launched yet, but they said soon. So I don't think in fourth quarter earnings call on February 4th or whatever it was, you say this if you're not going to have it out by the next one. So my prediction is, I also predict that on the Facebook thing. I think they're going to do some kind of agentic thing between next earnings call and I think UCP will launch before then. I have trouble with UCP because I want to say UPC because I say it so many times. And then sundar said in our areas like he was responding to a question from our friend at morgan stanley uh brian noack so brian asked about agentic commerce and sundar said in areas like commerce i think we spent the year working with the ecosystem to develop the underlying protocol that's going to be needed for this new agentic world the launch of universal commerce protocol at nrf in january with a bunch of partners founding partners i think has been super well received.
33:43 I'm excited now that we have laid the foundation of operability in which agentic commerce can work. And then finally, another person. So this is Michael at Moffat Nathanson. I think I've met him before, but I haven't seen him on the calls before, so that was interesting. He also asked about this. So Wall Street is really tuned in on this. So this is coming from the tippy top down. And he asked a question and Sundar answered, part of what's been good in designing the Universal Protocol is it makes it much easier for users to complete their transactions. But at the same time, it allows merchants to showcase the range of their offerings. If they want to make promotions, et cetera, all that is built into the protocol so they do as a reminder they do have an ad unit in there it's called the deal i think it's called the deal ad digital commerce deal ad unit so you can kind of say if someone gets down to your product you can offer a real-time deal in there anything else stand out to you in the google announcement yeah.
34:36 I mean just kind of echoing your point a in my mind google's version of the protocol ucp is more ambitious than chad gbt's like chad gbt is mostly targeted at ChatGPT specific use case, but the Google version to me is much more general and enables a lot of different use cases. And what I'm excited about is I feel like kind of the same phenomenon you guys had at Channel Advisor back in the day, like when eBay was trying to get people to become merchants, it was a tough sell. But after they became eBay merchants, it became super easy for Amazon to convince them to also be Amazon merchants. And adding that second and third and fourth marketplace became way easier.
35:19 And in that same way, like as we sit here right now recording this podcast, there are hundreds of millions of products in ChedGPT's feed. And so I expect that we're going to see all those products instantly show up in Google's much more ambitious feed very shortly here. And I think that's going to be a pretty exciting moment. Okay. And with that backdrop, it brings us to... Yeah. So Amazon news, I think we're only, what, about three hours into the podcast and we got to Amazon. Spoiler alert, as we've already heard from some of these other companies, Amazon had their earnings call last night and their stock is down 8% as of the moment we started recording here.
36:05 And it's a similar story, a lot of huge traction and success, but they also announced that they're going to make some huge CapEx investments this year on the order of magnitude of $200 billion. And investors did not love that, right? So that was kind of like the takeaway. You're a $2.6 trillion company. You're going to invest $200 billion next year. And that made investors nervous. Scott, what were your big takeaways from the earnings call? From Amazon? Yeah. Yeah, the... I've got so many. Let's talk about them. Yeah. I agree. I mean, were you surprised overall that the stock went down or were you?
36:48 Well, I didn't know what the CapEx was when I saw it. I was watching CNBC and my buddy Gene Munster came on and he was like, he was almost like babbling incoherently. He was so freaked out by the CapEx and I knew something bad had happened. I was like, uh-oh. And then like they had the little aftermarket stock thing up and it went reek and fell off a cliff immediately. And it was like, because the first headlines were like, beats revenue by this much, beats EPS, this part of the business is screaming. And then it was like CapEx to double. And then it was like, pulled the handbrake on the whole thing. Yeah, it is interesting, though. And I think we'll talk about this more. But like, all of these businesses are so capacity constrained.
37:30 You already mentioned it. Like, they all need more AI chips. They all need more power. They all need more cooling for all those chips. And those capacity bottlenecks are atoms, not bits, and they're super expensive to buy more, right? And so it feels like the way to unlock more sales is going to be even higher levels of spending. I listened to a podcast last month that was completely serious and not ironic, talking about how the big unlock in our category is going to be when the data centers move up into space where the solar power is cheaper and the cooling is much easier.
38:12 Speaking of space, Jassy was almost, the Amazon execs sound a lot like Warren Buffett in that they are very cautious about investing this type of money. And he was basically saying, we've never, we've been sitting on the sidelines on a lot of cash for a long time and we're going to deploy it because we've never seen so many high return on investment opportunities in front of us. So, so this number, this $200 billion number, which is the biggest of them all. It includes all the AI stuff. So it's more data center build out and chips and everything. It's also fourth generation of Tranium, which is their chip. It's also, they're launching like thousands of low earth orbit.
38:54 They call it Project Leo satellites to compete with Starlink. They also are coming out with some kind of a satellite phone thing. I'm a little less clear on that one. I thought it was more, I didn't understand there was maybe a phone coming out. Yeah. And then on top of that, they're building a bunch more fulfillment centers because as we'll get into the retail part of the business is like hitting a capacity thing too so they're getting and you know i know this is one of your favorite topics but they sound generally surprised about essentials so it's a really it's a really disconnected time because i've never jassy's like a super kind of straight laced guy and he i've never heard him sound excited i wouldn't know if the normal the way the normal person would would sound it wasn't excited but he was excited for Jassy.
39:38 It was like almost giddy that they found so much stuff to invest in. Also, Wall Street, it took a while for them to figure this out because this number was so big and surprised them. But basically, this makes their free cash flow go to zero or negative for the first time in a very long time. Maybe since like 2010 or something. No one's been able to kind of put that number out. I think people are still looking back to figure out when free cash flow would have been low. But it's not because the business is generating more than ever but they're spending it immediately on the other hand for all this stuff it's crazy. Yeah yeah so let's let's dive into some of the segments and i obviously vote we start with the the best and most important segment which is retail yeah uh you.
40:22 Want to take it or should i. Yeah yeah i'll i'll do the retail one uh because there it's it's mostly good news like a they they generally beat expectations so i think like they were plus plus two percent on north american retail, they're like plus 7% on international versus expectations. I care less about the expectations. General, their store segment grew like 11%. They don't disclose like North American online and offline retail together, but we all build models based on what they do disclose. And I'm waiting for the much smarter people's models to come out. But in my model, North American retail for the year grew 9.1%.
41:06 So put that in perspective, that's Amazon, the first or second largest retail in the United States growing by 9.1% last year. We'll get government data on the rest of the industry next week, but the U.S. Is likely retail grew at 3.1%. So Amazon is the first or second biggest player out there, and they're growing three times the industry average. So that. Is huge. Another super interesting thing is that almost exactly mirrors their growth in units. So it seems like most of their growth is coming from selling more stuff, which is not how most other retailers are growing right now. Most other retailers are growing by selling stuff more expensively.
41:48 And so that to me is a big encouraging sign for Amazon. And you hit it like the likely star here. They talked very little about growth acceleration in general merchandise or electronics or any of their traditionally strong categories. The unlock here is what Amazon calls everyday essentials, which is closely related to grocery, which Amazon calls fresh and frozen. And essentially, you know, during the pandemic, they said, hey, you know, we used to ship stuff from anywhere in the US to any customer in the US. Now we're going to go to a hub and spoke model with regional distribution. And so all the stuff that goes to the Wingo household is going to come from.
42:30 His area of the country, and that's going to be more efficient. And that pivot totally worked. And then last year, they said, you know what we're going to do? Instead of trying to open more grocery stores, we're going to put refrigerators in all those fulfillment centers that are now close to the Wingo's household and deliver fresh and frozen same day from that local fulfillment center to Scott. And so today, there's like 2,300 cities in the United States where they're delivering fresh and frozen grocery same day. And that is all of the unlock at Amazon. In every one of those 2,300 markets, nine of the 10 fastest turning skews, are fresh and frozen. So wherever they offer this capacity, customers are eating it up.
43:21 And to me, this is very reminiscent of history. Like there was this huge retailer that invented discount retailing. It was wildly successful called Walmart. And the founder retired. He got replaced by David Glass. And David came in and said, you know, the one category we're not in that's super important is grocery. We're going to add grocery to all our stores. And everyone told them he was absolutely insane, that grocery was way harder than all the things that Walmart was doing and way lower margin, and then it was going to sink the company. And today, Walmart's the largest grocer in the world, and grocery is like 60% of their business, and it essentially saved the company and helped it continue to grow. And that same thing I feel like is happening to Amazon this year, that they were a very successful general merchant that secretly wasn't selling a lot of grocery other than non-perishables like white paper towels and whatnot.
44:11 And this year, they've unlocked how to become one of the country and potentially the world's largest grocers. And so, you know, even though they're a very big retail business, they're still growing three times faster than the industry average. And that's a pretty big deal. Two quick questions. So the, if we kind of 11 or 12% is kind of where the Amazon online business is growing if we back into it. Yeah. And that, you know, I know it hasn't come out and you're excited, you're giddy over February 10th or whatever it is, but do you think historically that's pretty materially above what the rest of e-commerce is doing? Is that? Yeah, that's what I'm saying. So, so discount the 11% is a slightly like more complicated number, In the US, call that 11%, 9.1%. It's going to be somewhere between 9.1% and 9.5%.
45:02 So you got to take a national. Yeah. So call the US growth rate 9.1%. The government is saying the rest of retail is going to be growing at 3.1%, including Amazon. So all of retail is growing at 3.1%. So Amazon's growing 3x the total market in the US. And then explain, I think I know the answer to this, but I'm curious, like, why was the dude that took over for Sam, right? Like, why is grocery working for both these companies and everyone thought it wouldn't? Yeah, because A, while it's a lower margin, more difficult category because it has spoilage and all this extra complication associated with it, it's 25% of the consumer's discretionary wallet.
45:47 So it's the biggest category of spending and it's the most frequent category of spending. So you might be in market for a new TV every three years, you might be in market for new clothes every year, but you're in market for bananas every week. And so the way to get customer intimacy, the way to move up the importance in the customer's life is to be the provider of all of those groceries. And once you have pole position for that, you can win a lot more share of that customer. And so today, there's about 220 million households in the United States of America. 190 million households visit Walmart once a week. And that would not be true were it not for grocery.
46:31 So it's an LTV play and a frequency thing. It kind of flips it like now you're coming in every week for groceries, you'll pick up a toy or something in the other part of the general merch. Exactly. So now the opportunity cost to buy clothes from that grocer as opposed to going to a mall is much lower, right? The friction to get your Instapot from Walmart is much lower than from Sur La Tabla or wherever. Doesn't everyone have four Instapots by now? They do, but you need a new one every year. They keep getting better. You know what they have in the Instapot now is AI. And a microwave. Yeah. Yeah. I always worry about the device where it's like, oh, it does two things.
47:11 And if you push the wrong button, it's going to explode. Yeah. Don't do those two things simultaneously. Exactly. But yeah. So to me, it's funny, we didn't even really talk about it, but Amazon made an announcement earlier this year, we're closing all of our physical grocery stores except for Whole Foods. So they actually have about 100 Amazon Fresh stores or 60 Amazon Fresh stores. And then they have these Amazon Go stores that we've been making fun of for quite a few years now. And they're closing all of those. And so there was a narrative from some reporters that are like, oh, Amazon can't win grocery, they're getting out of it. And they're actually closing those grocery stores because their hypothesis that they don't need those stores to become America's food provider, or at least compete to be America's food provider, they've proven their hypothesis.
48:03 So that to me is, if I'm a traditional grocer, if I'm Kroger, I'm very worried. I think the fact that Amazon said, oh, you know what? We don't have to join Kroger. We can beat them with our preferred method is a very scary moment for Kroger. Anything else? Should I talk about Marketplace? Yeah. Yeah. Does Marketplace still matter? They do. In fact, it matters a lot. So Marketplaces came in at 62% of total units sold, which is an all-time high watermark. Very exciting. It's only up one percentage point, but hey, we're going to call it a win. So Flywheel Effect for the Marketplace, they launched in 2006, like 20 years ago. I think this may be, or was it seven?
48:50 I think we knew about it in a six and they launched a seven. I can't remember. It may be the 19th anniversary or the 20th anniversary of the marketplace. And it is still humming. The, that famous flywheel drawing is still working. The flywheel is still flying and just creeping up to, to another 1% there. And then that is good. Should I jump into AWS? Yeah, yeah, yeah. I wasn't sure if you wanted to say anything more about essentials. I know you love essentials and groceries, but I want to give you another bite at that. Yeah, no, but again, like if Amazon wins all these essential visits, it's an unlock for all the marketplace sellers, right?
49:30 So we talked about like, what else do you bundle into that trip? It's all the marketplace items that are. You know, much better margin. Interlocking flywheels, all flywheeling together. Exactly. Mega, meta flywheeling. All right, AWS. So this is the one everyone was watching. And this is the only profitable business at Amazon, right? Nope. They are all profitable, but people don't really realize that there's this many retailers still have it stuck in their head that Amazon wouldn't be profitable if it wasn't for that stupid cloud business. But that is fake news. So we're actually going to talk about the most profitable spoiler alert part of Amazon, which has been AWS has been dethroned.
50:10 But you'll tell us about that. I'll talk about AWS. So everyone was expecting an acceleration here, and it slowed down a little bit, and everyone was disappointed. So this was even before they revealed the $200 billion bombshell of CapEx. So AWS revenue growth came in at 17.5. It was good to be the scale they are. They're much bigger scale than everyone else that we've mentioned so far. Their operating income is $10 billion. And remember, Microsoft's total revenue is $30 billion. So their orders of magnitude higher than Azure and I believe Cloud. Which I think they pointed out a number of times in their notes. Yeah, Jassy actually said, I want to point out that it's way harder to grow a $100 billion business, 17%, than a $20 billion business, 30%.
50:58 Wow, your Andy Jassy impersonations are getting better. Yeah, thanks. I have to be less excited. I just want to point out to all the investors... The so the street wanted accelerating revenue and they didn't do it and the important thing is to ask why and it's not because they didn't have a lot of demand it's this capacity problem again and this is why they're spending so much so it's this kind of recurring story that is very common across all four companies we're talking about here margin declined a bit here and management said it's really based to accelerated they're investing so much in hardware now that they're hitting getting hit with this depreciation more quickly than they have before so it's not cash based kind of stuff and then they did say stock-based compensation is up and that that means to me they're it's getting more expensive to find a really good ai and cloud computing people and they're having to pay a lot more in stock comp than maybe they kind of thought they would and And for many years.
52:03 That was probably like endemic advantage they had, right? Like was. Yeah. Yeah. Yeah. If you wanted, if you were into highly available systems, you went to Amazon, period. There wasn't an alternative. And therefore, they could kind of pick and choose candidates. Now, it's an arms race to pick these people up. And you got Zuck out there, you know, throwing Zuck bucks around like crazy. They basically said the demand far exceeds capacity. They can't get enough power. They are. All the fabs are maxed out. So getting GPUs, regardless of whose they are, their own or NVIDIA's, it all goes to this one company called TSMC. We don't have time to talk about, but we're extremely bottlenecked as a planet on the ability to build chips right now.
52:46 And then they're just building out everything as fast as they possibly can. So investors wanted them to give a little more guidance of how it would accelerate. And they were kind of neutral. Neutral, I think, I think they have a lot of uncertainty in their ability to build, Things are out of their control, like concrete, rebar, dirt, power, and silicon, which is also dirt, but another flavor. So that was kind of the AWS one was really kind of the you-know-what in the punch bowl on the earnings call. Which glossed over the one that you're super excited about because it is your favorite category. Who do you love more, the ads business or grocery chasing? If you had to pick one and take it to a deserted island, which one would you take?
53:33 Yeah, I would take the grocery business because without the grocery business, you wouldn't have the ads business. But just one other thing I wanted to add on the AWS. The interesting thing to me is they talk a lot about this like donut or this barbell of demand that it's super like the bleeding edge is creating tons of demand. And everybody needs more AI workloads and more tokens. But they also are seeing like unexpected increased demand for traditional cloud. So there is this carry-on effect where like when you move more workloads to AI, you need more of your data in the cloud, which means you need to buy more traditional stuff. So I would actually argue like it's a nice time for AWS.
54:16 They're actually capacity constrained on both sides at the moment. But you transitioned us to ads and I have a love-hate relationship with this business at Amazon. It's an amazing business. It grew 23% for the quarter. It's up like 22% year over year. So growing much faster than AWS. It significantly exceeded Wall Street's expectations for growth. For the quarter, it was like $15.7 billion a revenue. But if you do the math on the year, I think we're up to about $68 billion a year. That's not even a run rate. The run rate would be bigger.
54:57 But the trailing 12 months, we're at $68 billion. And we don't know exactly how profitable that is. But if you conservatively guess that selling ad units, which have no cost of goods, is 75% gross margin business, which is conservative, then Amazon's ad spun off $51 billion in earned income last year. AWS that we just talked about spun off $45 billion in earned income. So the most conservative estimate of AWS makes it meaningfully more profitable, or I'm sorry, the conservative estimate for ads makes it meaningfully more profitable than AWS, which is exciting.
55:37 Now, I like to remind people, most of those ads are by third-party sellers. Like this is mostly still sponsored product listings from third-party sellers that are now, you know, competing with a billion other ASINs on Amazon, literally. And the only way they're ever going to be found is if they buy an ad. And so that 61% marketplace business that you mentioned really is the unlock for these ads. And that's an important point that I feel like a lot of Amazon's competitors miss. Like I talk to a lot of traditional retailers that don't have a marketplace and they're like, oh my God, Amazon sold $68 billion worth of ads. We need to be selling more ads. We need our fair share.
56:18 And I always have to point out, your fair share is much smaller than you think because. You don't have this marketplace. You don't have a billion items that are competing with each other for visibility. Like there's a viable business for you, but it's, it's not directly proportional to Amazon's. And so, you know, to my friends, uh, Curie Masters and, and Andrew Lipson out there, like you guys are living a collective hallucination by telling everyone that ads is a, the huge looker to business for every retailer. So I just want to get that, get that snippet in. Cause I know I'll hear from them. Good job. Snippet delivered. So anything interesting or surprising to you i will say the ads business starting to get a little more complicated because amazon prime video is a meaningful platform and they have a bunch of exclusive licenses and popular products and so they now are ads no longer just talks about marketplace and and retail ads it's now talking about ads on this broadcast network too one.
57:17 Thing i wanted to ask you about amazon ads was kind of a dog not barking thing because in facebook and google and i didn't cover this in my little google recap because i'm trying to keep this under three hours but sundar was going on and on about their p max d max. Or yeah which is their version of that their p max is their version of um advantage plus from meta yeah. I don't like the brand p max we can talk about that later. Okay the. Whatever it's called, And then here, Amazon didn't mention it that I saw at all that they're using AI. So is it not applicable here or they're just like making so much money they don't even try to optimize it yet? Could that be like a 30% bump to this puppy when they start applying AI to this?
58:03 No. So they actually are applying a lot of AI to it. They're not talking about it as much, but like for sure, I would actually argue Amazon is probably ahead of meta in terms of, AI creative generation for small advertisers. So they've automated all of these media buying and placement and bidding strategy tools. They have a pretty robust system. It's almost like, though, that where Google and Meta grew up before there was AI bidding. And so they kind of have, you know, choose your own adventure. Do you want the human to do it or do you want the robot to do it. Amazon forces you more to have hands off the wheel more often.
58:47 And so they maybe talk about it a little less. I would say it's built in there. I would say, and this is controversial, the ad people probably want to chime in on this, but like a difference to me between Amazon and Google and certainly Meta is that Amazon is inherently more bottom of the funnel. So you are already looking for products when you go to Amazon, and so those ads are helping you pick one product over another, but less often are those ads creating incremental demand for a purchase that otherwise didn't exist. And that's important, and if you're one of the brands competing for that sale, it's super important. The meta ads, when you were just doom scrolling on Instagram and you suddenly discovered something you didn't know existed that now you're going to buy, that is a more valuable occurrence.
59:38 And meta can charge more for that than Amazon can charge for the brand competition after the demand is already there. So to me, one of the interesting differences here is because Amazon at the moment only lives at the very bottom of the funnel is their ads are a little more tactical and less strategic. You know, Google and Meta, I would argue, have had a blend and they've leaned one way versus the other. And we have awful metrics in the advertising industry that make it hard to really understand the difference between just getting attribution for a sale that was going to happen anyway and truly incrementally creating a sale. But in my mind, these upper funnel platforms like Meta and for sure TikTok have more ability to create net new sales and therefore are potentially more valuable than the Amazon ad businesses at the moment.
1:00:32 Interesting. okay the i have more questions but we'll yeah we'll go on and we can talk about it part two the behind the scenes podcast for guidance amazon and at this point the market didn't care because that's during the aws is when they said the capex is going to be 200 billion so everyone's ears turned off during that this next port part but they said, That they're guiding Q1 revenue above consensus. And then margins looked like they were a little bit below expectations, which kind of people added into the AWS concerns. And I think what's going on there is it's not a structural problem in the business. It's the fact they just can't get the capacity online as much as the demand is out there.
1:01:14 So that was it. And then, oh, a couple other tidbits I pulled out. They got a question on tariffs and jassy basically yawned and said we don't see any pricing pressure from that it's been a non-issue for us they also were talking a lot about amazon hall having over a million items that's their like timu competitor is that right exactly. Right yeah and it's in like 23 countries now so they stopped talking about it because they launched it right before de minimis got turned off in the u.s which was unfortunate timing for them But it appears they've been leaning in and gaining traction. Yeah. And then they talked about, they got an explicit question about why are you blocking or why are you not listing on Google Shopping, which I thought was interesting.
1:01:59 So they basically said on July 21st, they stopped and they blocked all the crawlers. And they basically said, you know, we view that we have the largest database of products. We have the most reviews and we have the deepest pricing data. And that's a very important data to us and gives us an information asymmetry we want to control, which is a valid, this is kind of, you know, very clearly their strategy. They're basically saying, we understand how this world works. Data is the most valuable thing and we do not want to give away our valuable thing, which is data for free. So we're locking it behind this, you know, giant gate that no one's going to come into.
1:02:41 And then, you know, The other thing that's interesting is ads, because to your point, they're only constrained by the ad load and Amazon seems to see no end insight to the amount of ad load they'll add on there, which is, I do think it's, I do think they're going to create this self-fulfilling prophecy by juicing the ad so much that it's going to drive people away from the front end experience of Amazon. That's one of the reasons I use Agentic Commerce is for me, outside of Rufus, now I will say Rufus has gotten me back to using it. The core, the traditional Amazon ad experiences or search experience has gotten so polluted with ads, I can't find what I'm looking for. And it's way, way too high in my mind.
1:03:21 But because of that, the ad piece is growing about 23% in AWS, 17.5. If AWS wasn't constrained, I think it would catch up, but it has constraints with the physical world that ads do not. Then there were a couple more agentic commerce tidbits, and I promise we're getting to the end here. So Michael Morton, also from Moffat and Nathanson, he seems to be the guy that's asking this question to everybody, which I need to chat with Michael. I need to actually put that on my task list. I love that he's asking these hard questions for people. I'm not sure he'll get invited back, but we'll see. He asked Jassy, what do you think about all this stuff going on with agents? Explain your strategy more clearly to us because everyone can't, when Jassy speaks, everyone can't tell is he saying they're going to do this or not and then everyone's left we all have a split opinion kind of like you and I had a split opinion a little bit on the meta piece so Jassy said I'm very optimistic about the consumer experience that'll ultimately be what customers use for agente shopping and I think it's good for customers I think it's going to make it easier for them it's a big piece of why we've invested as significantly we have in Rufus so So he took the question and pivoted to Rufus, which is a smart PR strategy.
1:04:33 If you haven't checked out Rufus recently, I really encourage you to do this. I actually think he's right on this. I've gotten much, much better and it's gotten much, much better and getting better every month. Then this was interesting. They, this is a new stat as far as I can tell. We have 300 million customers who used Rufus throughout 2025. I bet that was super back and loaded. And then customers who use Rufus are about 60% more likely to complete a purchase. Okay. So logically, if that was true, you'll get rid of the search experience and do this, except there is a little fly in the ointment and it's that ad business. So you, you know, I think Google is close to making the switch and I think Amazon has not figured out how to monetize Rufus and they're not going to switch that until they do. So yeah.
1:05:19 It is, it does, it is an interesting decision point, right? Because if this really improved conversions that much, you would, and you didn't have that business, I think you would instantly get more, as many people in the Rufus as you can, could. Yeah. What do you think? Well, so I generally agree with you, although I feel like you tripped into the biggest fallacy in this whole thing. And this should be, we should do a whole podcast about this. Customers that use rufus are 60% more likely to complete a purchase that is true that doesn't mean that it's creating 60% more purchases uh causality yes correlation versus causation and we know for a fact we don't have to wonder we know for a fact it's correlation not causation because if it was correlation 38% of customers use rufus they're 60% more likely to purchase they'd be growing a lot faster than 11%.
1:06:12 It's evident in their data that the customers that like Amazon the most and buy the most from Amazon are more likely to use Rufus. So that's great. There's nothing wrong with that. But just don't confuse correlation and causation. So it's a nice vanity stat for Andy to talk about. It does not mean they have some new unlock that's dramatically creating tons of new purchases. So that's the sort of first thing. But that question to me unlocked the most interesting conversation in this whole deal, which is... Andy, you guys aren't on the third-party AI commerce experiences. Like all those hundreds of millions of Walmart products are for sale right now on ChatGBT. You're not.
1:06:56 Are you worried about that, right? And he had a two-part answer. He's like, part one, we'll probably eventually get there. We think the experiences are broken there and they need to get better. So we need to work with them to improve the data and the experiences. And we need to agree on the right value transfer between us and them, which is code for we need to make sure we get the right margins from that. But he kind of implied that'll all get worked out eventually and we'll find a way to have a presence on all those things. But he drew this distinction between what he calls horizontal LLMs, which are kind of general purpose LLMs that do a lot of different things, one of which could be shopping, and vertical LLMs. And he would call Rufus a vertical LLM. And he made a passionate plea for why vertical LLMs are better.
1:07:42 And in the long run, he believes customers are going to pick vertical LLMs. So when you go shopping, you're not going to want to go to a general robot. You're going to want to go to a shopping specific robot. And he would argue Rufus is the best version of that. I don't know what the answer is, but I'm way less convinced that I'm way less certain than Andy is. And I would argue Amazon's biggest advantage, the unlock that has caused them to win the most in commerce is assortment. That people don't want to go to a different store for every product. And Amazon has now taken a billion products and put them all under one roof. Like it's the number one advantage that Amazon has. It's the number one utility that customers choose Amazon for is that assortment.
1:08:23 And the robots make that assortment irrelevant. Like the robots don't care if they have to buy the items from three different vendors or one different vendor. And so I like whether Andy wants to admit it on an earnings call or not, these horizontal bots are a huge threat. Now, do they have less shopping data than Amazon has right now? Yes, they do. Do they have, you know, rougher commerce experiences? Yes, they do. But each side of this fence has different strengths and weaknesses. And I do think that's going to be the fundamental battle of the next 18 months is, you know, are people going to be satisfied to shop from the microchip that John Ives installs in our brain? Or are we still going to go to Amazon and use Rufus 10.0 to do all our shopping?
1:09:06 And the answer to that question is going to decide, you know, who the most valuable company is 10 years from now. Yep. Okay. So to summarize, here's kind of the answer to the AI bubble. Are we in a bubble? So let me frame it this way. So we had these companies say to us, here's the demand that we cannot satisfy right now, also known as a backlog. If we had the capacity, this is the revenue that would come from that capacity. Google, 240 billion. Azure, 368 billion. That's the biggest one. AWS, 244 billion. Add that together, that's $852 billion of revenue, like almost a trillion dollars of revenue, which is crazy.
1:09:47 Then here's the CapEx they want to spend on that. Now, this is a little apples and oranges because I'm going to throw Meta in here and Meta doesn't really have a backlog because they're just kind of dogfooding this to make their ads better. But they're going to do 115 to 135. Microsoft, 160 billion. Google, 185 billion. AWS, the biggest, 200 billion. Now, they didn't tell us what inside of there is Azure, fulfillment centers, and Leo. But you've got to imagine Azure is the lion's share of it. But Leo is pretty expensive because you're buying satellites. So maybe it's 150, but let's just call it 200 because what's 50 billion between trends, right? And that gives us 700 to 750 billion in CapEx.
1:10:28 So basically saying in CapEx, revenue is annualized. CapEx is seven years or 10. I think they're now saying these chips last, they have an effective life of, I can't remember if it's normally five and they're saying seven or seven to 10, whatever it is. So that is why they're doing it and if this math is right and we trust that these three companies aren't all lying with four companies i guess three on the demand side then this is kind of a no-brainer for so any any business person would make this trade-off every day if they had access to the capital and these guys do because they're pulling it out of free cash flow they're basically saying it is way smarter for us to invest this dollar here than anywhere else we see acquisitions share buybacks, interest rates.
1:11:13 Startup investing, everything out there. This has got the highest IRR out there. So I think it's a very clear signal if you really put this together that this is a, the craziest thing we've ever seen in my 30 years, you know, being in this world and understanding this stuff. And if we get, you know, 80% of this to come out, we could have five and $10 trillion companies coming out of this because they're basically adding a trillion dollars of revenue amongst these companies and this big step up function. That's, I do not think it's a bubble. That's where I come out on it. Oh, interesting. Yeah. So, and you're way more sophisticated investor than I am. And so I'll defer to you on that.
1:11:54 Like to me, I a hundred percent agree with you. The there's, there's a lot more gas in this car. It's, I guess intuitively in my mind, it's hard to call it like this. There's so much hype in this space. It's such the bleeding edge and it's growing so fast. And it's, everybody's talking about so much, like, of course, it's a bubble, like, like, is it the very front of the bubble or the very end of the bubble? Like, are we going to keep pouring gas in that bubble for five more years before a burst? And, you know, when when it does burst and normalizes, will it be bigger and smaller than it is today? I agree with you. I think there's a lot more room in it. And I think like this is inherently a huge market and a huge opportunity, but it probably is simultaneously slightly overhyped and there's probably you know slightly too much exuberism in it about the specific companies and where they are today is that could i be.
1:12:47 Right okay yeah and it also so i'm less qualified you know i'm not gonna go into cnbc and talk about the bubble thing but this is this is why i think people that aren't really pivoting towards this in our space because they don't think the companies are serious or are you know these companies this is the most serious thing these companies ever done you know the amount of money they're spending and their career you know they're all betting their companies on ai and then for every one of them agentic is a big pillar of this so yeah and and by. The way that's not just seven random people that's seven highly accomplished people with pretty good track records that are darn good operators.
1:13:29 Yeah, these are the best operators, you know, except Tim Cook. I put him up there and, you know, maybe the CEO of Exxon or something or something like that. And Elon, you know, these are the best operators on the planet. Like, you know, five of 10 saying this is where we're betting money right now. Yeah. So look, we may have set the record for the longest podcast. I'm so sorry. Hopefully it's been valuable. I'm just going to tease one other thing that we don't have time to get into. But also last week, Walmart changed leadership. So Doug McMillan, the very successful CEO, stepped down. There was no surprise about who his successor would be. It's this John Furner. But what has been super interesting is who backfills all of John Furner and everyone that moves up.
1:14:15 So John Furner goes from president of the U.S. to CEO of Walmart, Inc., who becomes the new president of the U.S. And there were like three totally viable traditional Walmart candidates that everybody was betting on. I had a bet on one. My bet proved wrong. We were all wrong because the new CEO of Walmart U.S. Is David Gugina, who was formerly the chief e-commerce officer at Walmart. So to me, that is big news. The largest retailer in the United States, the CEO is now the former head of e-commerce. We've been talking about these digital guys getting in the C-suite for a long time. And now it's the CEO chair at Walmart. So that to me is a huge deal.
1:14:57 And a lot of his lieutenants are now moving up. So one of his lieutenants runs all of marketing at Walmart. So it's an incredible new world. And in that world, Walmart just last week or two days ago passed the $1 trillion market cap valuation. So I think they're the first traditional retailer to be valued at over a trillion dollars. And that to me is amazing. I'll take the framing on the long episode. I think people love our Amazon earnings episodes and they've been dying for more content and we're delivering it. We're listening to our audience. To the 5,000 people that complained about our show frequency at NRF, be careful what you ask for.
1:15:39 Yep, just listen to it in three chunks and it's like having three episodes. And yeah, it's less work for our audio team to split it into three. So do it yourself. There you go. Awesome. Scott, always fun talking to you. And as always, I hope, though, our listeners got some value out of it. And we are still accepting reviews. As old as we are, the reviews engine is still open. So you can still jump on iTunes and give us that five-star review. And we're very grateful if you do. Yeah, and because it's a six-hour episode, I think it's worthy of six five-stars. So leave those 30 stars. Absolutely. And until next time, happy commercing. Happy agentic commercing.
Overcast
PocketCasts
Amazon Music
Audible
Youtube Music
RSS Feed