The Six Five Pod | EP 283: Mag 7 Earnings: Meta's Capex, AWS Comeback, NVIDIA GTC 2025, and Apple's iPhone Miss

On this episode of The Six Five Pod, hosts Patrick Moorhead and Daniel Newman discuss the tech news stories that made headlines this week. The handpicked topics for this week are:

1. Key Takeaways from NVIDIA GTC and Infrastructure Build-Out: Jensen Huang's keynote delivery. Nvidia's co-architecture approach to power systems, water systems, and manufacturing. Partnerships with Vertiv, Siemens, and GE Vernova for infrastructure development. Key partnership announcements from NVIDIA to build seven supercomputers, competing with AMD's dominance.

‍

2. OpenAI and Microsoft Partnership Restructuring: Microsoft's $12-13 billion investment for 50/50 partnership structure. Renegotiation allowing OpenAI conversion to a for-profit entity. Microsoft's potential 27% ownership stake valued at approximately $270 billion. Sam Altman's equity position and IPO preparation for a potential trillion-dollar valuation.

‍

3. Qualcomm's AI Chip Launch: AI 200 and AI 250 announcement driving 20% stock price surge. Strong Wall Street reaction despite limited technical details available. Credible entry into the data center market with scale-up methodology. 2027 timeline for scale-up technologies, including NVLink adoption.

‍

4. DOE Supercomputer Partnerships: AMD's billion-dollar partnership for two additional supercomputers. Continued dominance in high-performance computing with 64-bit precision. U.S. Secretary of Energy Chris Wright's recent recognition from both NVIDIA’s Jensen Huang and AMD’s Lisa Su. Highlights of government investments towards winning science across multiple domains.

‍

5. AWS Anthropic Trainium Partnership: A one-million Trainium chips commitment from Anthropic. Validation of AWS's custom silicon strategy. Recognition that all available chips are selling in the current market. Multi-generation improvement trajectory similar to Google's TPU.

‍

6. Google Public Sector Event Highlights: Google's military and government sector transformation under Google Cloud CEO, Thomas Kurian. Impressive Gemini for Government agent demonstrations. Seven-minute agent creation showcasing platform capabilities. On-premise GDC deployment with Lockheed Martin for air-gapped AI.

‍

7. Government Stakes Debate: Discussion of AI, quantum, rare earth minerals, and chip manufacturing.

‍

8. Federal Reserve Rate Cut: Fed Chairman Powell's extensive data center commentary.

‍

9. OpenAI Valuation: A trillion-dollar IPO valuation deemed "completely bonkers.”

‍

10. ServiceNow Earnings, Alphabet/Google Earnings, Meta Earnings: Unpacking tech’s earnings season.

‍

11. Microsoft Azure: 40% Azure growth with $400 billion booked business.

‍

For a deeper dive into each topic, please click on the links above. Be sure to subscribe to The Six Five Pod so you never miss an episode.

Or listen to the audio here:
‍

‍
Disclaimer: The Six Five Pod is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.

‍

Daniel Newman: Oh, boy. Those, like, 10 seconds right before the show starts are the best part. I'm so glad, though, the recording isn't actually on because it would be like, great blackmail material later. But let's tell everybody I think, you know, there's a couple of things that tend to happen right before the camera goes off.

‍

Patrick Moorhead: Yeah, I think I said I might be going bankrupt and cashless, but, boy, do I love to have a good workout. It makes everything good.

‍

Daniel Newman: I mean, I think the hyperbole and the. The kind of these comments that we make. You know, I do not expect to see you as one of those people walking around shouting on Second street soon in Austin. But, yeah, I do think that people tend to never understand what the real gyrations of being an entrepreneur are. The highs and the lows. And when. When it's high, it's like, oh, my God, you know, the world is our oyster. And then on those days that it's low, it is absolutely burning to the ground. And I. I don't think, you know, many people that have not, you know, built a company, by the way, finance their own businesses. I'm not talking about people that raise other people's money. I'm talking about building companies, putting it all on their own backs. So that is super fascinating. But overall, just the stuff that gets said right before. Because the other thing that tends to happen right before we open is you remind me to do a good job. Like every time you're like, Dan, seriously, like, like, you know, everybody that's watching, it's like, generally I. I don't plan to come and show up and do well. So. So Pat reminded me. You remember that. What was that tweet I sent you last night that was actually really funny. You didn't respond to it. But the one I sent you, Ben Horowitz, on the brilliant advice you get from your investors. Yeah, one of my favorite parts, one of my favorite pieces of advice I got, and I got like 27 times, they'd say, look, here's the key. Hire players. Like, okay, yeah, my plan was to hire a bunch of morons instead. Yes, I. I plan to come on and do terrible things every week. And I really appreciate that, that you always do remind me.

‍

Patrick Moorhead: But, yeah, I didn't, I did not remind you this time because I think I was the one who was scowling, you know.

‍

Daniel Newman: Yeah, well, you, you did come in with a pump. So, I mean, you know, we haven't talked as much about that lately. I think some of that has probably been due to the litany of little setbacks. Most of them I think are overcomeable. That's a word. But you've definitely, after a couple of amazing years of clean health, weight loss, muscle building, you did find your age for a minute, even though you look younger. Like, you know, I think that incline bench maybe went down just a little too far. Those leg presses, you maybe just pushed a little too much weight.

‍

Patrick Moorhead: I gotta tell you though, I'm shooting, shooting more peptides than I ever have before. And in fact, I'm taking it right into the injured areas which are. So I injured my hernia, I re injured my shoulder. But I have a thesis on it that I've heard from two people that I will share with you later, which is what I did a. I was doing a full gut, gut reset, like triple layer, the biggest one I ever done before. I was making fun of my doctor when he warned me about it. But it hit me last week like. And then my trainer and my nutritionist said I really think it was all the inflammation that you created while it was on that affected your shoulder and your hernia. And so it was crazy that both of them said that independently. So my trainer and my nutritionist. So we'll see. I ended it on Saturday and we will see if I bounce back in the middle. I'm shooting TB500.

‍

Daniel Newman: Nice.

‍

Patrick Moorhead: And HCG right into the affected HCG or BPC. Sorry, sorry. BPC.

‍

Daniel Newman: Sorry. I was like, I was like, I don't know if, if more test signals are going to help, but I guess having more tests could never hurt. By the way, I watched that Charlie Sheen special on Netflix. Just everybody out there. Wow, what a. We don't have time to go into that one. But yeah, you know, I remember tiger blood.

‍

Patrick Moorhead: Yes.

‍

Daniel Newman: So. So it was, he was at like, I think he told me he's at 2000 or 3000 was his test levels. He was rubbing on the cream and doing blows all day. That was what he liked, wow. It was testosterone and cocaine. 

‍

Patrick Moorhead: Wow.

‍

Daniel Newman: The story was great. Like getting it underneath it. I watched it on the flight this week, by the way. What a week. I mean you, you were out Sunday, I mean at lunchtime, you know, the GTC pregame show, which, God, you were like on every Gosh darn billboard in that place. I was there. I was like, I feel like I know a celebrity. Is that, like, who is that guy? You know, it is, you know, a talk show. It was great to see you everywhere, but it was a lot of work, right? I mean, that was not. That was. That was like, real. That kind of reminds me of back when we did that Dell event. You got like, pretty pre meetings. Every segment was thought about. Every interview was, this was not a, hey, just sit at this and talk to people.

Patrick Moorhead: I did not know exactly what I was walking into, which was a full live broadcast interviewing 16 people with interstitials Jensen coming up on stage. But. But it was, it was. It was well worth it. I met a lot of really important people. I got to know Brad Gerstner well, you know, a bunch of VCs. Probably the toughest segment for me, Daniel, was the first one where I'm on stage literally rocking five VCs and one industry analyst. Me. Many of them are billionaires, and I am not. And, you know, they use a different language, they use a different, you know, ontology and how they. How they look at the world. And I am pretty much a meat and potatoes guy, right, as opposed to what they're doing in venture. The interesting thing about them, I don't even think venture is venture anymore because they're putting in a lot of investments into public companies. Right, Brad? And Altimeter has a very hardcore investment into Nvidia, among other public companies. But, yeah, in three and a half hours, I interviewed 16 people across Venture, across infrastructure, meaning power, cooling, all the way to the grid, science and quantum robotics and manufacturing. And yeah, it was. It really was. It was really exceptional.

‍

Daniel Newman: Yeah, it was nice. I will admit. I got, you know, I came early, I brought my pom poms. I was there to cheer. I was really happy for you to be up there on the big stage. I was sort of, I'm not gonna lie, I didn't mind not doing the work. I always liked the attention. I won't lie about that, but I didn't mind. It sounded, listening to you, like a lot of work, like the scriptedness, you know, you and I are so fluid. We don't have a script most of the time when we talk, we maybe have like five talking points for our questions. But, like, you almost, almost had to play the role of, like, John Ford. You almost, like, had to, you know, but you did a great job. So at least in your path between now and when you become a billionaire, sir, you can always become a news broadcaster.

‍

Patrick Moorhead: You have my, my biggest takeaway. But by the way, the best events were the small events. Right. I was in two events where there were maybe 50 people, which include Jensen and all the, all the different speakers. But I did get a signed book from Jensen. Looks like him. Not an auto pen. And I was invited to an exclusive dinner or, or, or I don't even know what you call it, reception with big leaders where Jensen gave this amazing, inspiring stump speech.

‍

Daniel Newman: But then like five minutes later, he was in South Korea doing like arm lapped shots with the CEO of Sam.

Patrick Moorhead: I mean, no, he was supposed to do the keynote for the Google Public Cloud Summit, Public Sector Summit. And, and he, you know, he did talk about, you know, Trump wanted him to come to Korea with him. Right. And, and he, the day before, the day of his keynote, but I think he left early to go do that. I think what's not being written about is the time that he spent with President Trump there.

‍

Daniel Newman: So, yeah, there seemed to be some movement there between. And we'll talk about that here. And by the way, like, you and I could, could, could talk a lot, you know, and this is kind of fun to just riff, but I'm sure our audience wants us to get to business a little bit, so. Yeah. So the great show. You know, we always do our show, three segments. We get the decode breaking down the week we've started a little bit ahead of time here. We do the flip where Pat beats me in a simulated argument. And then we do bulls and bears, which again, there, there's no shortage of things this week because there has been a ton of earnings this week as well. Pat, we're going to hit gtc. We're going to hit on open AI surprise news. We're going to hit on Qualcomm launching an AI chip. So many more things. Let's stop teasing our audience and stop sharing all the blitherings and blatherings of our life. And let's get to the decode.

‍

Patrick Moorhead: Yes.

‍

Daniel Newman: All right. Well, you've sort of started to tease it. Couldn't help ourselves. But Pat, you know, why don't you do yours. There's no way the GTC breakdown could be the entire show. So in a few minutes, give us, give me the top sort of hits from, from GTC for you.

‍

Patrick Moorhead: Yeah. So first off, I've been to almost every gtc, so since its inception. Obviously, I didn't attend when I was at amd, but this was definitely the best one yet in terms of the keynote and I think the content Jensen was very clear, he was concise, he didn't drift. And you know, he's so funny. He made some jokes in these small groups that said, hey, you know, my family tells me not to be boring, right? And what he's trying to do is educate, right? I can relate to that. My best critics and supporters are my family on the pod. So the biggest thing for me is to build out the infrastructure. Okay. Right now, by the way, there's no math that makes sense to be able to build enough infrastructure yet related to power and cooling. So one of the most interesting things is what, what Nvidia is doing is they are co-archiving power systems, water systems and manufacturing systems to all with Nvidia technologies to do the build out. So you've got companies like Vertiv, Siemens, GE Vernova. Think of GE Vernova as grid folks and all the different ways and also power creation outside the fence. Think of G.E. vernova and Vertiv as inside the fence. Right? And Siemens being the software that basically manages these things. So that was a big thing for me. And it got me thinking, what do all of Nvidia's competitors do? Right? My guess is that they're going to have to. Back in the old days it used to be, hey, you're a socket stealer, right? Like you're playing into an ecosystem that somebody created, kind of like AMD versus Intel in the early days with CPUs. So I think the only choice that the competitors have to do is to work inside of the Nvidia infrastructure to make this work. Second, a big announcement. Just when I thought AMD was going to run the table in supercomputers as they did an announcement the day before Nvidia I think announced seven supercomputers Doe argonne. So it was just absolutely a mind blower. Something that I'm sure that you had your eyes on with your investment audience was this 500 billion AI chip bookings, okay. And there's a lot of questions on was this going to be for the year. But it was over the lifetime, I believe of Blackwell over five quarters. Is that, is that right, Daniel?

‍

Daniel Newman: I think it is the next five quarters and it includes the rest of Blackwell and possibly some early Rubin bookings was the way he sort of explained it. But it's this visibility between now and the end of 26 to the order book.

‍

Patrick Moorhead: Yeah. And my final comment to give you some breathing some air here is Jensen ended the show by saying Nvidia is helping to make America great again. And I think that encapsulates everything, right? It's about building, it's about manufacturing here in the United States. I mean, Foxconn is building gigantic factories here in the United States for AI servers. It was about creating enough energy. And if you remember, I was at an event winning the AI race where Trump showed up, where he did some EOs on permitting for things like energy. But here's the open loop, guys. There is not enough energy. The backlog of even natural gas generators for electricity. I mean, the lead time is five years. So I don't know how we're going to get here. There was a lot of talk about, you know, Moonshot right here. So that's more research to do. But what a show.

‍

Daniel Newman: Yeah, I, you know, in the vast wasteland of media appearances I made, I kept saying, I think you said something similar. That was probably Jensen's best GTC keynote. It was. He didn't do the drift that he does as much. He was still funny and picky at times, but he didn't drift into like, you know, into these sorts of tangents. And, and he stayed on. He had a lot to announce. He sort of broke all the different businesses down. You know, he broke down high performance compute, he broke down Quantum, he broke down Edge, Intelco, he broke down Autonomy and robotics, he broke, you know, data centers. And he really kind of showed all the different places that Nvidia is going to play. And I like the fact that he kind of took the bubble babble on, heads on. He took it on like directly kind of said what it is called, what it is, explained what it is. And I think that's important right now. I think he showed that he talked a lot about the collaboration of the ecosystem and all the contributors that are part of this. You mentioned some of the ODMs, Pat, in those participating in this process, the OEMs participating, the ISVs, the partnerships. And in each one he sort of highlighted a couple of different partnerships, highlighted Nokia made an investment there. People are like, oh, should he be investing? Well, yes. I mean, he wants the AI powered 5G 6G edge to be built on Nvidia. So a small investment in one of the leading players to help make that possible. I think that's smart money. I mean, that's a smart use when you have the kind of balance sheet that Jensen has access to to reinvest it into the areas you want to succeed. He's doing the same thing in Quantum. He's doing the same thing with you know AI factory build outs. I thought he did a good job of sort of bringing a relevant visibility to the Metaverse Omniverse with a real use case which is actually designing and building these AI factories using technology that is a really reasonable and useful example of how that technology can help. These projects are incredibly complicated and to actually be able to render them and simulate them and be able to validate them before you, before you shovel them. Put a shovel into the ground. That's really powerful. So I thought it was a really well done keynote. I think that half trillion dollar number is what sent the stock parabolic in the middle of it. I think I calculated they could do like 300 billion next year in revenue and that would mean over 150 billion in income at the rate they perform. And it is another indicator that the market is bigger than the forecasts. It's actually larger than all the forecasts we've seen so far. Which means more upwards revisions not just for Nvidia but for all the companies that touch Nvidia. So it's a big one. A lot more we could talk about but there's a lot more to talk about. So I'm going to head on to OpenAI and Microsoft. I think that's a worthwhile one. So you know a lot of people are kind of confused as to what that original partnership was. I believe Microsoft put something like 12 or 13 billion dollars in and two different investments over the up to this point that had given them basically a 5050 partnership with Open AI for their non. What's that? It's not super intelligence, what do you call it? AGI? Non AGI. And the idea was they were going to do a profit share or something with everything that up to AGI. Obviously there's nothing to share and profit when you burn as much capital as OpenAI is. And so one of the things is there's a non-profit. OpenAI and Microsoft have been in this ongoing somewhat tenuous negotiation to figure out what that structure looks like. And as OpenAI's valuation has risen to nearly half a trillion and I believe this week there was an announcement that the IPO might be coming as soon as next year for a trillion dollars. Satya could turn that 12 or 13. It might, don't quote me on the exact billion into something like 270 billion at that point because the renegotiation will now allow Sam Altman to convert to a for profit then convert to be able to do an IPO to about a 27 ownership stake in the company. Pat, that's a massive sea change and it's, I think, a good one for both companies. I think that this gets the halo or. Sorry, that's not right. The hangover of that original deal off of Sam's back gives him flexibility to do what he needs to do. I also think for Satya, it shows a kind of a. It was a master chess move to become the largest shareholder to have a ton of control over the company. And he benefits from every deal that OpenAI ends up making. And like I said, whether they make money or not, I expect that massive IPO to ring the register for Satya and team. So I'll give you a little back. There's a couple other open AI items.

‍

Patrick Moorhead: I don't know.

‍

Daniel Newman: If you want to talk about the Microsoft thing, I'll give you this topic to sort of run with.

‍

Patrick Moorhead: I don't have a whole lot to add. I heard, you know, I had a couple presses hit me up saying, hey, Sam doesn't have any equity in the negotiation for forward profit company. That just seems like complete nonsense to me. Did you pull anything out of that? Have you been following this at all?

‍

Daniel Newman: I have been following it. I, like you, have a hard time believing that to be the case. But I think until we start to see filings of how things are broken up and is it, you know, held in some other entity? Is there some sub entity that Sam has a piece of? I feel like that's probably been part of the ongoing negotiation, but I seriously doubt he's gonna walk away with this thing with, with nothing. I just don't.

‍

Patrick Moorhead: Such nonsense. Yeah. You know, OpenAI did make a claim that said it was going to have an AI researcher. That what I call is first principle, meaning it. It actually thinks about it. On its own. So think of it as AGI for research. You know, I'm a little skeptical given that you do have to train it how to think and it's going to train based on how people think. But I'm the first one to say that maybe I don't understand the technology well enough. They're certainly not going to do it with chat GPT5 in 2025. So we're looking at 2028, which means it could be that Chat GPT8 will need to be a new type of technology that's created to be able to. To do this. So the claim is that it just won't hoover data and learn. It's only going to be as good as the data. It will learn how to think. So I will see it when I see the research papers about a different way of doing this. But it certainly would be amazing because we could get cures for cancer, we could get, you know, new drugs quicker. I had some pretty incredible people on stage at the Nvidia event who one was genome and the other was pharmaceutical. And then the small groups met with Eli Lilly's chief scientist that talks about how much they're shrinking the time to trial. Trial time is still, you know, two, three years plus. But the ability to get there with a drug that's more likely to work than not is being improved by this technology.

‍

Daniel Newman: Yeah, I agree with that. Although I like the direction, I'm still skeptical on some of the time horizons. The generation to generation improvements we're getting are becoming more and more iterative from chat GBT to jet, like where the first few were so, you know, revolutionary I think we're seeing. And by the way, this is a good thing in many ways, like because we're getting that kind of meaningful improvement. You know, the pat. Lately the AI slop word has become the popular bear thesis. It's just slop. And there is some of that. But the thing about it is everything gets better every day, literally every single day as we continue to use this stuff, the outputs get better. And so while people are right that some of this stuff is slop, I see some analysts in our field that are posting X posts that are clearly just written by, by, by a chatbot. Right? Absolutely garbage. But, at the same time, like with what we've built on our platform, and I know you've been building some interesting stuff with Pico Patrick Jr. It's big now, man, he's crushing it. But you know, as we continue using it, the things it's generating become better and better and better. And the timeline in which this stuff's improving is like days, weeks now. So I think the whole idea that, that he can't get there. But I also do see, don't you see, like it's funny because him and Elon are always feuding, but don't you see a little bit of similarities and kind of the promises they're making of when things can get done versus like the reality, but yet they're still, they're still change makers. They're doing really impressive stuff. So it's like I, I, you know, like I said, I think the Don Lemon interview where Elon said that I build the future and you just, you just react to it. I can't take any more. You know, I got to be humbled a little bit by that because that's a lot of what we do. We react to the work other people do. It's still fun though. I still like what we do. Except the days that you feel, what did you say? Bankrupt and on the streets or whatever, however you started this, this pod.

‍

Patrick Moorhead: Yeah. Bankrupt and cashless.

‍

Daniel Newman: Cashless. Which, you know, that's, that's sure. It's too bad. All right. Hey, here's another one. You know, I have been quite outspoken that I think, you know, Qualcomm's future is not in handsets, that they have to make a pivot to edge AI robotics in the data center. Well, Pat, this week AI 200, 250 it came out. IT stock went surging. There've been some critics, there've been some compliments kind of. What was your read on this, on this launch from Qualcomm?

Patrick Moorhead: Yeah, Qualcomm has done an amazing job diversifying their business model. They're definitely up there in the top three of the ability to get into other, other businesses quickly and be successful.

‍

Daniel Newman: Right.

‍

Patrick Moorhead: You have the Automotive, you have IoT TBD on driving a ton of revenue. On the PC side. They definitely shook it up there. I like to simplify this by just saying if you want Wall Street to react favorably, get into a market that they are watching and putting premiums on and then do it credibly. Stock popped about 20%. Came down to, I think we're at about a 6% adder here. And by the way, that was with almost no detail. Okay, no technical details. They do have a customer called Humane, but there were no details technologically. And the ones that matter is the amount of processing power. But in a scale up methodology, the type of networking that they're using, you know, there's going to be a lot of options particularly in 2027 for scale up technologies. ESUN, UA link. I mean there's going to be a, they, they are an NVLink fusion adopter as well. So I feel pretty comfortable in 2027 that they'll have a, a, a scale up. What I find particularly interesting is first of all they're scaling their NPU technology, likely in this AI 200 which is in smartphones. I don't think there's anybody else that's doing that. They're either starting from scratch or they take something that's a beast and then try to narrow it down to be more power efficient. I find that super interesting. The other thing is this AI250 is going to be this near memory technology. There weren't any details on this, but what I will tell you is that I have talked to a lot of particularly RISC V accelerator companies and it's a very symbiotic relationship with the processor as well as the hardened NPU or AI accelerator silicon that gives you different levels of programmability but really good bandwidth. Then on the cost side, right, we're basically sold out until 2027. I think on HBM, this uses LPDDR on the AI 250. So, the amount of memory that you can put on any one accelerator with the lowest latency does give you an advantage. But I, I need to see, I'd like to see more details before I can, you know, Babe Ruth this or, or put my percentage of, or of possibility that this could be a, a big success having Humane as a, as a customer. I don't know if it's an ink deal or, or an loi. I'd like to think that there is some NRE that's changing hands where Humane and other potential customers are paying Qualcomm up front for all of the R and D work. But in the end more competition is better and I wish good luck to the folks at Qualcomm.

‍

Daniel Newman: Yeah, my readpad wins big while winning small. You know, we've got that 583 billion number we heard Jensen up the number for even just next year in terms of capex expected. You know, realistically we could see a trillion dollars a year in AI chip infrastructure by 2030. You know, there's kind of this breakdown. I don't think LPDDR is going to be training. It doesn't have the, without the HBM. I don't think it's going to compete on the, on the large training opportunities, Pat. But I think this is an inference machine and I think what Qualcomm is good at is efficiency. So if you look at what Qualcomm can do on a smartphone and what they're going to try to do in the data center and people will go, well, they didn't succeed in the past. Okay, that's fine. But AMD had some, some bits and starts with its data center work and by the way, Nvidia wasn't until it was so like let's, let's get over the fact of what things were. I think it's a different company with different IP, different engineering, a different opportunity. But I think Pat, 1 or 2%, I keep saying win 1 or 2%. Like this is all Qualcomm has to do to drive a $10 billion diversification to their business. And I think with these Middle east relationships, I think alone and this, by the way, you know, even like Grok how they've been able to succeed. Like there's something in that market where there's companies that just want these inference machines, these token machines. And I actually think there's a story of Qualcomm playing their cards right, that they are complementary to Nvidia, they can be complementary to the big training clusters and they can just say, look, we're just adding inference capacity and we're adding massive lower cost, higher efficiency inference capacity. I think that story could land, I think they could land that story with some of the NEO cloud, some of the, some of the hyperscalers, some of the Middle east deployments. Depending on policy in China, I'm sure they would be able to be successful in China if that ever opened up again. Qualcomm's huge there, so I think there's a lot of different plays. We are, God, we are going slow today. Let's hit the next one really quickly. Pat, you mentioned that Nvidia supercomputer, you know, announcements, but the DOE was busy this week. Another billion dollar partnership to make two more supercomputers with amd. This has been AMD. The word I use has been a juggernaut in high performance compute. Chris Wright has been very busy. I believe they got a big, he got a big shout out from Jensen, he's getting one from Lisa. And this is all about basically winning science, winning science in the United States. They want to, you know, everything from nuclear power, fusion science, drug discovery, energy, we want to win in these categories. The government is making investments and amd. You know, my quick read here continues to prove that it's, it's the company to beat in this category.

‍

Patrick Moorhead: Yeah, I got invited to the event on Monday where Lisa Sue was speaking. Unfortunately I couldn't go. I'm really interested. This makes a lot of sense to me. Primarily given AMD has a 64 bit precision part which if you are doing flops versus tops, it's the most important thing. You can't do supercomputing with 4bit. What is happening is there is a transition to add a lot of AI, hence what the big, you know, the Seven supercomputers that Nvidia announced. But no, it's very good to see, I'm interested to see for earnings what the impact is to AMD financially and.

Daniel Newman: Does the market care. That's another big thing. Like it just feels like right now and we get to earnings, we'll talk about like Amazon, these companies, like it's all about the data center right now. I mean this stuff's really interesting. I just don't know how much it, or how much it computes. Speaking of companies that reported this week, we're not going to hit the earnings just yet but what about the job cuts at Amazon? What do you, what do you make of it?

‍

Patrick Moorhead: Yeah, it's interesting. I, I, I still am not convinced that every one of these job cuts is because of AI. We hired so many people during the P word that I'm not going to use because you know we get filtered on it. But I still think we hired way too many people and we will see what the real reason for those layoffs is. I know you're bullish on the AI layoffs part.

‍

Daniel Newman: I don't think bullish is the right word. I just think I'm honest about it. At least honest in my own mind.

‍

Patrick Moorhead: What's more interesting to me is just when people said that Amazon was out and I have been saying absolutely continuously AWS, you know, announced anthropic is going to use 1 million trainiums. Right. I knew that they had some big announcements. I think this gives more credibility. It's so funny. I saw, I see so many people particularly on X dishing it out to AWS and I'm always thinking you people don't have any idea what you're talking about. And I, I don't think AWS is done yet. I mean we've got to reinvent, we've got Reinvent coming up and oh by the way, the largest build out company in the next few years is Amazon. What customers and what use cases are they building those out for? We will see.

‍

Daniel Newman: Yeah, it's, it's interesting. I think that they are replacing dollars meaning tokens are our productivity. And so what I do think is happening is it's becoming like we're thinking of it all as Capex, but it's actually they're thinking of it as opex. There was another note that came out this week that talked about how basically Amazon wants to automate like 75 of its workforce. Okay. So they've been very upfront about what they're doing and what their Intent is. Having said that, I think you're also right. Every. I think Jassy actually said to himself that this was Culture. They want to flatten the organization. It got too steep. We both work with Amazon. I think we could both probably add some comment to the bureaucracy that's been created as it's become a larger company. Hard to get things done. And so I think that that could be smart. And by the way, I've learned from leading business businesses long enough that sometimes you have roles that you think are really critical and then you eliminate them and you realize that nothing changes. I mean, just like literally there's a. There was. It's not personal. There was a person in the middle that, like, literally everything still gets done. It's like, whoa. So I think when you get a company to their size, they have to kind of keep looking at them. By the way, the market rewards this. So he has to be, you know, he has had the lowest performing of the Mag 7 stocks this year. At some point, they have to start returning to shareholders. So he's got a job to do. He's got to balance that. But culture, this could be good for both. Culture, this could be bad for Culture. As a quick note, though, there was also that anthropic Trainium news. That was great news. You know, I'm, I, I continue, though, Pat, to just say this very simply. It's not entirely yet an indicator of success for any chip company that sells out right now. It's all hands on deck. Every chip that can be built is being sold. This, by the way, is why I said Qualcomm will sell everything it builds. Intel, we heard this last week. Even their CPUs for XEON and older nodes are selling. It's like, because. Because cloud companies are looking at how to improve their footprints of CPU data centers right now. And it's like every chip that can be made is being sold. Having said that, I think you tweeted something like, don't rule out Trainium. I totally agree with that. Multi generation as we know it took. What did it take? TPU about three generations. Three, four was where it really started.

‍

Patrick Moorhead: That's right, Joe.

‍

Daniel Newman: So we're right there at that point now. So obviously Anthropic went all in on TPU, it's going all in on Trainium, it's all in on Nvidia, because they actually just can't get enough compute. So it's part of when. To me, Pat, I don't know what you think. It was partially a win for me. It was also partially just an indicator of how much demand there is for all this. Every chip that can be built.

‍

Patrick Moorhead: Yeah, I already talked about it. You were texting somebody when I was talking about it.

‍

Daniel Newman: So did I do that? Did I double down?

‍

Patrick Moorhead: No, I've already, I've already commented on anthropic.

‍

Daniel Newman: You know what? Sometimes, sometimes, you know, I zone out. 

‍

Patrick Moorhead: Let's hit this, let's hit this.

Daniel Newman: How was my, my take? Was it pretty good?

‍

Patrick Moorhead: Yeah, it was.

‍

Daniel Newman: The audience got it twice.

‍

Patrick Moorhead: So similar. Yeah.

‍

Daniel Newman: I'm only sure I want to show when the audience twice, they, they really lean in. All right, quickly Pat, you did the Google public sector event. What's hot?

‍

Patrick Moorhead: Yeah. So Google, particularly related to military stuff, they really hadn't leaned into. In fact, until TK came. They weren't doing anything there. And I gotta tell you, they absolutely brought it. One of the best examples I've seen of agents. Gemini for government. So let's say, you know, we're, we're or at. We're using OpenAI and we tap something in. You pull them down. What data do you want to use? Is it email, is it SAP? And then you go in and you create an agent. I created an agent in about seven minutes on this and I couldn't believe it. Now they say that they're going to send me some stuff where I can do that. But it's really good. Like it's really good. And Google is one of the only companies to really lean into on prem with GDC. Lockheed Martin, right, is doing an on prem air gapped AI using Gemini. I don't know what the hardware is, it's probably Dell, but I'm really liking what I'm, what I'm seeing with them. The last thing that they had in the keynote was Joint Chiefs of Staff, four star general reports to Trump talking about the mission. And that is so not Google. And I loved it. I loved it, I loved it. I met with the CEO of Google Public sector. Check out the three videos I shot there as well.

‍

Daniel Newman: Appreciate you doing that. By the way, next week I'm gonna go solo while you're off in Europe gallivanting about being fancy. But I do appreciate you taking one for the team on that one. I'm sure they came out great. I'll just mention that there was a mini Azure outage. Nothing like the AWS one. Don't think we need to pay a ton of attention, but I think watching these is gonna be an interesting thing. You had tweeted a few times when you saw some sites struggling. But we've got a lot of earnings to cover, and we still got to do a quick flip here. So here's what we're going to debate, Pat, is that, you know, the government's taking stakes. There's a lot of people that say that's not capitalism if the government starts taking stakes in companies. There was a thing about Quantum a week ago. We obviously know what's happening in AI, Rare earths, possibly other energy plays. So should the government take it. Be taking stakes in the future of technology and the future of energy and other critical industries? You all know me. I'm, you know, I'm a pure socialist. Pat. I want government control over everything. I just do not think that capitalism works, and I do not think the people have the ability to build businesses. We must depend on our government for everything. Okay, that's complete nonsense. But what is not complete nonsense is in every era, there are certain industries that are just too important to fail. The US Government has always had ways and means of making these kinds of participative actions and contributions into industries that are critical. But in the past, we've done it in ways that are actually, in my opinion, more socialist than the ways that we could do them. What am I saying, for instance? Well, the way we do grants, for instance, we give grants to companies, and then we never ask for them to repay, so they get to use grants for research and things that we think are important industries. Maybe it's AI. Maybe it was Quantum. Maybe it was APIs for pharmaceuticals. Maybe it's rare earths and fracking and drilling for energy. And we basically just give this money away. We give it away, no strings attached? We could say there might be strings, but there's never strings. No one ever pays Uncle Sam back. And they get to basically use the taxpayer money to build their enterprises at scale and work on projects that oftentimes end up looking like the rail system in California that's built, what, 10ft of track for $1.5 billion? But the difference here is that these are existential. So let's just start with something like AI. Leading the world in AI means leading the economy, global economy, for the next three, four, or five decades at the very least. We don't even know what superintelligence and AGI might look like. But just AI alone is going to control the future of the world. This means energy. This means the data center. This means the infrastructure. And by the way, this means the ability to manufacture leading edge chips. So taking an equity stake in intel when you have no other option in the United States to build leading edge AI chips was the necessity. We had to show that we would back just like TSMC did when it got government funding early on and has been backed not only dollars given to TSMC but a number of other subsidies. We needed to do the same thing for Intel. The CHIPS act did not work that well. And the bottom line that makes this a very simple decision is returning value to shareholders. Now where I don't agree is the government should not take a stake and take a board seat. They should not have decisive controlling action over these entities. And that of course will always be in question. But the way we've done deals so far, the government is not actively taking a stake in the sense that they're going to sit on boards or drive decision making. But what they are saying is, hey, we want to put dollars into the critical industries that we believe are important for our long term survivability, success and prosperity here in the United States right now. Every one of the bets they've made so far, rare earths, AI chip manufacturing and also in quantum which could be existential with things like breaking quantum encryption that could cause massive fraud and issues within the financial systems. Are all critical industries doing the same thing in categories like drug development, cancer research? I also believe the government should take a stake, but we shouldn't do it through grants, we shouldn't do it through tax subsidies. What we should do it through is investments like a sovereign wealth fund that backs companies that can help us get to where we need to go. And then when these companies succeed, the government has a better balance sheet, has the opportunity to cash out on that success, reinvest those dollars and return value to shareholders. It's very simple. We really aren't doing anything different. We're just doing it smarter now.

‍

Patrick Moorhead: Oh, my little commie bestie. That's so cute. I love it. Yeah, let's just nationalize everything while we're at it, you know, Love it. So no, no, seriously here, I think in markets that are oligopolies or monopolies like foundries and rare earth, it makes sense. There's literally two rare earth metal companies, material companies, and there's one US foundry that does leading edge with IP and that's Intel. That makes sense. You have a tremendous amount of companies that are doing great Ionq, doing great cutting deals left and right and driving, you know, real revenue with not just research grants, but from auto companies as an example. I mean heck, you attended IBM's Quantum event. They're certainly not, not struggling. So who would, who would the government actually invest in right? If, if, if, if they were and, and my God, your, your choice of getting it wrong when, when you have in, in the mix here. Oh my gosh, you know what if some, you know, cubit special cubits are probably four different cubit standards out there. What are you gonna, what are you gonna invest in? You're gonna invest in all of them. So, and invest in all of them. You're really not investing in any of them. So I, I think we should be doing like, like we do with science. This should be part of the Science act, not the CHIP act, right. Tax credits, price challenges, procurement pilots. So equity certainly is the best way to buy a seat at the table. But picking who you would invest in, you're going to be, you're going to be wrong unless you invest in 10 companies. And then, you know, what do you actually, what are you actually achieving with that? So, I'm done winning. Let's move on.

‍

Daniel Newman: Do you believe that? Was that your real take?

‍

Patrick Moorhead: Yes. Yeah. I don't think the government, I don't think the government should be investing in, in, in this.

‍

Daniel Newman: Yeah yeah.

‍

Patrick Moorhead: I mean we're so far behind, so far ahead of the Chinese on this. It's not even funny. On which AI on, on Quantum.

‍

Daniel Newman: Oh well I mean obviously they have potentially to be able to filter through them for sure. I still think I like it, I prefer it with intel getting a return. I just, you know the grants money of the CHIPS act was all just going to get burned and only potential winners would be chip would be shareholders of the company including it could be the U.S. but yeah, I think that it's the problem is, is the modality that people think we don't do this now. I think there's a lot of unpacking DARPA and different ways that money gets filtered to companies that we think are interesting that never gets repaid. It's all tax money. Anyways. We've got a ton of earnings this week. It was a super mega killer earnings week and you know, we just had a GTC in the middle of it and a bunch of other news to splash. But Pat, you want to do it.

‍

Patrick Moorhead: Little bulls and bears, let's dive in, buddy.

‍

Daniel Newman: All right, well, I'm glad there weren't very many earnings. Let's just do the super fast review of two, two items and then let's hit the earnings things.

‍

Patrick Moorhead: Okay. I want to make sure you've got the time.

‍

Daniel Newman: Yeah, that's why I'm going really fast. Fed cuts rates. No surprise here. Powell basically said Data center like 55 times on the call. Basically said it's not a bubble, it is a secular thing. Which is interesting from the Fed. That's my interpretation. Not his words exactly. Did talk about job softness and weakness and potential risks related to the job. And also threatened that they might not cut rates further, which seemed to create a bit of a consternation. But also said they're going to end quantitative tightening which means the, the, the, the printers are going to come back on soon. Guys like crazy. We're up this high and we're talking about printing money again when inflation still hasn't been beat entirely. But I think we all agree at this point we're just going to try to outgrow it. We talked about the open AI valuation, the trillion dollars. Pat, I don't think you and I have much to say. I think it's insane. Like if, if, if open AI is worth a trillion dollars and Nvidia is worth 10 at least. Like there's no way that they're worth a trillion because you remember Meta is only worth like one and a half.

‍

Patrick Moorhead: Yeah.

Daniel Newman: Do you have, do you, do you think a trillion's crazy?

‍

Patrick Moorhead: No, it's completely bonkers in, I mean can you imagine the pe? They don't make any money.

Daniel Newman: You can't, there is no pe, there's no earnings. It's, like, infinite. Yeah. Well then like what's Google worth?

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: Anyways, it just, I can't like, I'm the most bullish guy I know. Maybe. Other than your friend Dan Odds that you were on with yesterday. Nice job. I'm the most bullish out there. I'm the most, not a bubble guy out there right now. Like, but at the same time, like if there is a bubble that is, what is. It's like open AI. Yeah. Drooling. I got so excited that open AI's value is a, is they. You gotta make a dollar. I mean you can't be worth a trillion if you can't make a dollar. Pat, that means your company's like 30 trillion at this point in terms of how much more money you make than he makes.

‍

Patrick Moorhead: Okay, can't wait.

‍

Daniel Newman: So Nvidia got 35 trillion. We've kind of hit on that. Any, any, anything interesting about that to you? Just, you know, you know, I think we hit that pretty well at GTC.

‍

Patrick Moorhead: I think we hit that already.

‍

Daniel Newman: All right, well then let me just, let's do the mega rundown of earnings. I'll hit, we're going to hit all the, the, the mega mag sevens. But first just ServiceNow. Pat.

‍

Daniel Newman: ServiceNow. I did talk to Bill McDermott. Another rule of 50 quarters. So while enterprise SaaS and enterprise software has been beaten up to death, ServiceNow just continues to perform their steady, consistent, strong performance. You know, fastest growing. The only scale enterprise software company that's breaking Rule 50 right now. So a lot of credit to ServiceNow for doing this. Their generative AI number is only like half a billion right now when they talk about that number. But Bill said something really profound to me. He said, you know, effectively we are workflows, company workflows, automation. So effectively every dollar of our revenue is actually enterprise AI. And that's why I keep saying that I think the repositioning for ServiceNow sitting with companies like Palantir, Palm, possibly even with IBM where their software kind of sits, is that they are the enterprise AI software platform companies now. And that's where these companies all want to be. So a really good quarter from ServiceNow.

‍

Patrick Moorhead: Yeah. Every CIO conversation I've had over the last 18 months, pretty much just narrowing in on workflows is the common denominator of what to go in and fix. You can create something brand new or you can use a company like ServiceNow who has done workflows for years. Many HR workflows are already top end by ServiceNow connecting systems like Workday and SAP and add that to items. So the difference here is even though there's only a half a billion dollars in AI revenue, it's, it's the, they're doing so well in their other businesses and that's, that's, that's a two bagger here. So the new areas that, that they're, they're running into plus AI is, is really the story for this company.

‍

Daniel Newman: Yeah, yeah, positive Pat. Okay, Alphabet Google. For those of you that don't realize it's Alphabet Google, what do you think?

‍

Patrick Moorhead: Ads in Cloud beat debunking the whole AI cannibalize a search meme, at least for now. I'll admit I am surprised right that they keep the ads going plus AI overview. Congrats to those guys. But when you've got the founder coming in and coding and pulling together the team, I mean, that's pretty freaking, freaking awesome, right? Cloud up 34%. Backlog up 49 billion in three months to $155 billion. Remember when there was discussion before TK came in, on Google Cloud going away? I, I think everybody is hiding about, about, about those statements. So TK and the team have done a great job and let's not forget that Google has been doing AI bigger than every other company, maybe with the exception of Amazon that did a lot in machine learning.

‍

Daniel Newman: Yeah, absolute smash out of the park beat from Google. Great result. The market loved it. Victory lap. Victory lap. Victory lap. I've been, you know, sometimes you laugh at how much I've been pounding on the Google train, but I just never believed and maybe it's the millennial slash Gen Xer in me that hasn't fully committed myself to chat GPT yet and I just don't see the I, I don't see and the numbers now are continuing to indicate that it's just not fatal for, for them and so great job. You've been very, very consistent. Let's talk about meta. Meta was a bit of a weird one because they had this one, they had this one time tax deal non cash and their earnings instead of being like 6 they were like a $50 but they actually would have been 7 they would have beaten if they didn't take the one time tax. But this one was all about Mark's capex. You know I guess there's a little bit of a difference when your capex is all spent on your own build as opposed to building capacity for others. But Mark has said this quarter after quarter after quarter and this is the easiest way I justify it is he'd rather overbuild than underbuild. These CEOs by the way every one of them raised Capex, Google, Amazon, Meta, Microsoft, they all committed more capex because they all know it's existential for these companies to win in AI, to win, to have the technology to be able to deliver to capacity. And by the way every company that hasn't been able to pay including like Amazon early on under building, look what happened to them. They're finally getting out of a hole that took them multiple years because they committed early. These other CEOs are smarter. The market sometimes whiplashes you a little bit. I think that the dip will be gobbled up. They issued a bond by the way. A $25 billion bond.

‍

Patrick Moorhead: Yes, yes. I think your thesis, your thesis was that hey, I think somebody knew about this.

‍

Daniel Newman: Well these are the things anytime you do an issuance, generally there's a sale it creates. Even though these people get paid first before the shareholders get paid and generally speaking when that happens it creates that negative. I bet you over the next few weeks you'll see a lot of that again. Assuming the rally continues and the bottom doesn't fall out, you'll see a lot of that gap get filled. But I, I, I think Pat Meta is the user of AI. They make freaking fortunes using the technology. So when everyone's like no one's using AI except ChaptGPT, it's like Meta uses AI every stupid ad you see every time that you go on there and the thing you and I are talking about peptides and now all I see I'm going to Facebook are peptide ads. Like this company is using AI to make everything super personalized. It's a wonderful example of how AI has infiltrated our lives without our permission and is being used by mega companies to make mega bookoo dollars. Anything any add on Meta? 

‍

Patrick Moorhead: No, I mean listen, from a cost basis they're just uncompetitive related to Google who's rocking the TPU for every internal workload. Right. Mtia is a nice start. I like the way you talked about what was going on with Trainium if you need four or five turns to crank out something exceptional. Ameta did an Aqua Hire and you know, I'm hearing MTIA 5 is, is even, you know, better. But my guess is they're also buying a lot more GPUs than Google is to be able to accomplish what they need to accomplish. And rule of thumb that I'm hearing from Daniel, is an XPU is about 30 to 50% less cost the same at the same performance level, albeit a little bit less flexible.

‍

Daniel Newman: Absolutely. All right, we got a couple more. We are getting close to our time Pat, but we got to give some credit where credit's due. Microsoft.

‍

Patrick Moorhead: Yeah. So absolutely crushed it with Azure AI contributions actually rising. And here's another company that's using AI. This is not just for OpenAI, it's for Copilot that they're monetizing Crpu uplifts a big medium term driver here. Security and Windows look really good. I was really surprised that the stock went down afterwards, particularly in the very good way that Microsoft talked about how they were looking at Capex which was basically we're not going to add Capex unless we absolutely know that we can monetize this long term and then the stock goes down. Didn't understand that Pat.

‍

Daniel Newman: You know the down and dirty was that quarter was freaking amazing. Like absolutely 40% Azure growth.

‍

Patrick Moorhead: I know.

‍

Daniel Newman: Order book, you're giving out two or four. They say $400 billion in visibility in orders and these orders in the book. Not, not even like what they think they are, this is actually. Forget it. Visibility is the wrong word. $400 billion of booked business, not even including things they think they're going to book into the future. They're building out capacity fast. My biggest beef with, frankly, Pat, is just that they need to speed up their own infrastructure development. They are going to get margin hit by the fact that they are 100% dependent right now. They do not have the competitive answer to Trainium and TPU. And I think that's gonna cause some strain on margins down the line. We'll see. But that's the area I, I think they need to have, they need that choice, that selection.

‍

Patrick Moorhead: I said the exact same thing two minutes ago.

‍

Daniel Newman: Yeah, well, I'm just agreeing.

‍

Patrick Moorhead: Okay. All right.

‍

Daniel Newman: Sorry.  Am I allowed to not have the same opinion as you?

Patrick Moorhead: Of course you are. I just.

‍

Daniel Newman: That wasn't even me zoning out. I heard you say it. I'm just doubling down. Like when people wanted me to nitpick. People want me to nitpick it, I had to nitpick it. You know, I nitpicked it when I talked to them about it too.

‍

Patrick Moorhead: I'm like, by the way, hey, I just wanna, I wanna say that Brad Gerstner between, like as he's talking, he's looking at his phone. It was amazing. Like I thought that was just people like you, but maybe it's just rock stars and maybe, maybe Daniel, this portends to you becoming a billionaire someday.

‍

Daniel Newman: I'm just trying to cross the, you know, the 250 line for now. Cash, cash, cash.

‍

Patrick Moorhead: Like I don't want $250,000. Right. Network.

‍

Daniel Newman: 50 million 250. I want to be worth 250 when I grow up. All right, two more quick ones and Apple. Pat, you went on CNBC, talked a little about Apple yesterday. It kind of did a quick dip and then it did a dip and rally.

‍

Patrick Moorhead: It did. And guess what? Apple's down today. Imagine that they listen, they internalize the call and they are one of the only markers that that is down today. And, and they're down big. I called it. Sorry, Dan Ives. Sorry, sorry about that.

‍

Daniel Newman: And is it called because of the low iPhone number? Was it the low iPhone number?

‍

Patrick Moorhead: What they are doing is their investors are piecing together the low iPhone number. What? Tim Cook said that he should have been able to ship more but supply chain challenges in the same breath said hey we're going to have an amazing holiday quarter. I think they're piecing it together and investors just didn't see enough evidence of what were those supply chain challenges. Why didn't they put more in there particularly when it came to tariffs. Like I'm wondering if they have India iPhone issues in there but yeah, I mean they crushed it. Services are just services that are 60% of the iPhone revenue Daniel and that is absolutely amazing. Like I said on the show John asked me what are the growth areas. I think sustained cheaper max plus buying a movie studio is I think the future for Apple. Nobody cares about Apple intelligence. I will cop too. I did make a mistake on Apple intelligence. I did some post research on this. Maybe 20% of the phones in the installed base can run Apple intelligence. I thought it was more but it is low. I still have, I'm very still skeptical if Apple can pull off something that's truly compelling for the brand new Siri.

‍

Daniel Newman: Got it. Yep, yep, yep, yep, yep. I don't have much to add on Apple. I came out, I said the low iPhone wouldn't sit well and then I got a lot of from people because it went up I a minute later I still think when you have a new launch I know it's not their main iPhone quarter but like when you have a new launch you want to see a strong reaction. It seems like they had one product do well and the rest of that new lineup has not done particularly well. Yeah that was kind of the overwhelming channel checks and everyone came into it. They missed the number by a lot. I mean it was like over 49. It was like over 50 million was the expectation or whatever and it was like they missed by a whole million point something units which is a lot of units. 

‍

Patrick Moorhead: You know people might say Daniel. How do you have supply chain issues in the first week? Like how does that happen? Right. It was shipped for a week for the quarter. So I, I totally just don't understand that.

Daniel Newman: Well that and that's what I mean that like that was the strongest selling week.

‍

Daniel Newman: Yeah. Of the new device because that was the first week of shipments it should have been like the blowout week of getting stuff out. But next quarter I do expect it will be better for iPhone we'll see but there's a lot of reasons to believe that last one Pat Amazon look this was the one I think that was on because what happened was is like after Google Microsoft Meta especially meta kind of spooked everybody on the more capex I think people wanted to see like was it all share shift with Amazon having the biggest cloud? Was the share shift coming over from Amazon to Microsoft and Google rather than actually sort of Amazon growing? Their numbers had been down into the teens. They're growing at half the rate. Granted they're bigger, a lot of people kind of struggle with those two things but they're growing at a much slower rate. They needed to be over 20. I kept saying they need to be over 20. They gave 20 like 20.2 and this is the first kind of re-acceleration of the aws business since 22. So it had been slowing and now it's re accelerating. It seems to me that some of the capacity Project Rainier came online. They needed to show that they'd kind of overcome the hangover from the slow start with Nvidia the Trainium success. I think they said they're 150 quarter on quarter with it sold out. Was it through 26 parts or 26? Basically they are all there so these are some good tea leaves that they're starting to turn the corner. And of course the commerce business overperformed. The ads business is doing great. But Pat, to me, discord was all about AWS.

‍

Patrick Moorhead: It's amazing to me that people really discounted AWS here. We truly believe that we can't get enough capacity. Why on earth would people have said that it wouldn't have been AWS? There was a piece of research that came out of a company called Data Center Hawk planned self build expansion in terms of gigawatts. Guess who's number one by a. By an amazing amount. It's AWS at 28 gigawatts, Microsoft at 21, Google at 12, Meta at 11. Okay. And you know, I'm gonna do a victory lap. I was right and all the bears out there about Amazon will were wrong. And we'll see what they have up their sleeve.

‍

Daniel Newman: Yep, absolutely. And Pat, we went long, we went strong. We're missing important nonsensical meetings within our businesses right now to be here with you all. Hope we appreciate that it's Friday here but it might be Saturday, Sunday, Monday. We don't know because we don't know when the producers are going to get this thing out. But having said that, what a week. Attack. We appreciate everybody being part of our community. It's been a heck of a month Pat. Put October in the books. It's Halloween. I'm dressing up as a gym bro tonight because why wouldn't I?

‍

Patrick Moorhead: Yeah, buddy.

‍

Daniel Newman: Stay strong, stay long, travel safe. My bestie and everybody out there, we appreciate you. We'll see you all later.

‍

Patrick Moorhead: Bye bye now.

‍