The Six Five Pod | EP 287: The AI Cloud Shake-Up: AWS’s New Silicon, NVIDIA’s Strategy & Apple’s Innovation Question

This week, the AI industry delivered a cascade of breakthroughs—and a few warning signs. From AWS unveiling its most powerful silicon yet, to Apple confronting a talent exodus, to misinformation shaking the markets, the tech landscape is shifting faster than ever. On this episode of The Six Five Pod, hosts Patrick Moorhead and Daniel Newman unpack the announcements, strategies, and tech news stories that made headlines this week.

The handpicked topics for this week are:

1. AWS’s AI & Silicon Leap: AWS kicked off re:Invent 2025 with major hardware news, including Tranium 3 Ultra servers and the 192-core Graviton5 processor. By integrating CPU, DPU, and XPU technologies on a single card, AWS is signaling its ambition to lead the next era of AI cloud performance.
2. Enterprise AI’s Next Chapter: With 95% of enterprise data still locked behind firewalls, we’re only scratching the surface of what AI can do for business. Salesforce’s 3.2 trillion tokens consumed on AgentForce shows how quickly AI adoption is scaling for real enterprise workloads.
3. Apple’s Leadership Challenge: A growing executive exodus raises important questions about Apple’s long-term innovation trajectory—especially in AI. Pat & Dan discuss how this talent drain could shape the company’s next chapter.
4. Market Misinformation Fallout: A wave of inaccurate reporting recently triggered significant market reactions for Microsoft. The incident highlights how fragile the information ecosystem has become—and why responsible journalism matters more than ever.
5. NVIDIA’s Ecosystem Play: NVIDIA’s partnership with Synopsys reinforces its commitment to optimizing the full chip development lifecycle. It’s a strategic investment aimed at strengthening its lead in AI and robotics.
6. Marvell’s AI Momentum: Marvell reported impressive results, with 37% YoY growth in its data center and AI infrastructure business. Strong cash flow and a focus on connectivity position the company well as demand for custom AI silicon accelerates.
7. Salesforce’s AI Demand Signal: Salesforce continues to emerge as a bellwether for enterprise AI. Surging AgentForce usage and massive token consumption metrics reflect real, measurable demand for AI-driven business transformation.

‍

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: I'm so glad that it's no longer, I don't know, do we know? I'm sure that just compelled the audience to want to listen to the rest of the show. I know. Me too. Well, welcome everybody back. But, is that your thing? Welcome everybody back to the 287th edition of the Six Five Podcast. I wanted to start this one out differently. I just couldn't do another start out the same. Hey everybody, we're back. It's 287, monotonic. We are gonna be passionate today. We are gonna be interesting today. We are gonna talk about all kinds of things that no one's talking about, AI, robotics, GPUs and TPUs and Tranium chips and oh wait, everybody's talking about that stuff. But that's what we do here, Pat. We talk about what's going on in the market. But first and foremost, I just want to ask you how your biceps are doing today. Are they bigger than they were yesterday?

Patrick Moorhead: Doing really well. You know, I've had some really good lift lifting days for the last four or five days. I don't know what's kicking in. Maybe the hormones are, you know, finally coming together. I've got that together, but we will see. I mean, listen, you can have a good workout day. Everything else sucks and everything just just just comes together, Daniel.

Daniel Newman: Yeah, no, it's good. I'm glad to hear that. I think we worked out a couple times this week, bright and early on the West Coast. Good. It was so good. Yeah. Hit a chest day, hit a back day. I know. It's great. You know, three days, a big part of our week we spent at AWS reInvent. It was good after the long holiday, I got out on Monday, pretty early start to the day, but it was a big event, a big one every year, Pat. And we are getting closer and closer to the wind-down period where I think you've got a couple of nice vacations planned ahead, which you've earned after a heck of a big year. And, you know, get that end of the year wrapped up, which I can't believe we're already in December, dude. I cannot believe it. We are already at the end of another year. We're gonna be heading into 2026 soon. Oh man, dude, crazy times. But hey, we got a lot to talk about today, Pat. We got a big show. We'll talk about reInvent. We'll talk about some strategic alignment news. Of course, some strategic AI stories dropped this week. We'll talk about Apple not being able to keep any AI people employed. What the heck is going on at Apple? We'll talk about an information report that somehow got headlines. At some point, I don't know how many more times they can get headlines and then refuse. But how can we not talk about that? And then there's a few other things of time allotted we might be able to get to with acquisition by Intel. And then we had teams from all over the world this week. So if we have a few minutes, we might talk about some of those things as well. We'll have a normal flip. This one will be another debate for you to beat me at each and every week. Same story. I appreciate that. And then we'll get to markets because there's a lot going on there. And despite the fact that the year is winding down, there is a ton to talk about in the markets, Pat. But we got a lot here. I think we should just go right into our decode. All right, Pat. Set the stage.We were in Vegas. We were at reInvent. What was your take?

Patrick Moorhead: So this one's going to be tough because there was so much news, so many announcements out there. And I'm going to, you know, I want to dive deep on a few things. The first thing I want to dive into is chips, because we love chips here on the Six Five Pod. You've got the new Tranium 3 Ultra servers, went GA out there. And if you are in the AWS ecosystem, it's absolutely a good choice. Per-chip performance is higher than TPU, but TPU is a full scale up and scale out system. So the aggregated performance will be higher and training three got a slower start out of the gates than the TPU. But they are very much in the game and you just have to look to the $10 billion opening ideal if you need any reinforcement of that. It's not that they're just going to use Tranium. They did talk about using GPUs, but I find it hard to imagine. After all the action that Anthropic has and stuff they've done with Tranium 2. And you look at Project Rainier. So yeah, AWS is very much in the game. At the very end of the show, they announced another chip called Graviton5. And Graviton5 is really an absolute beast. I mean, 192 core processor, folks. Okay, you got a question, you know, even the need for a two socket design, although there are for some scale up applications like database and stuff like SAP, but 25% higher compute than Graviton4 using Arm Neoverse 5. and a lot of just great stuff in there. I don't know why they saved this for last, because quite frankly, competitively, it's the most differentiated chip that they have out there right now. They moved the needle. 55% of all new CPUs inside of the AWS ecosystem are Graviton. And the other thing, which was really cool is if you, you've got an entirely, um, AWS, um, ribbon, uh, rack, um, when you look at nitro Graviton and tranium. So I need to get underneath some of the details. They said that Graviton was more than a head note. It was integrated. and I couldn't help thinking about, you know, Grace Blackwell type integration, but I gotta do more research on that. So I'm gonna take a breath, Daniel, and give you some oxygen. We'll just bounce this back and forth.

Daniel Newman: Yeah, for sure. I mean, you hit a lot of the silicon. I thought it was very impressive seeing CPU, DPU, XPU on a single card. I think they're making strides there, or they're heading directionally there. Your point about Graviton, I mean, I kind of say, I tweeted this, Pat, I said AWS is to the CPU, Graviton is to the CPU, what Google is to the TPU. I mean, where they're at in their cycle. And let's not mistake, I mean, we need a lot of CPU compute to go with all this AI acceleration. So that's a really good promising technology piece of the puzzle that they've solved. And others are very early in that. They're very early in their custom CPU chips. I think that a big part of what I saw though, and I just walked away, Pat, just feeling so confident in my outlook for AI. You walk around, you talk to that place, there is no slowing down. And I'm talking about at the enterprise level. There was a data point this week that came from Mark Benioff from Salesforce that there was 3.2 trillion tokens consumed on AgentForce so far. People probably don't realize, but AgentForce has barely been adopted yet. it's so early in terms of the pivot from CRM to agents at that company, which is a pretty big bellwether for applications. A lot of the focus of AWS was all about making development of building agents, building AI applications, using generative AI, building frontier or custom models that you can use for your enterprise. This whole 95% of enterprise data still sits behind the firewall, meaning most of what we're seeing, I call these LLMs, Pat, they're like parlor tricks. They're like the first phase of what we can do with AI. We've barely scratched the surface of using AI at the enterprise level. And AWS has a really good opportunity there because they are still the largest cloud from the first compute era. Now there's some debate about whether they're the biggest AI cloud or the biggest GPU cloud. where they will land in the next era. But they have this massive, almost $140 billion business. They, by net revenue dollars, are growing a ton. I get it's not as high as a percentage, but it's a much bigger base. And they have the opportunity to get more and more of those workloads, more and more of that data that sits in their storage buckets and in their SageMaker, ML tools and stuff to be utilized for the future of AI applications. And so I think this show is really about them making their prove-it moment there, development environments, tools for streamlining agent development, tools for building models and customizing and tuning models. So they've got the infrastructure, they've got a lot of the tools needed. And Pat, we've sized that market, the tooling market, the AI platforms for that could be as big as an $800 billion market, we think, by the end of the decade. Because that's how much companies are going to have to spend to actually get their AI efficiency and productivity out of the data. And so there's a big opportunity here. And we think that AWS did a pretty good job of showing their wares. But it's very competitive. And the prove it moment is going to be showing up every quarter in their numbers.

Patrick Moorhead: That's right. And a couple other things that caught my eye AWS is calling these frontier agents, which is kind of a spin on an agent that's scalable, autonomous, works for a long time, and they brought out three in particular, an autonomous agent for software called Kiro, a security agent, and a DevOps agent, three of the ones that you would expect AWS to bring out first. Interestingly enough, they also have introduced a product that is really, I would say, a one-for-one competitor to Copilot, where essentially you go in and you pick the data sources you want. You can create agents on it. It can do things for you. I might actually sign up and kick some tires. It has like 500 connectors. And that's been the biggest limitation I've found so far when I've tried to set these up is the lack of connectors. And as an example, ChatGPT doesn't connect with OneDrive, right? And if all your files are in one drive, you're, you're screwed. Right. Um, and on the other side, you know, co-pilot doesn't connect, uh, to Google drive. Right. And, and so, so that's a limitation, but at least from what I understand, those limitations aren't true in the AWS and AWS, uh, world, which is, uh, which, which is, um, pretty good, uh, transform. AWS Transform is another one. You know, I've been hot on agents taking enterprises from legacy environments to modern environments. And, you know, whether it's VMware into, you know, EKS or something, and I'm not calling VMware legacy, some people will. Taking, you know, IBM code into Java, right? COBOL into Java. But they have now, I think, 10 different examples that they gave. I think this was called QDeveloper before, but they renamed it into Transform. I like the new name better. But I really thought that was clever. Final one I want to put in here is AWS has truly gone sovereign, and sovereign in the way that I define it. Sovereign for AWS used to be logical and not physical, but what they're doing is actually putting AWS infrastructure. I'm trying to figure out the difference between this and Outposts, but a lot of research to go. I think Oracle always had the cleanest version of this. It was called Cloud at Customer. Literally, it was copy and pasting on-prem, sorry, hyperscaler, infra, and you put a chain link fence inside of the customer. You could run all the Oracle apps on it. You can do one rack as small as that. So my team's doing a lot of research on it, and I'm sure yours is, to see the flexibilities and the gotchas. But I I don't know if this is real yet. I'll be honest with you. And what I mean by that is I've seen this story before. AWS came out with Sovereign. It was logical, not physical. We've seen Outposts. And we'll see how much investment gets put into this and how much they could take. But what absolutely shows Is that on prem infrastructure is still required particularly in this age of an AI sovereign cloud that i think was the second biggest talk of the town in davos when we were there.

Daniel Newman: Yeah. And then there's a lot more here. Did you take a bite of food? Right. Just like naturally as you were doing that.

Patrick Moorhead: I took a, uh, it was a, um, nitric oxide tablet.

Daniel Newman: Oh, very cool. I mean, like you didn't even take, you didn't even wait until your break. Like you were just feeding the machine.

Patrick Moorhead: I love it. It's, it's incredible. I've got a, so what I do is I put all my pills and stuff here. I may have to crank, I didn't have a chance to eat breakfast. So I might, might pick up a meat stick on occasion.

Daniel Newman: You should eat something because, you know, you gotta get after those good workouts, you gotta get the proto. Well, look, there's a lot, there's a lot to that. I do agree with you. The idea of, they kind of did this physical AI factory concept on, on, in on-prem, which is just another, you know, cloud on-prem. I don't know what you kind of call it, but like, I do think that's going to be a solution. And we're seeing a lot of kinds of sovereign deals showing up in other numbers. You know, you went on Bloomberg, talked to HPE yesterday, but a lot of their AI book is sovereign related, slower. They have to be more methodical, they have to be more careful. The infrastructure has to meet, like you said, not just logically, but in many cases, physically, especially in Europe. Very specific about how they want to do things. Something to keep an eye on. Let's hit on a couple other topics, Pat. NVIDIA had a busy week. A few different news announcements. One came out early in the week with a strategic announcement with Synopsys. This was kind of to me like it was a dollar investment announcement that came with a little bit of like we're doubling down on what we've already committed to. I think that there's a $2 billion investment. There was an article that came out this week by CNBC that basically said Nvidia has a cash problem. They have too much. A problem that you and I wouldn't mind having. But the net of that is, I think Jensen continues with this extreme co-design and everything he's trying to do. He's investing in his ecosystem. He's investing in every part of the stack that allows to optimize silk and optimize software. And of course, Synopsys is a super important partner. And this, of course, doesn't just look at data center, which, of course, you can do a lot of the design and simulation and chip development for data center. But with Synopsys and Ansys now together, a lot of that future of physical AI, they're going to be one of the really important key partners down the line for omniverse and for robotics. You know, I don't know that you would have much more to add to that, but I'm happy to give you the floor. But that was, to me, it just felt like kind of a public doubling down of what was sort of already a pretty symbiotic relationship.

Patrick Moorhead: Yeah, I would agree. Nothing more to add here. NVIDIA had made some pretty big announcements with cadence and with synopsis. And this is the front end to all chip development, all physical design development, and all package development. So it just makes sense they would invest.

Daniel Newman: Yeah. And so, you know, otherwise, there was some more talk about the China and state by state regulatory agents and came out kind of talking and basically said that Trump enabled. I mean, it was actually very positive. And I think something that, you know, that basically we have. we, the policies and the regulatory and the deregulation that's been done for energy and stuff, basically that this administration, I don't know if it's, if it's, you know, suckling at the power teeth, or if it is truly a belief, but I think the policy has enabled the US to maintain global AI leadership. I do believe we're still well ahead in the chip space. And I do believe that there is some deep down desire to be able to sell more H200s or Blackwells to China, and there seems to be more policy still on the fence there, Pat.

Patrick Moorhead: Yeah, for sure.

Daniel Newman: All right, let's keep moving to talk about Apple.

Patrick Moorhead: I think you came out swinging a little bit on X. Yeah, listen, another week, another big executive departure. I have to tell you, though, If I look at the Apple apologists out there, it's funny. The initial hot take was, oh my gosh, this guy, Dai, who runs design for interface design, is leaving. It's the end of the world, and this is a big deal. And then about 24 hours later, and I don't know if it was because the Apple PR team went out there and went on the attack. They're an awesome PR team, by the way. But then you saw people saying, oh, this is the guy that did this interface. He sucks. His team doesn't like him. Good riddance, right? Apple is so good at that. But if you look, I mean, they've lost 10 key people out there. It feels like they're losing an AI executive every month somewhere. Serious brain drain there. My big point is, at what point does the board of directors come in and make a leadership change at the top? I don't think they will because they're a pretty, how do I phrase this kindly? They're a board that follows what Tim Cook does. And they're really not going to push in. And unless the stock drops 25% or something, I don't think you're going to see that intervention. And quite frankly, I don't think we need to see it. I mean, Apple, regardless of their lack of innovation, is still crushing it financially. You're probably going to think this is a net, but until the green bubble goes away, Apple is still going to have the advantage and there's no screaming reason to leave the Apple ecosystem at this point.

Daniel Newman: Yeah, I would agree with you there. They've got deep moats. They've been a bit of a defensive name because they are stable. They have not taken the risks. People probably love that their CapEx investments have been small, especially those doubters that this is an important investment right now. And there's definitely a certain bit of the market that thinks this way. But there's something going on, Pat. There's something going on that this is not a company where people left. It's not a company where people left to go to a Facebook. I know it's meta now. It's not a company where people left to go to Google. I mean, if you were at Apple, that was the crown jewel. That was where, you know, if you ended up at Apple, you were at the best of the best. You were the best company with the most innovation that had the most secure long-term position, the best stuff to own if you were going to get employee options. Like there was nothing better than that. But I mean, at the same time, like all this disdain, it's like, I think Apple, is almost worth what NVIDIA is worth again. Like it literally has crept back up despite not doing much of anything. And so it's pretty wild how Apple just has so, you know, and again, I guess when you're spending like hundreds and hundreds of billions of dollars buying your own stock back for years and years and years, you can keep a low float and you can keep your investors happy. But the innovation thing is a real problem. I mean, it's a real problem. Like at some point they're gonna have to do something. I don't think they can just co-op Google's Gemini or OpenAI and satisfy people. At some point, I think it's going to be a challenge. I just, maybe some point as far up into the future that they can just keep playing this game. I mean, Tim Cook's done great in terms of returning value, and that's his job. Despite the fact that you and I as technical people, technology people, don't love some of the things that we see. Speaking of not loving things we see, I mean, the FUD machine's been out in full effect this week. You know, it's great when, I know, I know, I can't help myself. You know, it's great to wake up every day to another story about AI and not, nobody needing AI, nobody wanting AI, and that everybody's investing in, you know, dark fiber. But here's the thing, Pat, is that's not the case. And Microsoft, which by the way, typically tends to play it pretty cool, even when they have kind of what I would call inaccuracies reported to the market, They came out pretty hot after this information report saying that they were lowering targets. What was your, sorry, this one's actually, I'll start this one. You know, this one was really interesting to me because it came out hot, the market like tanked. By the way, tanked the whole market. The whole market went down on this news that supposedly they were lowering sales quotas or growth objectives. And by the way, the news changed a few times. They changed the headlines a few times, but then Microsoft basically came out and said that we just flat out, this is inaccurate. This is not true. that the piece was conflating different pieces of information. And by the way, I thought it was pretty ironic. I don't know if you saw this, but a day later, Microsoft announced a price increase on a number of its commercial software, which wasn't all directly, but it was a number of the bundles that included a lot of these AI software features, which was really interesting. But Pat, I mean, it's like, This is more to me a story about how information gets misconstrued and gets out there and creates major ripples in markets. You know, there's been some of this around GPU, TPU, where you and I have spent a lot of time educating this last week. This is another case where a story comes out. It's inaccurate, misleading. It actually causes meaningful pivots in share prices. I mean, God, is it just gotten this easy to manipulate? What did you read on the whole story, first of all, and then kind of what do we do about this constant barrage of misinformation? Any thoughts on that?

Patrick Moorhead: Yeah, so to me, this didn't reek of malfeasance, somebody trying to tank the stock. I know a lot of people have talked about that. But it's just, first of all, the thirst for a hot story, which I understand, I mean, particularly given what's the right word, the decrease in page views and trust in the press, actually decrease in page views. Trust is a different thing. It was just more about inexperience, right? Which is trying to I hate to think you have to work at a big company and understand how this works, but there's a lot of moving parts when it comes to quotas. Don't diffuse quotas with what you have committed to the street. Typically, your sales quotas are higher than is committed to the street. There's also probably 500 different AI products inside of Microsoft. you know, you're always moving some down, moving prices down, you're doing bundling, like, like you talked about Daniel. So there are just so many, but the fact that Microsoft came out so strong very clearly indicates to me that, that they tried to get, to get the information to, um, reconsider it, but, but they just, they just ran it anyways. And that's a big surprise. I mean, Microsoft rarely does that. Like they've got a really good relationship with, with the press. So they must have, and that, I mean, that gave me a lot of confidence to know that there was something that was awry here.

Daniel Newman: Yeah. My, my big beef with the whole thing is just, You know, one is they said they put in their comment that they proactively had already refuted it before published. They told him it was inaccurate before and they still went ahead and ran it. So where did the information come from? And then at what point is a publication that causes major losses in the market held accountable for publishing inaccurate information? I don't know. It's an interesting problem right now because it's not just them, but it's happened a lot. I mean, remember when they did the whole thing about NVIDIA's mask set, where you called out, and I think they said something, but like the indication of the article was that there was like some major delay. I think it was like a week. I think it ended up being like a week and it like absolutely crushed, uh, you know, caused a crushing reaction again. And this just seems to happen a lot. And it's not just this publication, there's others, but like, you know, the power of, of, of influence is certainly becoming bigger and bigger and its ability to, to, to, to move markets. But I mean, look, you know, um, Microsoft is a conservative shop with this kind of thing. So I read pretty seriously to the fact that they refuted it as a pretty strong signal that it was not correct. Because I think if there was any truth to it, they probably would have let it lie. Anyway, so that's a deeper topic maybe for another day. Maybe there's a flip. about whether I can tweet rocket ships and fist bumps on every topic. And if that's a violation of I'm doing pumping or not when I do that, but I am giving you a fist bump.

Patrick Moorhead: That'd be a good topic.

Daniel Newman: All right, buddy. Only because my number of rockets versus my number of rolly eyes is getting a little out of control these days. Don't get me started, Bestie. I like when you give me shit. It's my millennials. It's the millennial side of me. But don't make me reply to your tweets with a picture of my little play school weights. And that was epic. You may look bigger than me, but you know.

Patrick Moorhead: Listen, the power of AI, Daniel.

Daniel Newman: Yes, the power of it. We already did that. We need to start upgrading our videos of us lifting. So we're like, now I'll make it 700 pounds, like on the bench.

Patrick Moorhead: Yeah, by the way, I tried that with Sora too. It gave me like a quadruple chin. I said- It made you huge, but fat. It was the weirdest thing. I said, get, you know, make me a video of me bench pressing with four plates on each side. And it gave me like four plates, but four chins. It was the weirdest thing.

Daniel Newman: But it couldn't quite get all the way there. Yeah, it makes a lot of mistakes like putting two Jensen's in my photo. Yeah. That is that is weird. It's like this seems like a basic thing it could get right and it doesn't. You want to talk about the Marvell deal or do you want to just hit that when we get to earnings?

Patrick Moorhead: Because we can hit the earnings.

Daniel Newman: All right. So why don't we, Pat, because, you know, you know, we love to debate. Why don't we talk a little bit in the flip today? Trump's administration came out, now they're all in on robotics. They're all in on, you know, they're moving on. Are we done with AI? So like question, you know, is Trump's pivot to all about robotics a smart move at this time? Is it the next thing after AI?

Patrick Moorhead: Well, this point flip, what happened?

Daniel Newman: I won. I won. I am. I am the guy that's saying that's a smart move to robotics. Look, I'm going to make this argument short. I'm going to make it sweet and I'm going to be victorious. China is kicking our ass in robotics right now. Drones, robotics, autonomous vehicles. We don't see it here because we don't let it come here. And that's one of those things where they don't let their people see Google or Meta, and we don't let our people see how amazing their next generation autonomous vehicles and robots are. And so right now, the US understands that, well, yeah, we have some advantages in the AI space. That's done. That's been well set. The policy is in place. Energy is being built out. Our chips are being well controlled. And we've got that particular opportunity under wraps, we will win the world's AI war. But what we're about to lose is the world's robotics war if we don't start to fight it and put more energy and more investment into that space. This is all about China. This is all about not accepting defeat. This is all about The fact that our vehicles are being exported less, our technology and robotics is falling behind. We may have Optimus and we may have Boston Dynamics, but we are so far away from robots in everyone's home. I don't care what Elon Musk says, I'm still waiting for him to drive one autonomous vehicle across country. He promised to be done in 2016. Hasn't happened yet. Doesn't look like it's going to happen anytime soon. So BYD for 40K offers cars that are twice what a Tesla Plaid is. we're falling behind. And this is all about physical AI. So when he says robotics, he's talking about all the things that are autonomous, edge connected. This is drones, this is vehicles, this is robots, and we need to get big, get invested, and make sure that we win this particular next phase. Not to mention, if the president wants to keep the market going up and up, talk about a pump. The guy basically said, we're going to keep the markets hot. We're going to keep them at all-time highs. That's appreciated. Well, you know what we need? We need more TAM. You know where more TAM comes from? Another $3 or $4 trillion at least that will come from unleashing the power of physical AI and pouring more investment into that space. You think that data center AI is big. Just wait till we see the size of physical AI at the edge, at scale. This one isn't even a hard argument, Pat. This is all about winning the war. We've already won the AI war. We're going to win the robotics war, but we've got to turn our attention there. Smart move by the Trump administration.

Patrick Moorhead: So listen, I know that we can all walk and chew gum at the same time. That was my dad expression, by the way, that he used with me. you know, here we are clamoring for capital clamoring for energy on what I would say is an even more important thing, which is getting the AI thing straight. And I do know that Edge AI, industrial AI is AI, but I think that we should be getting the first part done first. I mean, right now, enterprises, it's half-baked. Downstream usage right now is still a question mark. And capital is finite. Where is all this money going to come from? Who is going to pay for all of this in the end? I know, Daniel, you've been espousing about all the debt that we have. What is going to be the reduction in spend somewhere else to be able to fund this when, quite frankly, we haven't even solved onshore manufacturing? Thinking that a 10% investment into Intel is going to solve everything is not going to happen if the big chip makers don't fall in line. And I'm curious, like, what happens to the labor force? It just accelerates the whole people versus robots discussion. And quite frankly, accelerating investment in this space is accelerating the Terminators taking over our planet. I'm trying to say this with a with a straight face here. And if we thought getting Faster permitting was hard to drill and reboot nukes and stuff like that. Imagine the state by state regulation, the state version of OSHA coming in and applying these robots. One carve out I will absolutely put out there is for the military. 100% full stop, overinvest. I want 20 companies doing robots for defense. So with that, Daniel, you can cry and weep.

Daniel Newman: I think I won. Maybe. But not really. Did anybody care? Who's been winning more of these? Have you been sharing? At Patrick Moorhead. Don't at me. I don't care. At Patrick. Look, I think we can walk and chew gum. I do think that, but I do think we probably should make sure we get this AI thing right and we get our manufacturing thing right. By the way, we got to make more chips to support this. And I'm pretty sure we can't, because I'm pretty sure Micron is now taking whatever little capacity it had to make consumer memory and turning it over to make more AI data center memory. I just mean, where does that come from? I don't know. And Gemini keeps breaking because there's not enough TPUs. I don't know how the hell they're going to sell TPUs to anyone else. They need more for their own stuff right now. It's like, we can't build enough for what we're trying to do. And we're not doing that much yet. Wait till we're doing more.

Patrick Moorhead: Yeah, man. Gemini's breaking all over the place. It's crazy. It's really good when it works.

Daniel Newman: I know. I know what I'm saying is like, you know, you sure you want to give a million chips to Danthropic right now? You might want to for making sure that my nano banana works, you know, because because I don't I don't know why I keep getting two Jensen's. So I want four plates. Seriously, four plates, not four chins. I'm working on chins. Now that I've got less beard, it's going to become obvious. You know, that was it. You're looking sexy as heck. Thanks, buddy. Thanks. You know, seriously, though, like that's the one upside of a beard is when you have a kind of a I don't have a I don't have a bad job when you don't have a really strong jaw. You know, the beard sort of covers that little chin sag you start getting with our old age. Hey everybody, got a little chin sag going on here. Maybe I should grow my beard back. I'm all right. I'm all right. Don't, don't, don't do it. No? No. All right. I'm doing like the three day stubble thing. I'm kind of enjoying that. Like not shaving, but not shaving. You know what I mean? Shaving. All right. So, hey Pat, like markets, tons of stuff happening. And even like, since we came up with this list, stuff is happening, but you know, my favorite part, I don't know if it's your favorite part, but let's get to the bulls and bears You know, I love this part so much is because it's after I lose a debate, I get to talk about something I like.

Patrick Moorhead: I know you're a markets guy.

Daniel Newman: I get that. No, I just mean, like, I tend to lose every debate. You took the wrong side of that argument, though. My point was, is I'm just tired of losing debates. And at least we can talk about something I feel good about.

Daniel Newman: Come on, man. Come on, man. Oh, man. Come on, man. Well, Pat, rate cuts are coming. Jobs unexpectedly fell. We're not getting full data, but ADP data. By the way, guess what? PCE came in from September, came in at expected. Rate cuts are coming, but what do you think? Private payrolls fell a lot steeper. They were expecting up, they ended up down.

Patrick Moorhead: I don't believe anything anytime from anybody, but I do believe that we are, what's interesting is we're deporting a ton of people and theoretically we needed more people to do the jobs. So the only thing that I can see here is maybe it's retail, right? We're going to have a soft retail season, which might make sense based on consumer confidence. But that's, yeah, that's about my value add there, buddy.

Daniel Newman: Yeah, I mean, look, I think, I still tell you my biggest fear of the economy is I don't know if rate cuts can solve the job problem. Yeah. I don't, I mean, like, I think companies feeling more confident in spending, and I think people feeling less in debt and feeling like they can move, like giving mobility through lower mortgage rates and stuff. But like, I just don't know if we can solve it. But inflation's cooled, at least according to PCE data. And again, I don't, I don't really know everything.

Patrick Moorhead: It's very concerning though, Daniel is it was small business that led the cuts, you know, and also this is the first negative print in a long time on, on, on ADP. So, you know, potentially right. The people who want to go after the administration will, Well, we'll hit them on everything, but I'm really interested to see if they can cut that by small business vertical and see where most of the action is.

Daniel Newman: Yeah. Well, I mean, look, the pressure is real. Small businesses and the technology is good. And by the way, did you see that MIT thing? It said 95% of AI didn't work. And then they came out with a new study that said 11% or 12% or so of jobs by like 2027 could be automated with AI. Yeah. Yeah. People like they can't even agree with themselves. I mean, how are we supposed to do like when you get that kind of like mixed signal?

Patrick Moorhead: You know, you got to have a smart human in the loop. People with experience that that understand this stuff. I know a few guys.

Daniel Newman: It's like, by the way, it's like when, you know, when she bats your eyes at you and then says no, when you ask her out for coffee. Yeah. My wife does that to me all the time. I can't figure out why. Anyway. Let's talk a little bit about Marvell, had earnings this week, got up on response to its earnings. This company has been sort of, you know, there's been a lot of sort of criticism. Matt Murphy, their CEO, hasn't been maybe bullish enough about their prospects. You know, let's be candid. Doing custom AI is hard. This is a company that by the way does a heck of a lot more than just custom AI chips, but this is what the value of the company so much over the last couple of years has been about. Basically, how much tranium. Well, the good news is this week you got a lot of validation tranium to is. is completely booked. Trainium 3 is completely committed. Trainium 4 gets announced. It looks like based on some of the comments about expectations going forward, that Marvell thinks there's going to maybe pull forward or some big volumes on Trainium 3 plus 4. That was a really good result. I mean, a company at 37% year-on-year growth, strong cash flow, And, you know, data center revenue grew by 38 percent, Pat. So now it now represents almost three quarters of the company's revenue, which is a big part of that multi-year pivot it made out of consumer and out of other spaces. Data center is ruling the roost. So, look, this is a company, Pat, that I think, you know, we still think about 25 to 30 percent of the trillion plus of AI accelerators by the end of the decade are going to be custom. And we still think that this company could probably win in that 15 or 20% range, maybe as high as 25 of that percentage. So there's a lot of upside here for Marvell. And of course, a lot of it's hinging on the success at Amazon. That's not the only design they have, but that's the one that's kind of most notable. And that's the one that's kind of getting the most attention in the marketplace. But there are other big parts of their data center business is networking. They're gonna be important to the speed of light as we're trying to connect data centers, scale out and scale across. And then, you know, Pat, they also made an acquisition, which I'll let you touch on.

Patrick Moorhead: Yeah, so I don't usually write about financial stuff, but I wrote a lead up after a JP Morgan conference that they did. And everything that happened this quarter in the guide was baked into that. Investors just didn't pay attention or just didn't trust it. When you've got big CapEx increases from the hyperscalers, the downstream impacts are going to possibly hit hit hit Marvell. And I said for a long time, people are way too focused on XPUs versus the connectivity. And if you look at even the, you know, the P ratios and the valuations for some of the smaller networking companies, And you would lay that on top of Marvell's businesses when you slice and dice them. It just looks sideways. So I think people need to re-look at them as an AI connectivity company first. And I think the XPU stuff is gravy and understanding it's going to be lumpy. So it's very similar to what AMD experiences in the, you know, when NVIDIA, you know, crushes something, right? It's, you know, gosh, AMD, why aren't you doing something? here. But no, they had an exceptional quarter and they had a great guide. Revenue was above consensus. EPS was above consensus. I guess you'd call it quadruple beat their guide. They raised quarterly, they raised annual, and they pretty much did everything that the company said they would do in the JP Morgan run up their celestial rights adds another potential $10 billion opportunity and optical interconnects. And you and I are going to be at their industry analyst day next week and we can learn all about it.

Daniel Newman: That'll be fun. I'm sure everyone's gonna be super interested. So tune in. I think we'll have some coverage there too. We'll be having some exact conversations on the six five. Yeah, that's right. So Pat, you know, Salesforce, you heard me talk about 3.2 trillion tokens, but how'd they do?

Patrick Moorhead: Yeah, so listen, I'm not going to pop the champagne corks on downstream demand and the ability to monetize that yet. But some very positive AI signs coming out. You have 3.2 trillion tokens, AgentForce, which is the prime Application and then underneath that i call the precursor data 360 air are reached one point four billion dollars a hundred and fourteen percent. Year on year i know a lot of people can make fun of a you're calling everything agent force therefore everything is a i but agent force plus data 360 is absolutely absolutely. I so good. a good path. Again, good early signs. We have to see if this continues quarter after quarter. I think if it does, you're going to see investors believing in this trade even more. And Salesforce is kind of the bellwether for downstream AI usage. So it's important for the overall AI trade.

Daniel Newman: I agree with that. They're definitely one of those key Metrics, because like as people pivot, like there is this kind of like build your AI thing, but a lot of companies are going to look into their app ecosystem as the low hanging fruit to start to implement AI. And this is one of those applications that so many businesses are running on. I look at them, I look at Oracle, I look at SAP, I look at ServiceNow, like these are places where AI starts to show up. The question that we all have is like, is AI just kind of this catch all for things that are being done that were always being done and now we're calling it AI or is it truly Now, but the thing is, like, agentic is truly measurable. Like, are we having tokens generated by agents that are doing processes and performing actions? That's a real important metric. And point is, like, 3 trillion tokens by one company. that's barely been deployed at scale. I mean, I just don't think people are bullish enough for how big enterprise AI is going to be when it actually really starts to get lags. I think we're still looking at LLMs and subscriptions to open AI as the TAM for AI. And I just think that completely misses the point. You and I got to talk to Antonio Neri, CEO of HPE yesterday. You did a good hit on TV about it too. I'll just kind of quickly say, look, this company wants to be as committed to be and has become a network first company. So yes, they have servers. Yes, they do security. Yes, they have, you know, hybrid cloud, but this is built around enabling enterprise network. I liked the conversation we had. This is one of those companies that you have to get kind of underneath the strategy to understand the results because when you see companies like Dell's AI server book growing and then you see their book growing at a smaller pace, it's a conscious decision and you have to be okay with that if you're looking at this company. But like their conscious decision is they want to be in sovereign deals and enterprise deals and higher margin neocloud deals. They do not want to chase the some of the types of deals that are super micro and dollar. So prolific at winning. And that's really what they're focused on, a huge growth in networking. But you do have to account for the creative nature of Juniper showing up in their numbers. It's not all organic growth, but it's still a huge number. And of course, enterprise networking is going to be a huge opportunity and bottleneck for successful deployment of AI across enterprises, which you heard me just saying. Company got margin enhanced margin growth, but it is seeing deal flow somewhat constrained because these deals for enterprises, especially these sovereign deals, which again was like 61% of their book, are taking a longer time to go through. One last thing I'll say is their ARR number looked great. Another big gainer. I think they're in the three plus billion now on their GreenLake ARR. That was always a big part of the strategy, always a number I've been watching. So like I said, largely good. I think everyone would like to see revenue grow faster, but it was a record quarter.

Patrick Moorhead: Yeah, listen, I think on the plus side, this was the juniper validation. It was going to be a creative day one. And here we are. And I think there were a lot of activist investors who just hated this thing. came after this. I'm not saying it's free and clear because we've got to see growth, but I think this is a very good sign. On the overall AI number, these are big lumpy deals, some big sovereign AI deals with countries, even potentially some Neo clouds out there. I think the server backlog, which is massive, I think will indicate where this is going. It's interesting. I said on Bloomberg last night that I guess it'd be a day after this, it'd be Thursday night. I think we could see some improved reception when we wake up in the morning. And sure enough, the stock was trading down 6%. And it's actually up today. And I do recognize the market's up. Today, overall tech, but not up, kind of leaps and bounds. In fact, Cisco's off, Dell is off, but HPE is up. It was trading down 4% when I was on Bloomberg. And so, yeah, I mean, so far, so far, so good. From a sovereign cloud perspective, I mean, HP is in a great, is in a great position. They're very strong in Europe. They're very strong in the Middle East. And I also thought Antonio's tone, he's very bullish on 2026, right? A little bearish on the immediate Q1, but very bullish on 2026, which I thought was, thought was good. I have to dig into the transcript on the questions about GPUs. I haven't gotten there yet. But I think that the challenge of getting GPUs likely reared its ugly head here.

Daniel Newman: Yep. So a couple more, Pat. Let's run through them. Dell also had its earnings, and I think Dell had a good one.

Patrick Moorhead: Yeah, they crushed it. I mean, it was all about AI servers, and they shipped $3 billion in AI for the quarter. And the guide was really good. The stock rate, I mean, they were up 4%, which was good. They're likely taking share from Supermicro at these hyperscalers, which doesn't surprise me at all. And also, Dell just has a reputation for quality. I've checked in with some of these Neo clouds, and it's a quality thing, right? And the speed at which Dell can deploy these. PCs, right? Consumer market's down. They're not widely tuned into that, but they were down like HP, Inc. was down, but I do think that this Windows 10 end of life is adding business and Dell is taking advantage of that. You need to do a double click on storage. There was the talking point that said, hey, compute goes first and then storage will come in second. I think storage was flat again, but I do think there was commentary that PowerStore was holding share against Pure. I'm wondering about Vast, DDN, and guys like that though. Yeah.

Daniel Newman: Yeah, that's a part that we'll talk a little bit here in a few minutes about pure storage as well. My read on Dell is just absolutely crushing it on the AI data center play. I mean, huge numbers, huge growth. It's going to always have some contested margins, but they're creating cash. They're seemingly managing the business. They're seeing some strength in their CPU server business as well, which looked pretty good. The PC thing is just like an afterthought right now. It's so funny, but it just is. I mean, they've done well in commercial, but they really have focused on premium and commercial and they're not putting too much into right now that overall big consumer. The dude you're getting at Dell era feels like it might be over. I do want to give a shout out though to Michael Dell for what he did for this Invest America. I'm sure there was some great value from a tax standpoint, but philanthropically, what, 25 million accounts back?

Patrick Moorhead: on Tuesday, I got you gonna you and I got and we worked out. Michael and Susan Dell gave $6 billion. Yeah, they did.

Daniel Newman: Yeah, more than maybe most of us will ever be able to do in our entire lifetimes. But at least I got a pump. Worth it. All right, we got one more go one more. where we let you all go and you can't go until we're done. That's the rules. I didn't make them up, sorry. Pure Storage, another one, Pat. Company performed well. It grew. What was it? Beat on revenue. It was pretty much in line on EPS, raised a guide, and yet I think you said it so well in your tweet, the stock falls. hard to say, make that make that make sense.

Patrick Moorhead: And I was down 7% dude.

Daniel Newman: Yeah, I mean, what a beat down. I want to look at it right now because I want to see like kind of how it had recovered. Remember last quarter though, they just crushed it. And um, You know, I'm trying to I'm trying to find their like right now quote.

Patrick Moorhead: So give me one second. Yeah, they so on August 22, they went from 50. Let me see 58 to 80. Right. So they're down 20% this week.

Daniel Newman: but they were up like 70% left. Like it was like some cute. So basically I think what was, there were some, probably some big options bets and some big swings that they would have another kind of bang blow up quarter. And what they have is a good quarter. They had a good quarter, but like they have some bigger investments making, they're growing in the right areas, their cloud stuff is growing, which is where they've been super focused. But I think that this is one of those after like MongoDB had a huge week and they got like, people were looking for some massive outperform here and what they got is a performance. And honestly, I think that it's had such a run up that it kind of got penalized for just doing the job well. But like you said, price follows sentiment and the market, when I say it's a weighing machine or a voting machine. So over time, it seems like this is a bit of a weighing moment where it's looking at the company and its growth and its ability to execute and its ability to maximize and capitalize on AI. And I think everyone's saying it's good. And remember, it's still up like 25 or 30% from the last quarter, but maybe they said it didn't in this moment show enough to deserve to keep that momentum that it had. But like I said, Pat, I think overall, we know Charlie well over there. He's just continued to do a nice job in executing. And sometimes he can't help when it runs too far and he also can't help when he executes and people sell the stock. So I think it's still on the same path, still gonna be an important part of this I mean, storage is constrained too.

Patrick Moorhead: Let's face it, storage is just another constrained part of this AI stack. Yeah, the sell side just yawned this off. And there were three things that if I were looking, they talked about significant incremental investments in R&D and sales and marketing to close AI opportunities. They also talked about stopping disclosing specific hyperscaler shipment volumes. And we're talking about new business models that they're exploring with the hyperscalers. And but anyways, I think this is just short term, short term churn. And like I said, sell side, they just yawn this off.

Daniel Newman: So a quick note on our way out the door, Pat, Anthropic IPO. I don't know if you have much to add, but I called it it's going to be a $500 billion IPO. You're going to get in on it? Are you going to get in on it? I mean, we don't have any conflicts with them. I could. I just, I don't know, Pat. I don't know. But I do love the fact that they have a moat. Their moat is cold. Code is a real thing and it's going to be a big, big part of the future of development. I just, is there enough monetization to justify this? I mean, come on, Arvind Krishna came out through cold water on the whole AI trade. I mean, thanks Arvind. You know, he came out and basically said that there's not enough revenue to be generated for what we're spending in CapEx. And I mean, Arvind's a smart guy. And he talks like, I don't know about Michael Burry, but Arvind doesn't know what he's talking about. So that does give us a reason to have to look. Yeah, true. But like, but you know, I do have a high tolerance for risk. So maybe I'll throw a couple of shackles at Anthropic if it does IPO, but I know my team uses it pervasively for the future of AI platform, like Cloud Code is our toolbox.

Patrick Moorhead: Yeah, Claude's good. Yeah, I got- I know Pico likes it, doesn't he? Doesn't he? Well, he likes it as his primary for everything, but he uses cursor. Okay, another big one.

Daniel Newman: Another big, probably future IPO coming at some point. Yeah, totally. All right, Pat, well, this was a fun show, even though I might've been in a bad mood, I pretended to be in a good mood, the whole thing. I do appreciate everybody. We'd love all of you for tuning in. Like I said, send all your negative comments to Pat. He likes to engage with trolls. I personally don't read my replies, so I will never see them. But in all seriousness, we do really appreciate you being part of the community. Episode 287 is in the bag. See you all very soon. Bye-bye.

‍