Google Cloud Goes Full Stack, Amazon's $100B Anthropic Bet, Intel's Foundry Moment & More

The handpicked topics for this week are:

1. Google Cloud Next 2026: Full-Stack AI and New TPUs — Google Cloud Next has cemented itself as the second-biggest AI event on the calendar, with Thomas Kurian declaring the proof-of-concept era over and enterprises now in full production mode with agents. Google unveiled two next-generation TPUs (the 8i for training and the 8t for high-throughput inference) and reinforced its full-stack differentiation from infrastructure through Gemini Enterprise Workspace. (The Decode)
2. Google's Agentic Security and MCP Push — Google made a significant move into agentic security, combining Wiz and Mandiant into what Pat calls a sleeper announcement of the show. Google also committed to placing MCP servers across all of its data surfaces, meaning even non-Google platforms can tap into Google data without full lock-in. (The Decode)
3. Google Distributed Cloud and On-Prem Agentic Orchestration — Google took the biggest first step Patrick has seen toward a true agentic orchestrator that spans on-prem enterprise and public cloud through progress on Google Distributed Cloud. No other company has yet attempted cross-environment agent coordination at this level. (The Decode)
4. Amazon's $100 Billion Anthropic Commitment — Amazon formalized a commitment of up to $100 billion into Anthropic, including five gigawatts of Trainium capacity, making it the largest non-NVIDIA silicon commitment in history. Anthropic's valuation crossed $1 trillion just weeks after a $350 billion raise, a pace that has left even veteran analysts searching for new language. (The Decode)
5. Adobe Summit 2026: Enterprise Agents and Jensen's Endorsement — Jensen Huang took the stage at Adobe Summit to deepen the NVIDIA-Adobe partnership, calling agentic workflows the new front end for SaaS rather than a replacement for it. Adobe reported $250 million in Firefly ARR and 45% quarter-over-quarter growth in agentic tool usage, yet the stock continued to disappoint investors expecting hypergrowth multiples. (The Decode)
6. Apple's New CEO: John Ternus and Tim Cook's Legacy — Apple named John Ternus as its fourth CEO, closing the book on Tim Cook's 15-year tenure marked by custom silicon success, services expansion, and operational excellence, alongside misses in Vision Pro, the abandoned car project, and Siri's failure to become the AI front end it should have been. Ternus is a continuity hardware candidate, and the most consequential decision may prove to be keeping Johny Srouji over all of hardware. (The Decode)
7. Intel Foundry: Elon Musk, TerraFab, and 14A Validation — One day before Intel's earnings print, Elon Musk publicly confirmed TeraFab will use Intel's 14A process, delivering the first verifiable public wafer commitment on that node. Intel then reported a 23% stock surge, 22% data center growth, and EPS of $0.29 against a $0.01 street consensus. (The Decode)
8. The Flip: TSMC vs. Semiconductor Equipment Makers — Pat and Dan take hard opposing stances on who holds more power in the AI supply chain: TSMC with its control of over 90% of advanced AI silicon and irreplaceable process expertise, or the equipment oligopoly of ASML, Applied Materials, LAM, and KLA without whom no leading-edge fab can operate. The real answer, they conclude, is deep interdependence, though TSMC's combination of talent and leading-edge control gives it outsized leverage today. (The Flip)
9. Intel — Intel's earnings were a blowout across the board, with data center up 22%, EPS of $0.29 versus a $0.01 estimate, and guide raised, driven by CPU price increases, customer pull-ins, and packaging volume growth. Hosts discuss whether the stock at current levels is pricing in foundry revenue that has barely begun to materialize on the tape. (Bulls and Bears)
10. GE Vernova and Vertiv — GE Vernova posted a beat on revenue and EPS with orders up 71% organically and a $163 billion backlog, while Vertiv reported sales up 30% and raised forward guidance to $14 billion. Both companies reflect the acute power infrastructure demand tied to data center buildout, with Patrick noting their growth was likely already baked into share prices heading into the print. (Bulls and Bears)
11. ServiceNow — ServiceNow beat across the board with a Rule of 57 growth result and AI run rate up to $1.5 billion, 50% above its prior target, though margin headwinds from three acquisitions and on-prem impacts from the Middle East conflict weighed on sentiment. Daniel argues the market has not yet accepted that workflow automation at enterprise scale will not be replaced by vibe-coded alternatives. (Bulls and Bears)
12. IBM — IBM posted a triple beat with Red Hat up 13%, software up 11%, and Z mainframe up 48%, the latter driven in part by AI-assisted COBOL modernization tools making the platform newly relevant. The stock slid after hours despite the results, continuing a pattern Patrick describes simply as silly season for enterprise infrastructure names. (Bulls and Bears)
13. SAP — SAP beat on revenue and earnings with cloud revenue up 19%, cloud backlog up 20%, and total backlog up 25%, reinforcing that enterprise ERP customers are not moving away from core platforms. Daniel and Patrick agree this is another data point showing enterprises are building AI on top of existing software stacks, not tearing them out. (Bulls and Bears)

Want the full breakdown from the ground at Google Cloud Next? Check out our coverage at SixFiveMedia.com.

Be part of our community — hit that subscribe button and let us know your take on this week's pod.

Disclaimer: The Six Five Pod is intended for informational purposes only. The hosts and guests are not financial advisors, and nothing discussed should be construed as investment advice. Always consult a qualified financial professional before making investment decisions.

Patrick Moorhead:
We absolutely know what we're talking about. Welcome back to The Six Five Pod. It is episode 302, two more episodes than 300. You can do the math. I can do the math, but it's great to be back. We are back here in Austin in our offices after traveling to Adobe Summit, Google Cloud Next. Daniel, it's great to see you.

Daniel Newman: Yeah, man. It's good to be back. It's been a while. I have not seen you since like 11 last night.

Patrick Moorhead: No, I know, you know, getting in late and then doing an early workout, you know, I started at 7 a.m. What time did you start this morning?

Daniel Newman: Um, don't pick on me because, you know, I'm like normally the 430 guy, but I actually slept till six. I was tired. I was tired. I didn't get to bed till probably 1230 by the time, you know, I live, I live pretty much in Houston. So we landed at like 11, but like by the time I actually got to my house, it was a lot later than that. And my daughter was in town and she was leaving this morning. And so I spent a little time with her. And then my wife is driving up to Baylor for mom-daughter weekend. So she's leaving and spend a little time with her. And then the dog is leaving because I can't stand having the dog around when no one else is here to take care of her. So she's leaving. So just going to be me and Matthew this weekend, man. We're chilling. We're going to do dude stuff. We'll probably walk around the neighborhood with our shirts off. get a tan and, you know, cause all kinds of other chaos. That's what we do. It's like when you have Pico, right? When you guys are home alone, you create trouble, right?

Patrick Moorhead: Yeah, I may or may not clean up everything within an hour of Paula returning, but…

Daniel Newman: It's like a frat house and then it's like that one hour, it's like home alone when Kevin cleans up the robbery scene. No, totally. Just a one gold tooth left on the floor.

Patrick Moorhead: So yeah, hey, we got a great show. We have so much to talk about. So many things going on in the market. So much AI infrastructure, AI services news. We think you're gonna really love the show. We're hitting Google Cloud Next. We're hitting Amazon and NetTropic, Adobe Summit. We attended two events, as I said before. Apple has a new CEO. And wow, Intel Foundry kind of lands Tesla for TerraFab. So much stuff going on. So hey, let's dive in to the Decode. All right, Dan, we attended Google Cloud Next, a lot of infrastructure, a lot of stuff going on all the way from infrastructure to agents and pretty much everything in between.

Daniel Newman: Yeah, massive week this week. We got there, what, on Tuesday? Yeah, we got there Tuesday and got there into the evening and we were immediately ushered off to a top secret private event. and they moved back the curtain, and what did they have to announce, Pat? They had to announce, oh, it's my turn to say. They had to announce two, not one, two new TPUs. It's on, baby. And all the headlines come screaming out that it's, Google's about to take out NVIDIA. I'll pine more about that later, but let's just say that's not actually the case. But what they did do, and they, did launch were these, you know, next generation TPUs, 8i, 8t, you know, one big trading chip and one high, low latency, high throughput, big memory inference chip that's going to be designed for what's coming next. And, you know, they talked about this inflection that we're at because the week was all about agents, right? I mean, it was so focused on you know uh Thomas Kurian gets on stage first thing he really calls out is we are past the proof of concept stage now we are on to production and I think that was what we spent all week talking about this kind of layer cake and what they really do and this is really Google's thing and as I was putting out fist bumps and arm muscles and rocket ships about Google all last year before it reached these highs in the market. You know, my thesis was really what it's become. It's got the infrastructure layer, they've got the networking chips, they've got the CPUs. But then you go up the stack, they've got you know, the, you know, frameworks, the developer ecosystems, the easy plug and play, agent builders like Gemini Enterprise Workspace, and they got security, they brought in Wiz. And of course, that, you know, was added to the Mandiant family and the rest of their security, they got governance and policy around it and, you know, the company really is on fire. And by the way, Pat, I mean, I know we didn't sort of talk about the week, but has this not become a circus? I mean, has the event not become a circus in the sense that, like, the volume of people on the ground at Google Cloud Next, you can't even walk, you know, you can't even get to the show floor anymore. It's become so chaotic. I think you did the keynote in a special, you went to a special laboratory where you could focus. Literally, you can't even get a seat. I think that has a lot to say about two things. Obviously, this AI inflection and Google, arguably, is becoming the second show to GTC now in terms of the biggest AI moment. The other thing is just how well Google Cloud is doing. Google Cloud has been a rocket ship for the company. They turned profitable over the last four to six quarters, but now their acceleration is just incredible. You know, Pat, I think they it's too many announcements to hit here in a short kind of decode segment. But, you know, I think what you talked about and we have a good wrap up that we did from the ground. And hopefully our team can link that in here for people that kind of want to get the whole breakdown. But they really did go full stack. And I think that is Google's differentiation. They've been doing it early. They've been doing it often. What is it, 15 years of infrastructure development there? This wasn't a new and surprising entry. This is a company that understands it wants to control its own destiny. It wants to have the full product from data, security, infrastructure, all the way out to the applications. But they also have a really rich ecosystem. So no matter what you want to use, you can build it in Google Cloud. So it was a good one for TK, a good one for the company. impressive show, and that TPU V8. I look forward, hopefully we can get it in the lab soon. Hopefully we can get it in the Signal 65 lab soon. I'd like to see how this thing stands up.

Patrick Moorhead: Yeah, you know, Google really sharpened their message on this one. I went to either the first or second Google Cloud Next in history, and they've just gotten just a lot better in not only what they offer, but how they talk about it. And You know, as you mentioned, the headline was really, you know, we're the only company that does the full stack and listing what that full stack means. And then the secondary message is, is we're open because you really have to have both, right? If you're a closed system and you're the best, then people worry about lock-in. And it just, that's just the way that the enterprise world works. Yeah, TPU, really a 10th year anniversary. I was actually at Google IO where they announced the first TPU and it just keeps rocking. There's no definitive measurements of GPU versus TPU. I'm looking forward to that. And what I'm looking forward to is, first of all, more cross-platform benchmarks coming from more than one outlet and also hitting a variety of workloads. And there is an initiative called PRISM which I haven't had a time to deep dive in. Hopefully, Ryan Shroud over at Signal 65 is all over it. But essentially, it's an open source evaluation tool for conversational analytics. And it is a way to compare TPU to GPU, at least on the Google at least on the Google platform. So really good stuff. And like you said, so many, so many announcements. And, you know, some of the things for me, a bigger one is agentic defense, really a sleeper announcement here. This is essentially the culmination of that $32 billion WIZ acquisition. Had a good conversation yesterday with the the COO of Google and the president of the security division. Hopefully, y'all can Connect to that in the in the show notes Little stuff that you know seems little but I think it's super sexy is MC basically MC peeing everything on an enterprise agent platform. And not only is it the ability, if you're creating your own agents, to put your own MCP in there, but also wherever there is Google data, they committed to putting an MCP server where you could plug into it. So even if you're not locked into the Google platform, you can tap into that. And let me see, what else? I think some of the other highlights for me, I think you mentioned Gemi Enterprise Platform really answered the control plane question that I think the street has been asking. And I think the final one for me was some really good progress on GDC or Google Distributed Cloud. Google Distributed Cloud is essentially having Google capabilities that have been typically in the hyperscaler cloud available on-prem. And it's not every single service, every single capability. But they even brought in some capabilities to more intelligently be able to span data from on-prem to hyperscale, and also agent coordinators that go in there. And I've been a real pain in the butt, always bringing up, well, What about an agentic orchestrator that cuts across on-prem enterprise and the public cloud? Nobody's doing that. And I think Google took the biggest first step that I saw. A side note here, Anthropic was there. They rented out a complete restaurant. And I was just thinking to myself, for what Anthropic has pulled, and at least my conversations with IT leaders out there have been, they're using what Anthropic pulled on 4.6, sorry, 4.6 and 4.7 as an example about they can't trust Anthropic. Um, and, you know, afraid of getting your prices raised service degradation. And it was funny, all this talk about security and, you know, meet those Scott, uh, basically, uh, hacked through a really, really simple, um, uh, method. Daniel.

Daniel Newman: 

By the way, we didn't talk much about it. We'll probably come back to another time. But one thing that is interesting is what's coming next for Google on the model side. Yeah. I do feel like they had a big leap last year, but then in the last few months, it's been very anthropic, open AI heavy. Feels like Google is going to have something up their sleeve.

Patrick Moorhead: 

Yeah, I mean they have to. I think they were in the spotlight for 60 days. And I don't know if they're running out of compute. Um, but all I can tell you is my experience with the workspace agents, uh, broken tools, wrong days for stuff that I've, uh, I've asked for on, and these aren't agents that I cooked up. These were, uh, pre-populated gems that, that Google provides, um, in there. So, uh, I hope, uh, I hope that gets fixed. Um, and, but until then. I'm going to keep a TBD. And to be clear, there's three different modalities, right? You can get agentic inside of Workspace. You can get inside of Enterprise. And then you can go all in. And at work, it's very super, super granular. And I use two versions, the starter and the middle one. And I just haven't had the results that I would like. I'm glad you brought that up, Daniel. Speaking of more compute, it is so funny. It seems like this was weeks ago. I mean, just like a new deal. Amazon committed up to $25 billion more into Anthropic, $100 billion. In in in total in total commitment there and daniel this just gets back to your rocket ship name of you know there is there is never enough compute you know why do we keep having this. This debate you know that the debate is shifted from. You know, we have too much compute with the deep-seek moment and everybody's sold off. Then we kind of went to the, there is no downstream benefits, even though, you know, the SaaSpocalypse is happening, which, you know, what was super ironic. And now I think what we're debating is, How do you pay for those tokens and make profits? And if you think about that, that's really an important movement that we've had. And I think it just shows the maturation of the conversation. And by the way, I think this is the largest non-NVIDIA silicon commitment in history. And I know it was 100 billion over 10 years. I would like to know more granular. I know what was committed you know, was five gigawatts of trinium 234 plus capacity. And by the way, that's roughly the power draw of five nuclear reactors. That's a little little fun fact for all.

Daniel Newman: 

Thank God Nvidia and Oklo partnered this week deeper.

Patrick Moorhead: 

Yeah, that's hilarious. Other things, native cloud console inside of AWS, right, seems to change the distribution model a bit. This is a big deal, right? A huge commitment with with Anthropic and AWS. So, you know, I think, you know, the deal, it puts pressure on, I think Microsoft a bit, like, what's going on with custom silicon? You know, Maya 200, you know, where are some of the big deals for the custom silicon going in? So anyways, it just seems like each week, new multi-billion dollar commitment, in this case, 100 billion over 10 years.

Daniel Newman: 

Yeah, my read on this is that, you know, when Maya's ready and they have ramp and volume, it'll sell. I mean, I'm not even saying it because it's good right now. Like, I can feel like I could safely say that if they can produce a chip that can deliver some level of inference metrics for users, right now it will sell. I mean, we're seeing these $20 billion deals with Cerebris and we're seeing like, you know, and by the way, interesting specs from that company, but like, you know, not much of real data in the wild yet. So, you know, these large commitments and basically, I think you call it like Pat's rule of everybody's talking to everybody. There is a certain amount of like everybody's buying everything they can get their hands on right now. Like there's literally, I mean, did you see, I mean, we'll talk more about Intel later, but like what they're charging for what they thought was dead inventory.

Patrick Moorhead: 

Yeah, no, it's it's wild.

Daniel Newman: 

It's wild. So these things are even like borderline competitive, they will sell. I don't even know they have to be competitive if they just work. So so but I mean, on the on the Amazon side, I mean, look, they're, they're big bets. Anthropic's making big bets. I haven't seen as many kind of the circular complaints. I don't know if people are just normalizing this now. You know, in the beginning when Nvidia did these deals, like people were losing their freaking mind about it. Now, I mean, they're basically putting money in. Anthropic's planning to spend money back. I always say this is just good venture, you know, especially because of the prices in which Amazon bought equity into Anthropic. They're making massive returns as this thing crosses a trillion dollars. Remember when they did a raise at $350 billion a few weeks ago?

Patrick Moorhead: Yes.

Daniel Newman: 

They literally went from $350 billion to over a trillion in valuation in a month. It's just, it's insane. It's insane. This is, you know, this is good, but whatever. I mean, every day, you know, MetaBuy is buying Graviton. I mean, look, everybody's buying everything they can get their hands on right now. So I feel like a broken record. I just, I don't actually have anything smarter to say than that.

Patrick Moorhead: 

Yeah. I mean, once you say the same thing continually for a long time, you know, it's like you got to find a different way to say it or just pick a different emoji, Daniel.

Daniel Newman: 

Yeah. Well, you know, Some people did say that Intel at 20 would end up becoming a generational opportunity. And some people said that they would end up being made by the government at some point because we would need US-based foundry capacity. I mean, some people call things early and sometimes, you know, those people should be rewarded for being right.

Patrick Moorhead: 

You know, it's tough being right all the time, Daniel.

Daniel Newman: 

It really gets tiring.

Patrick Moorhead: 

Of course, I'm just gonna gloss over the giant things I got wrong. Maybe not, I do try to point those out.

Daniel Newman: 

I really don't, there's no value in it. I'm kidding. I'm kidding. We put everything out there for everybody. The good news is I don't delete anything. So you can go back and you can look at all my wrong calls. You'll spend a lot of time sifting to find them, but they're in there for sure.

Patrick Moorhead: 

It's pretty easy to find, do that with AI, by the way.

Daniel Newman: 

Yeah. Well, just go through Daniel Newman's Twitter and find every post where it's become factually, and actually that gives me something to work on later. I might actually have to do a whole roll, a reel, a reel, a roll, and a thread of all my dumb picks over the years.

Patrick Moorhead: 

You should, but hey, before we do that, let's dive into the next topic. Adobe Summit 2026, enterprise agents working with big labs. Daniel, company and the stock, it's so funny. Great financials and the stock just tanks. I know it's not bulls and bears, but it's like, we can't just gloss over that.

Daniel Newman: 

Yeah, well, you know, um, they look at 45% quarter over quarter growth of AI use of their agentic tools. So it's like, I just, this is like the world's biggest disconnect for me. And we'll end up talking about this more in bulls and bears, like when we get to some of the other SAS and software companies. Um, I just think people's eyes and ears are in violent disagreement right now. You know, they're seeing the numbers and it's like, looks normal, hearing the Dario rhetoric, world's going to end. When you have a company that goes from $300 billion to a trillion dollars in value, you tend to listen to the guy running it. But they had a big launch. I mean, by the way, Jensen came on stage to talk about a deeper partnership on what they're doing with OpenShell and Nemotron, basically building models directly into their CX enterprise. And Jensen continues to talk about Adobe as the marketing manufacturing system. And by the way, he said, SaaS agentic is the new front end. So he's basically saying that it's not going to be a rebuild from the ground up. It's going to be agentic workflows built on top of SaaS. Maybe that's what Benioff is talking about when he talks about his headless solution. And by the way, I don't know about you, Pat, but I'm building things every day that sit right on top of my SaaS applications that are pretty much more of the semantic experiential layer that I want. because it's become that easy. I'm not going to completely reconstruct the software from the ground up. But anyways, they had, you know, their customer experience platform. You know, they announced these partnerships that they announced with OpenAI, they announced with Google Cloud, they announced with Anthropic, they announced with AWS, IBM, Microsoft and NVIDIA. So I'm going to continue to say it's Dan's rule of it's and, not or. That's mine. And in this case, that's what's going on here. You know, I think their TAM is growing, their revenue is growing, their profits growing. Some of the stuff they talk about like these, what do they call it? The content supply chain, this thing. I think people underestimate when you are an enterprise, how complex it is to create, distribute, put paid, manage advertising campaigns, all the stuff that comes together for running an organization. that yes, you can kind of vibe code some of these things. But if you're doing this at enterprise scale, if you're a massive brand, and you need consistency, I don't think you're going to build this from scratch. I just don't think that's what's going to happen. So, you know, I think they landed the agent story. I think this rewriting though, is going to take a while. I just don't think people are, again, it's the eyes and ears thing Pat, until people really see that the world has chosen to build on these platforms instead of build instead of these platforms. We're going to have some turbulent times, but I think the businesses will continue to be robust because it's just too much complexity to do any of this stuff from scratch.

Patrick Moorhead: 

Yeah, that's You did great on this topic. I might just not say anything. No, I'm just kidding. Adobe, they're trying to reposition themselves from this digital experience thing to customer experience orchestration. And that's a very, very much a different TAM. But you do have to recognize their success. $250 million in Firefly ARR, right? AI first ARR up 3x. And I think people want 100x, because they're comparing it to not the financials, because both OpenAI and Anthropic are losing money hand and foot. They want to see that top line. Jensen getting on stage was super interesting. And I think people just kind of misunderstood maybe what he was trying to say. People were like, Jensen basically called it a Adobe factory, right? And there were some potential negative connotations to that. But the reality is what they announced there together was a way to keep on brand on steroids. So if you have a product, let's say a Ferrari or a Birkin bag or a high end watch, and you're trying to do visual representations of it in doing different ad copy, it has to be 100% correct. Like, it can't be 99%. It can't be 95%. It has to be 100% correct. And essentially, taking Omniverse, where physical products are designed, pulling that into the digital marketing world, it just makes sense. I think Adobe told every CMO out there that their creative stack needs to be rebuilt as an agent. Jensen was on stage to agree, which I think is the closest thing to anointing you to get an enterprise tech. So I also agree, Daniel, that for four or five key reasons. The market is kind of misunderstanding that, but we will hit that more in bulls and bears. So let's move to the next topic, Apple. a new CEO, third, actually fourth CEO in the history of the company. Tim Cook's 15-year run. First thing, I want to talk about Tim Cook. And I don't know if it's a mea culpa, but I was probably too focused on the new segment innovation You know, the failure in cars, the failure in goggles or glasses, the failure in core products initially with maps, as opposed to looking at the scale and the lock-in and the refinements around the business that Tim Cook's leadership drove, I would say the number one thing that Tim Cook did is he made the investment into custom silicon. And remember, at the time, there was nobody else who was really successful at custom silicon. The Japanese had been doing, you know, when they own consumer electronics, they had been doing a lot of their own silicon. But that was years ago, that was like 80s and 90s. So For them to be able to pull that off and do it so well, hats off to Tim Cook. Services were another area, kind of beyond backup and streaming. Taking the bet to get into video, I thought was going to be very hard for them. So this is where I would say I called it wrong. Innovation is in the eye of the buyer, right? Lack of innovation did not mean lack of people buying and spending more money. So let me get to the real topic here. John Ternus, insider, he's been there pretty much his entire career. I said on CNBC, he's a continuity candidate. He's not some visionary reset. He's a hardware guy. I do think we will see some massive innovation in hardware. I don't see this accelerating AI, but quite frankly, hasn't hurt them, it hasn't mattered. Like what's gonna come in and take out or reduce Apple's footprint? It would be A, a collapse in China, or B, some new pin or device, but I guarantee you Apple has 10 of these prototyped, right? Nobody's gonna catch them by surprise. The number one, win is Johnny Sruji making him in charge of all of hardware. Johnny could have gone and been a CEO and made billions of a chip company. He's staying. And I think that is super important for Apple and probably the most consequential decision, I think, over the next five years, at least.

Daniel Newman: 

Yeah, I mean, this is one of those relative 18, 1900% return over the course of his tenure. But he spent $600 billion on buybacks. And so there's a there's some question about, you know, how much innovation the company could have Delivered if those dollars have been spent in R and D instead of on, you know, reducing the size of the float as a relative basis, you know, compared to Google compared to. Vidya, his results were pretty terrible. But in terms of being maybe the world's best caretaker, I think he maybe achieved that result. The goal was continuity, preserving the culture of Apple, which I think he did well. He maximized the return on its best product, expanded the company in some areas into services. I'd say most of the hardware that came out under his watches except for the watch, maybe, or flops. I don't know when the last time you wore your Vision Pro was, Pat. The car got shut down. The home never quite turned into anything. I mean, lots of things that never materialized. But what he did do was he left before things really started to fall apart, meaning that if he is wrong, the fall hasn't really happened. There's been some criticism. Apple Intelligence was a flop. But he's leaving a strong balance sheet. He's bringing hardware people. And I mean, you're bringing hardware to the front because he needs to figure out what the future of his hardware strategy looks like. I don't think the phone's going away anytime soon, nor is the laptop, but I think Apple doesn't want to miss whatever next is. It doesn't want to be outflanked by, you know, Jony Ive or anyone else that's going to build something that's going to become the next thing for the way people interact every day with AI. And the last big question mark just comes down to, what is their AI strategy? They have Siri. They had Siri well before, you know, AI was what we think of AI as today, and Siri should have won. It should be the way, you know, I think you and I even said it, Joe said something about it yesterday. Like if Siri didn't suck, you know, we would just talk to Siri and it would, you know, it would orbit between the different apps. Like, hey, I need a car. We'd figure out which of the Uber and Lyft and different tools would be the most, you know, efficient in that given time. We would understand what our preferences are. Siri should be that front end. The fact that Siri hasn't become that front end is probably the biggest miss under under Cook's watch. But, you know, he's been a good steward to the shareholders. And I mean, Wall Street loves that. So he did a good job. He delivered. But whatever happens next, Apple needs to rotate. They need another jobs. They need an innovator. They don't need another caretaker. So that's you talked about Shruji. You know, we talked about Tarnas. Like they need to be innovators. Apple needs to return to those roots because it's had a decade and a half of caretaking.

Patrick Moorhead: 

Good stuff, Daniel. Let's go to the final decode topic. We're just ripping, kind of ripping through, maybe not. Intel, I mean, my gosh, you're up 23% right now as we speak after ripping earnings. We're going to talk about it, but it's just been a momentum story here. And, you know, Intel, apparently Elon is saying that he's going to use Intel Foundry. for his TerraFab and 14A.

Daniel Newman: 

Yeah, well, everyone's waiting for that Intel 14A customer, and they wanted that 14A go-ahead confirmation, and Lipu didn't even have to deliver it in his earnings. Elon Musk did it for him. I mean, how good is that? We heard about the TerraFab. We knew that Intel would be involved, but getting Musk to come out and basically become your spokesperson one day ahead of your big print before a 25-ish percent rip, is a pretty good thing. The TLDR here has been people have kind of acquiesced the idea that Intel would win some packaging deals and they might win some lower volume chips or network offloads or different things. And we'd heard a few announcements on 18A, but there was questions about 14A, there was questions around winning any real wafer type deals. And basically, as far as I'm concerned, This was the first person to make Intel was President Donald Trump, when he gave them $10 million and said, you are going to be the US foundry. And now the second person is going to be the world's soon-to-be first trillionaire, Elon Musk, when he basically said, we're going to go to market with Intel. So I don't have a lot more to add here. Let's just say not a bad spokesperson team to have President Trump and Elon Musk in your corner, basically committing your future. And the market is clearly, now that the company is well over 100 times forward earning, clearly believes that Intel is going to grow into this.

Patrick Moorhead: 

Yeah, here's the way I look at it. If Intel can execute that, they absolutely will use it. Let me read the full quote, and maybe the producers can flash this up. We plan to use Intel's 14a process, which is state of the art, and in fact, not yet totally complete. Given that the time TerraFab scales up, 14a will probably fairly mature or ready for primetime. 14A seems like the right move. We have a great relationship with Intel, a lot of respect for the CEO, the CTO and the new team there. We think it's going to be a great partnership. So, you know, some, they're not weasel words, they're just very pragmatic, right? Like probably, Fairly mature, right? So seems like the right move. So that's likely an LOI that's getting put together. Congratulations to Intel's continued success. There were two analysts who basically said, if Intel could do this, then they will come roaring back. People thought we were crazy, but there we go. Yeah, there were some really smart analysts that called this stuff, right? Absolutely. Hey, let's go to the next. Decode is over. Let's jump into the flip where we do a simulated debate, taking hard stances on both sides, kind of like Crossfire style for those of you who are in there. So here's what we're going to debate. Who has more power in the AI supply chain? TSMC, or the equipment companies like ASML, Applied Materials, LAM, and KLA. All right, Daniel with a beard. you are saying that TSMC holds the real power in this relationship.

Daniel Newman: 

Never updated fat Dan. Nobody cares. Nobody cares. Yeah. I mean, look, this one's pretty simple, Pat. TSMC has all the power. Sure, Intel won potentially a 14A customer sometime three years into the future, five years. I mean, it's like quantum over there. The TSMC right now, if you're talking 2nm, 3nm, 5nm, over 61% of the Q1 wafer revenue, and they basically control the entire leading edge. They have 90% of the world's advanced AI silicon, they have NVIDIA, AMD, Apple, Broadcom, Google, just to name a few. And basically every single one of them, what do they depend on? They depend on TSMC. So the bottom line is where's all the pricing power? Well, the fact is, is TSMC is the only company that can do what it does. And so when you have that kind of power, you're able to simply put a comment out into the press like they did this week. It says, hey, we're not going to move to the newest ASML equipment because we don't think we need to. And it's too expensive. And then guess what happened? ASML stock plummeted because everyone knows that there is only one company and it all goes through that company that's going to be buying the vast majority of the leading edge equipment that these equipment makers make, whether it's ASML or applied, these guys are all beholden to what goes through TSMC. And not only does TSMC control all the, you know, the wafers themselves, but basically, you know, they are the systems integrator too. I mean, look at co-ops, that's the real note, you know, you basically see that everything that's being done in these AI chips, again, now is even more dependent on the specialization that TSMC has that no one else can do. Equipment vendors, they sell tools. They basically, you know, they have the parts, they have the pieces. And sure, if other companies come through that can do what TSMC does, which there aren't any, then yeah, maybe those companies would have a little bit more asymmetric power. But as of right now, TSMC, they basically sell the AI economy. Every hyperscaler every major fabulous designer, all running all of their leading edge and next generation technologies through them. What happens if these companies stop selling the TSMC? Their revenues go to pretty much nowhere, to zero, because there's no one else to sell the equipment to. So even if they can build more, even if they could offer it, the expertise that TSMC has is so unique that they could sell it and all these machines would do is sit idle. We've talked about it countless times on this show. There's not enough talent out there. And yes, you and I may be cheering for a strong Intel. And yes, maybe in the future, Intel could fill some of these gaps, but isn't going to happen today, which basically leaves these semi-cap companies powerless and in need to continue to supply the equipment to TSMC. So sure, ASML, what you do is cool. What you do is unique. And yes, no one else can do it, but it really doesn't matter that you're that unique. It shows up in your stock price. It shows up in the continued apathy about your stock is that people, they would rather invest in TSMC because they are the real controller of the AI economy. That's all I have to say about it.

Patrick Moorhead: 

Daniel, you're essentially trying to argue that a car with no engine would be a great car, right? I mean, it's not going to run, right? And TSMC essentially runs on ASML, applied materials, LAM, KLA, and a bunch of other companies out there. And let me do the drill down. ASML is the single point of failure for every advanced node on the planet right now. Here's the other way to look at it. If TSMC no longer existed, would there be people who could eventually fill in the gap? The question is, yes, Samsung and Intel. If ASML went away and didn't do anything, who would fill their lithography gap? And then, you know, let's move out of lithography. Let's move into non-litho steps, applied materials, LAM, and KLA control. every non-litho step a fab needs here. And I think the equivalent oligopoly has a better margin structure and far less concentration risk than TSMC as well, which I think really buttresses the argument. I'm usually not this light on words, but when I know I've won an argument, there's really nothing more to talk about. I rest my case.

Daniel Newman: 

Yeah. Thanks for another easy one.

Patrick Moorhead: 

Appreciate it. That was good. So what do we really think?

Daniel Newman: 

These things are, these guys are, they're, you know, we're interdependent. I mean, they need each other. I mean, I think you and I probably tend to believe that TSMC has more outsized power because of their combination of control of the leading edge and their talent right now, but- It depends. They can't do it without the equipment, so it's like- Yeah, I mean, it depends on the definition of power.

Patrick Moorhead: 

You could replace a TSMC in three to four years if you pumped a gigantic amount of capital. I guess it's not them going away, it's just that they stopped innovating, right? You could replace them. I mean, I don't know, Chinese are trying to replace ASML with EUV. I'm not talking about, you know, the refined version of DUV. So yeah, this is a I don't know, this one's an easy one to me. I don't know.

Daniel Newman: 

Well, easy. What? What is it?

Patrick Moorhead: 

Easy? It's the equipment makers just because TSMC wouldn't do anything like I don't know.

Daniel Newman: 

But I mean, the equipment wouldn't be useful if they didn't have it. It would take half a decade to spin it up.

Patrick Moorhead: 

You know, you could sell it to you can sell the capacity to Intel. And by the way, this is why the US government isn't going nuts around the IP having to, you know, you know, TSMC foundry IP having to be done in the United States because they know there's about a 90% equipment match up between what Intel is doing and what TSMC does. So hypothetically, you could take, you know, let's say Taiwan craters and no new IP is coming out of TSMC. You could have Intel run Intel process inside of a TSMC factory in Arizona. I'm not saying it's easy, but if you look at the brands of equipment that they use, It is very freaking similar to that. And again, this was a good conversation here. But Daniel, we have 12 minutes left.

Daniel Newman: 

Yeah, and I actually stopped close to the top.

Patrick Moorhead: 

We're going to jump into bulls and bears where we talk about markets, not market texture, but what's happening out there. Let's dive in. I mean, where do you start, Daniel? Intel right now, like it's ripping at 23% right now. Talk to me.

Daniel Newman: 

I mean, look, could the analysts have been more wrong? Thank God a few of us saw it when we saw it. But like, what, how many holds and sells were there versus buys? on Intel. I mean, they and by the way, the consensus was 12 for they ended up at 13, almost six, growing on not only over consensus, but on a year over year basis, earnings 29 cents versus a one cent estimate on the street data center went up 22%. This thing's ripping after hours, guide is up. They confirmed the TerraFab, but they had a laundry list this month with the Google deals and other deals that they were able to announce. This company is selling their dead inventory, like I mentioned earlier, for huge profits. They bought back their fab capacity in Ireland because they know they're going to be building even more chips. It seems like history is out the window. The world is willing to look past all of its mistakes. Sure, it might have gotten out of memory too soon, and it slowed down production of data center CPUs before the engectic inflection. How did we all miss this, Pat? How did every one of us miss this CPU inflection?

Patrick Moorhead: 

Oh, we didn't. Some of us did hit it.

Daniel Newman: 

Yeah, I think we all, fairness, we all hit it, but nobody saw it as early as it really, if we had been beating the CPU drum 18 months ago, Intel would have slowed production down. But the bottom line, Pat, is it's very, very strong. That 22% data center number was a big surprise. Market loves it. I think we might be getting into overvalued land for a little while here. Like it might be actually overvalued at 80 a share, but I think the world is betting on Foundry and Foundry is really not priced in yet.

Patrick Moorhead: 

I don't know. I mean, Foundry, I mean, fundamentals. I mean, you said it in your tweet yesterday on EPS. not EPS, yeah, forward EPS. So it's not factored. Sorry, it is factored.

Daniel Newman: 

No, sorry. What I'm saying is there's no real numbers in the number yet. You're right. It might be baked into the price. It's not actually baked into the tape. We're not seeing much revenue. I mean, it was like $175 million of outside foundry revenue.

Patrick Moorhead: 

Yeah. Yep. Yep. No, I hear you. I pretty much knew that data center was going to rip. And the question that I had is, how are they going to do this, right? There's really, there's three levers, right? There's increased pricing. And the second is take away from earlier processors and CCG and deliver more. Um, and, um, I think the third, I forgot what I was saying. So maybe there's two ways that, um, that they could do this. So, and that's exactly what we saw, which was, um, DCAI was CPU skyrocketed. There were price raises, there were shifts, there were pull-ins, but there really wasn't enough time for Intel to get the smokestacks, uh, going again, uh, in, in. that made a difference. And the other thing that was interesting is stuff that didn't bend before, customers took, right? Let's say you didn't hit the frequency numbers or another metric might've been off, customers took it anyways. And I thought that was an interesting takeaway. So yeah, CCG, even after a really good launch, even though we all knew Panther was gonna be low volumes, They were cranking out a lot of prior gen. And then, you know, this is a momentum stock here, right? It is a, we believe this. I think we knew months ago that they were going to get some significant packaging volume because Lippu said they were increasing CapEx for it. And they had multiple, I think it was at the Morgan Stanley conference, David Zisner talked through that. So no surprises there. And then, you know, the Elon Musk, the Elon Musk comment, is really I would call the first verifiable and public wafer comment on 14a. I do believe that Apple's doing work to get ready. I do believe that NVIDIA is likely doing work. And it's that Apple and NVIDIA, not only do they need Intel the most, by the way, for different reasons, But they also have the size of the design teams to be able to pull off a dual foundry, or even if Apple or NVIDIA has them doing less than leading edge somewhere, let's say on 18 AP. I could see them spinning out a design. So yeah, just a rockin' stock right now. It's pretty amazing. So hey, let's go into a sector that I'll be honest, I ignored for a long time. But GE, that's power, right? Data center power. We keep talking about the thought that we're gonna run out of We're going to run out of power in 2028. And by the way, you have to look at inside the chain link fence in the data center, outside and connecting to the grid. And yeah, I know there are some implementations that don't even have to connect the grid when you're looking at natural gas. I had the chance at NVIDIA GTC Washington, D.C. to meet with the CEOs of both Vertiv and GE Vernova. And, you know, when these two folks are getting up on stage, this is an important phenomenon. And you look at GE Vernova, right, a beat on the top line, EPS was a beat, orders were up 71% organically. Listen to the backlog, Daniel, $163 billion backlog, and that is absolutely nuts. And we saw some similar action with Vertiv, right? Sales up 30%, a beat on EPS. And forward guidance was a beat as well, right? Raised to $14 billion and adjusted EPS was raised to around 650. Yeah, I mean, these stocks didn't rip as much as I thought they would. But then again, you know, their growth, their growth numbers, I think, were likely baked into the share price.

Daniel Newman: 

Yeah, I don't have much to add here. There's, we don't have enough power. We don't have enough chips, we don't have enough memory. We don't have enough workers to build the data centers. So I think this is an interesting category that should continue to grow. So keep moving because we got like five topics in five minutes.

Patrick Moorhead: 

So let's go man service now. What do you got Dan?

Daniel Newman: 

Yeah, okay. You know, Every quarter, I'm fortunate to have conversations with Bill McDermott, talk to him about it. Rule of 57 growth, they did have some headwinds to margin that were attached to closing three acquisitions and rolling them in. Also some headwinds into the Middle East as some on-prem impacts with the ongoing war. They beat on everything. The market's just not done digesting this idea that some vibe-coded parlor trick is going to displace the entire work automation, the entire, I guess you call it workflow automation, I don't know why the word's escaping me, that is service now. But I think the downside was all about that headwind. on the margin. Because otherwise, I think the idea that companies are going to build their own, it's just not realistic. AI run rate is up to a billion and a half, 50% above the prior target. So now Assist is executing. And I think if the Middle East clears up and we see the margins improve, I think we're going to stop seeing the baby get thrown out with the bathwater at some point. But I think it could take a few more quarters.

Patrick Moorhead: 

Yeah, you can check out. I've got a long post on LinkedIn and a little bit on X. But fundamentally, I agree with you. The thought of, you know, not only vibe coding, but testing, deploying, securing and maintaining all, you know, the core enterprise app. I just say, I just say good luck. There will be some who will try. uh, and fail. There will be some who might use 10 features out of 100 and be successful. But it still takes a tremendous amount of work. And quite frankly, you know, can you trust a company like Anthropic to lean on everything for you. And not just talking about the coding tools, but you know that they're gonna wanna get into the deployment of this stuff and pretty much rule the world. Let's get into IBM. It's interesting, very different businesses, but a similar result. IBM is a software and infrastructure company. So it's a little bit different, but pretty much, Pretty much a similar street reaction. I think the headline from investors.com said it all. IBM stock slides late despite earnings beating expectations. I just love that, right? I was thinking, what a silly season. They had a triple beat. Red Hat overall up 13%, which is higher than 8% before. Software up 11%. Data with Confluent up 19%. First quarter that's in. Oh, don't remember. February 23rd, the mainframe was dead. with Anthropic improving COBOL conversions to stuff like Java, mainframe, Z mainframe was up 48%. I cranked out a long post on this as well, but patience. Patience, Daniel. That's where we're at on here. I think you nailed that. Cranking through here, SAP, what's going on? What's going on there? Another software company.

Daniel Newman: 

Yeah, I, you know, this is still one of the core businesses that core technologies that run so many businesses, you know, strong backlog up 25%. You know, and I say that's kind of the forward booked AI business. So despite whatever the market's doing, the market continues to buy, consume, use. You remember, it's gonna take multiple years for people to, even if they wanted to transition off any of these things, it's not gonna happen overnight. They got multi-year commitments, they got massive dollars invested into these tools. But they beat on revenue. They beat on earnings. The cloud revenue is up about 19%. Their cloud backlog is up 20%. Like I mentioned, their broader backlog is up 25%. I think these are all solid numbers, Pat. I don't have… This is another one to me that indicates that there's an eyes and ears mismatch. People just aren't moving away from these enterprise software tools. They're building AI capabilities with the open AIs and Anthropics and partnering with the Googles and Microsofts and AWSs and building on top of these things. And maybe that just further reiterates what Jensen said, agentic plus SaaS, agentic plus software, not agentic instead of SaaS or software.

Patrick Moorhead: 

Yep, we live in a world of ands, not ors. Sure, a few things have died, but it's mostly they keep on going and we keep adding things to it. Obviously, there is growth in different areas, and I think that's the big question here. I don't think anybody is thinking that the terminal value of these companies is zero. I think it's a question of growth and where the software dollar investments will actually go. So Daniel, good take. Nothing swingy around the room. to add here, but, you know, the stock crashed in January on a disappointing guidance. And they just, you know, they just showed cloud acceleration quarter and reiterated every number. And final thought, enterprise ERP is boring on purpose. And that's how this one ends.

Daniel Newman: 

Yeah, absolutely.

Patrick Moorhead: 

All right, everybody, thanks for tuning in to the Six Five Pod. Hope you have a great weekend. Maybe we'll publish this thing this weekend for your delight. I did get some comments that, hey, I really liked when you published it on a Friday so we could watch it over the weekend. We're talking with our producers on that. If you have an opinion, let us know. Be part of our community. Hit that subscribe button. Take care.

‍