The Six Five Pod | EP 284: From TPUs to Trillion-Dollar Pay: Google's Ironwood Launch, AMD & Qualcomm Updates, A New AWS Deal, and the Race for AI Compute

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

1. OpenAI's Multi-Cloud Strategy: OpenAI announces a $38 billion, seven-year deal with AWS for GPU capacity. Deployment timeline: immediate start with completion expected by 2026. Pat & Dan discuss the strategic approach of partnering with multiple cloud providers (Oracle, Azure, Google, CoreWeave, AWS) and the benefits of multi-cloud dependency for OpenAI's leverage and support.

‍

2. International Chip Export Dynamics: The Trump Administration blocks Blackwell B20 chips to China. Commerce Department approves Microsoft UAE deal for 60,000 Nvidia AI chips. Hosts analyze the dual approach as part of broader trade negotiations and discuss China's actual chip needs versus political posturing.

‍

3. Google's TPU and Axion Availability: Ironwood TPU and Axion ARM VMs reach general availability. Notable deployments: Anthropic, OpenAI, MidJourney, Salesforce, and Ford. Mention of ARM-based Axion CPUs serving as head nodes for TPU systems and the validation of XPU strategy despite ongoing market skepticism.

‍

4. ARM Holdings Earnings: ARM’s triple beat on revenue, EPS, and Q3 guidance. Continued expansion beyond mobile into data center and automotive verticals.

‍

5. AMD Performance Analysis: AMD gets a triple beat with 22% data center growth and strong client segment performance. EPYC processors are showing exceptional momentum.

‍

6. Qualcomm's Diversification Success: An exceptional quadruple beat across revenue for Qualcomm, EPS, handset, and guidance metrics.  Automotive revenue exceeds the $1 billion milestone, and their data center strategy is gaining traction despite limited technical disclosures.

‍

7. Coherent's Infrastructure Play: Strong performance driven by insatiable data center demand. Transceivers and co-packaged optics are seeing massive growth.

‍

8. Lattice Semiconductor Results: Server business up 85%, networking up 63%. Data center and networking strengths are offsetting industrial and automotive market weaknesses.

‍

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.

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.

‍

Patrick Moorhead: All right. Boom. There we go. Daniel, I gotta admit, man. It's been a long march. & straight week on the road. Somehow my voice has decided to leave me here. But, no, we're. We usually don't say exactly when we're broadcasting, but it is Sunday morning, and hopefully all of you can see it on Monday. Yeah. Best you have. I. I've missed you. You know, even though I've. I've traveled to some pretty cool places in the last two weeks, I have missed my bestie, my companion. And research does show that if you know those men who do not have besties, just, they don't live longer and they're. They're pretty miserable. But I'm sure you read that and you were like, wait a second. Morehead's, like, miserable most of the time. Or let's say a little bit of the time. It could actually be worse.

‍

Daniel Newman: It's interesting to figure out if that's possible. What's keeping you from being even more miserable? I don't know. Yeah, that research was interesting, though. Basically, that we are just performative is the, you know, we perform at home, and if we don't have someone, a friend that we can sort of take out.. Our full selves on. So, hey, man, look, I'm here for you. I really didn't go anywhere. I, I. You were distant. You know, your distance was incredible last week. You know, you're up to bat and back and over just a few days, and I just was, like, zigzagging all over the place. But we got a lot done. I mean, you know, it's nice when we can all go to the same places. It's very convenient. But sometimes you have to divide, you have to conquer. But, yeah, man, and listen, Sunday mornings, I mean, what else could we want to do? What do we want to do besides POD for our amazing community? I mean, I can't think of anything I'd rather do on a Sunday than do this.

Patrick Moorhead: Yeah, I know. I hear you. You know, I mean, I'd be, like, chilling in bed, getting ready for F1, getting ready to cry about my team. That really sucked this weekend.

‍

Daniel Newman: Max, 16 today, and Yuki, like, 20.

‍

Patrick Moorhead: Yuki's 25th. Hey, I did watch Rush. I watched four straight hours of F1 last night. Just watched the stress melt off. I watched Rush. Great movie if you haven't seen it. And the other one was the new F1 movie.

‍

Daniel Newman: I watched the F1 movie.

‍

Patrick Moorhead: I feel like I did watch Rush.

‍

Daniel Newman: Was that the one about Nikki? Yeah, yeah.

‍

Patrick Moorhead: Russia was better. Rush was better than the F1.

‍

Daniel Newman: The latest must feel a lot more authentic. It didn't have like, you know, like F1 was cool, but like some of the scenes and the cars like flying like 700 meters in the air and like the things like the thing.

‍

Patrick Moorhead: Catching on fire, like, when's the last time somebody really got injured in, in Formula one?

‍

Daniel Newman: I mean, it goes back to the Senna era. That was like when that all sort of the pivot was after Senna, which by the way, another show to watch. If you saw them. Have you seen the miniseries on Netflix, the center one? If not, it was really good. There's like, it's like a three part, maybe about three hours, three and a half hours long. All about, you know, sort of his. So, yeah, I mean, but today. And I think Lando's gonna get a little closer to clinching the, the, the. The driver's championship as Oscar has recently forgotten how to drive. And then, you know, I don't know, it's weird how like one week or like one or two weeks, like in, in the U.S. like, Max was so good, the car was so dominant. And then like, you know, it does go to show though that, you know, building a car for this season is not about any one race. It really is to build something that can be good on all these different tracks. So I think that's what McLaren has gotten right this year. They built a car that even when they're not great, they're top 10, they're, they're getting points. But you know, these things come in waves, dude. But yeah, I mean, I look forward to football, a day of racing, and of course, knocking this pot out before another busy week gets started.

‍

Patrick Moorhead: I know. So let's dive in here. So we got a great show for you, folks. And we are Episode 284 OpenAI diversifies its cloud Washington D.C. titans chip rules again Microsoft's UA deal clears commerce hey, going from China, I guess, to the UAE Google turns on Irwood, TPUs and Axion as as even head nodes China's Kimmy 2 rattles cost assumptions or does it really? And if we have time, Apple's cheap map MacBook chatter is alive and well again. Apple and of course partner motions out of Cisco. And a lot of earnings.

‍

Daniel Newman: Right?

Patrick Moorhead: We had a Brutal tech sell off and we broke down earnings from ARM, AMD, Qualcomm, Coherent and Lattice.

‍

Daniel Newman: So let's today, don't forget we got a flip. And, and oh, oh, and the flip.

‍

Patrick Moorhead: Yeah, no, I appreciate that we're going to debate Elon Musk's pay package, so stay tuned. But let's dive into the decode, baby. Am I calling my number? This is so out of order here. We must have.

‍

Daniel Newman: Yeah, you know what, what happened was, is, you know, we like had the docket and like the top, the top topic that was, you were gonna have me start with was something that I felt was kind of yawn worthy. So I'm like, let's not start that with that one by the way, just. I do want to call you out before. You know, you did mention that there was a market meltdown. Did you put money into the market? I think the only reason I could think the market performed so badly. We can talk about that later. But I was just thinking, was it possible that you timed it as perfectly as you ever did?

‍

Patrick Moorhead: No, I timed it as perfectly as I usually do. I bought it on Monday. Open good job. And watched everything melt.

‍

Daniel Newman: So it'll be, it'll be better. 

‍

Patrick Moorhead: Open AI seemingly cut deals with everybody, right? Oracle, Azure, Google, CoreWeave, a bunch of different Neo clouds. And I'm sure I've forgotten somebody on the hyperscaler side, but. So the meme was, wow, AWS is so far behind, they don't know what they're doing. They are not doing anything. They're the old guard and the new guard is everybody else. But here we go. Boom. OpenAI and AWS now have a seven year, $38 billion deal for GPUs. And it starts now. And the entire build out should be done, should be complete in 2026. So hundreds and thousands of Nvidia GPUs using AWS's network and these super pods they put together that are seriously kick butt. So I just thought that, I mean, listen, this is a big deal. It just shows that I kind of, I'm kind of buying into this whole notion, Daniel, that OpenAI by having everybody on the hook for something, they're all going to be pushing for it, right? And anytime OpenAI needs help in this, somebody will come in and help them, help help them out. So I, I don't think there's any losers here at all. There wasn't any discussion about Trainium. But you know, after the gigantic deal with anthropic, the million TPUs out there and OpenAI's embrace of XPUs overall, I can't imagine that this wouldn't be in the cards because the native OpenAI XPU is a few years off. So yeah, OpenAI is going completely multi cloud and bringing the big checkbook to boot.

‍

Daniel Newman: Yeah, this has always been a bit of a race to maximize capacity, making sure that it is in a position that it never is out in the cold. You sort of pin all these clouds and all these compute makers against each other and yet come out with the advantage of being the chosen one. You know there was something this week that was kind of like a comment that someone made that maybe wasn't completely accurate about wanting a backstop. The CFO of OpenAI saying they want a government backstop for, you know, but the question is, is, are they too big to fail? I mean, yeah, that might be the case. Like right now at this point they've now tied, hitched their wagon to Oracle, to Microsoft, to Google, to aws, they've hitched their wagon to AMD, to Nvidia, you know, on the chip side also Broadcom or you know, to do custom chips. I mean they are basically just, you know, smattering out and all these companies are now dependent on them being able to come through. Yeah, because they're all committing, you know, you think about like they're all committing these huge revenue numbers and you know, if there's any doubt in the market that exists, that really exists about is it a bubble, it's really whether or not OpenAI can pay the obligations. I mean that's the one that's like where does the cash come from? Yeah, it was interesting. I actually talked to this week, I was at an event and I talked to a few OpenAI investors that were in open AI at varying spaces but at varying numbers. But one of them was over 300 million. I mean a big fund. And I, I just said to him, I said, you know, do you think there's any valuation where there won't be money chasing equity in this company? You know the, the, of course these people are talking about their own book. So it's not like he's going to be like oh yeah, it's going to collapse at 500. But I mean the amount of confidence in which, you know, these people indicated the over subscription level of these kinds of deals is pretty incredible that there's like so much money chasing in. And what I'm saying is like, you know, the trillion dollar IPO is not crazy at this point. Like a trillion dollar IPO for a company, by the way. It doesn't, it doesn't create, you know, as much income as McDonald's. It doesn't create as much as, you know, it doesn't create as much income as, well, it doesn't create as much income as you and, and me in our companies right now. It creates less, just to be clear. But the scale is enormous. So anyways, you know, I think they're, they're, they're addressing the need. The need is capacity and they want to have as much as possible. And they realize at the end of this game it's whoever has the most compute wins. But I also think, and I'll leave it at this thought, but I also think it still comes down to OpenAI believing it's going to be a lot more than just models. It's going to be a hyperscaler. They want to be the software company for enterprises in the future. They want to be the device company, they're going to be the search company, they want to be the social media platform. Like they believe that they are the super Corp or the future. Watch, Mr. Robot. They are Evil Corp. They are. That is the goal. They want to be Evil Corp.

‍

Patrick Moorhead: So I love that. Good breakdown, Daniel. So, hey, we saw two things on the international EX export for GPUs. Trump basically said, no, we're not selling black wall for China. This cut down B20. But Commerce did approve Microsoft UAE deal for 60,000 Nvidia AI chips. So Daniel, what's up with the, what's up with the duality here?

‍

Daniel Newman: Oh man, this is like, this is all a big theater. I mean, this is all part of the bigger trade negotiation that's going on. I mean, suddenly over the last two weeks, President Trump and Xi Jinping are besties. They love each other. They have ultimate utmost respect for one another. You know, Trump's reducing tariffs on China. You know, I mean, what's next? They're going to start letting them ship the BYDs into the US and we're going to be shipping them first generation Blackwells. And you know, I don't think that's happening. But I, what, here's the thing is, I think in the end this is all a lot of tit for tat. And this is going to end up with something that's going to be, just, by the way, the entire Trump trade. People call it the Taco trademark. I think it's called negotiation. Like people want to make, like, what do you want them to not chicken out at any level? Do you want them to be held accountable and just blow the economy up. Of course he's not going to do that, but he's, you know, he's pulling those hard levers and he's landing somewhere in the middle. And, you know, I could see the situation where the, you know, the B30 does land or something else. But, you know, my take is we're going to land where, you know, if you kind of really do the generational thing right in age 20 versus like a Blackwell Ultra is like four generations. It's like, you know, about four generations behind. Because you had, you know, the first generation Blackwell is. You had the H200, H100, H20. So it's like the fourth generation behind B30 is what, about half the throughput? It's about half the throughput of Blackwell. You know, it's probably closer to like an H200, you know, in terms of its quality. And so, you know, my reading is, like I said, I see a situation where when we get the rare Earths deal that we want in place and feel comfortable that B30s could be released into the wild. But then the other side of this argument, Pat, is like, does China even want them? Because China's doing such great work. They're building these amazing Huawei chips that are much better than anything the US can build. And then they're training models for $12 using two Pentium 2 chips, or what did you call it, three Oura rings. And so maybe they don't need our chips anymore. And so that's the other big risk point. But here's what I think. I think Jensen's playing a great game of leadership in terms of what his role is to drive Nvidia growth. And his goal is to sell every chip he can to every part of the world that he can, you know, within the constraint. I don't think he's trying to do anything that would be, you know, nefarious or where the chips would be used for negative. But, like, opening China's market to Nvidia is good for him, good for investors, period. But I think the US continues to believe that we have a technological lead that we don't want to give too much too soon to China. And I do believe in what Jensen is saying is China's research arm is incredible, meaning that they have the volume of researchers, and if we give them the blueprints to do this, they will beat us like they have in EVs, like they have in drones, like they will in humanoids. I do see all those things. So I think we need to keep our, and protect our lead here. But I think in the end we're going to land one or two generations behind where they will be able to get some access to some of these tune down chips. But I still think there could be more tit for tat negotiations before that happens.

‍

Patrick Moorhead: Yeah, it's interesting Daniel. I'm feeling like this one is a lot more real than anything I've, I, I've seen before. And I don't know at this point what is, what is part of the negotiation. Like what does, what does Trump want or what does Commerce or Treasury actually want out of China? There is a massive imbalance in trade or you know, is slowing them down. You know, truly the, truly the strategy, I mean make no mistake the People's Liberation army is not going to use an Nvidia chip in, in their ecosystem. First of all, there's this so-called back door which I don't believe and why would they do this? And yeah, you could integrate any into some missile system or something like that, but you're not going to choose a US based chip or system to do that. You have to build an entirely brand new stack and if you're going to do that you're just going to use Huawei or, or one of the other XPU companies that are, that are, that are out there. And I do think we should enable the certified CSPs. And what I mean by certified, it's knowing the who, what, when, where. I researched this while I was on the plane for 10 hours coming back from Europe and you know, there are people who, who do KYC down to palm print on, on accessing systems in a country that's, that's questionable. Daniel, do you find any, any irony or you know, is us allowing shipments to the UAE with Microsoft on Blackwell, is that kind of a, you know, a second prize in the BB contest here?

‍

Daniel Newman: Oh those beauty contests. So much work, so little return. I think the point is, is that, you know, we've largely gone to the UAE, including President Trump and the administration and said hey, we are in a good place to collaborate, to partner and to invest in building this out. You know, the trade situation is very different there. Of course, you know, oil would be the most interesting potential trade risk and they of course want to sell oil to us. But I think the China relationship is more complex. China's a real threat to our global economic leadership and domination. The UAE is not. Yeah. So I think that that's a concession. It's something that of course allows some of the growth. Because right now, I mean the one thing that is true Pat, is like China, we may or may not be able to sell directly into China. We could argue whether anything's getting into China or not right now. But like Jensen and Nvidia can't really make any more than they're making right now. Like they're shipping everything they can make. So if all of a sudden China did open up, like where the capacity is going to come from and where is it? Where are they going to, you know, where are they going to reroute capacity from? Like the capacity that's going where. Because if we started using what TSMCs, I'm pretty sure I heard CC way come out this week and say basically they're going to have record quarters forever and that, you know, basically everybody comes here and only wants one thing and that's more wafer. So even if we open it up, where does more revenue come from? How do they even, you know, I don't know but like at least the checks I'm doing, I don't think we can make more right now without building more fabs. And you can't do that in a week.

‍

Patrick Moorhead: Yeah. And at the end, you know, once China burns through all those TSMC wafers that they die banked, Huawei dies banked. Where are they going to get the HBM? Where are they going to get or will it be like these LP DDR designs? I mean by restricting China. What I am with you on is that this BS about them, not.

Daniel Newman:.

‍

Patrick Moorhead: Not needing Nvidia makes absolutely no, makes absolutely no sense to me whatsoever. But they do need them big time.

‍

Daniel Newman: Just the plot, just the, you know, again, just agree with the software itself, it's just there's too much work to be done, you know, before can or anything else. I mean like I said, because if software was easy, you know, Trainium would be further along. If software is easy, AMD would be taking more market share or growing faster because just people would be eating it up. But like the software is sticky and these researchers there are no different than the ones in the US that don't want to use anything else. It's like, you know, necessity is the mother of invention. Like don't get me wrong but like I'm just still not sure that they don't have access. So, and if they're training, by the way, if they don't need, if it's only $4 million worth of training now on a couple of seven generation old Opteron servers then you know, they could do that in the cloud. Like they could easily do that without actually having to bring hardware into China.

‍

Patrick Moorhead: Yeah, yeah.

‍

Daniel Newman: So I don't know.

‍

Patrick Moorhead: Now let's move to our next reference.

‍

Daniel Newman: Pat, I just did that for you.

‍

Patrick Moorhead: I appreciate that. You do one every time. Hey, let's move to Google making Ironwood TPU and axiom ARM VMs generally available. So this was announced I believe at Google Cloud next, but these are available and we did some interviews when we were there. The 6.5 was there to cover it and some notable things for me that stood out. And by the way, TPUs are very different at this point. Right. With all the wins that they're racking up at this point and all the people who are developing for TPUs aside from Google, I hope this puts the XPU conversation to bed. I mean I'm still debating people on X and maybe that's, maybe that's, you know, I should just ignore it who are just saying nobody needs XPUs for about 12 different reasons. And it kind of reminds me of back in the day of these public cloud deniers, right? They're going to give me all the reasons why people aren't going to go to the public cloud. And then you see this monster growth and you know, there are choices. GPUs are going to be best for flexibility. You'll probably get more, I would say, versions of software out of them, they're easier to redeploy but if you narrow in and you know, you know what you want to do, they are going to be cheaper and they are going to be more efficient out there. So I think they talked about search, YouTube, Google, Workstate, workspace inference. That's, that's done. There's, I think there's a, a scaling path that they put in there as well, kind of hinting at the, at the future. So yeah, I mean it's real. XPUs are real. You can deny it all you want. Google is absolutely ahead with TPU in terms of at least the forecast that I have seen. And I would say that AWS is second and Microsoft is still trying to pump out Maya with Cobalt. The final comment that I thought was interesting too is that the head node for this combination is an ARM based Axiom CPU.

‍

Daniel Newman: Yeah, I mean this was where it was always going all along. Google opened the door to. We saw Anthropic recently made a massive commitment to build on TPU. We know that all of these clouds are going to go diversified. You and I have kicked this idea around. This is not like a core victory lap, but let's just call it a simple enough victory lap to say we always said it would be and not. Or yeah, it always was and always will be. The biggest cloud companies on the planet are going to build these and then of course if they build them and make the software flexible enough and usable enough, why in the world would they not make capacity available? It's not going to be destroyed. There are not too many stupid headlines out there that Google's TPU, you know, is going to, you know, be out there to compete with Nvidia at the high end. It's like, yeah, you know, there might be some competition there, but that's really not the point. It's really not the reason they're building it. They're building it to vertically, vertically integrate their business, increase their margins, increase their control in their applications, optimize their workloads and of course do it with as much margin, like I said, as possible. And you know, as these things get better, of course they're going to take their innovations inward and roll them outward. That's how the cloud has worked quite a bit. In fact, that's how the cloud actually became in existence as they built something for themselves and said holy crap, we could rent this to other people. So I'll leave it at that.

‍

Patrick Moorhead: Yeah, it's interesting. We've got anthropic, we've got open AI, we've got Mid Journey. So you know, we will, we will see. And I believe Salesforce, Ford Motor Company and a couple others are using it, which again is, is, is fascinating. Then somebody will say, well Nvidia can't, can't make enough. And I guess my response is that TSMC makes all of them. So it's not like Nvidia couldn't be the winner with everything. Right. They can take the wafers and the packaging from, from, from everybody else.

‍

Daniel Newman: So yeah, until Intel starts making high volume Samsung. Right?

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: Didn't Musk say he's going to work with Intel?

‍

Patrick Moorhead: You know, Musk says a lot, but he said something kind of like that. But we will see. Maybe he's just negotiating with Samsung. I have, I have no idea. Right. Probably.

‍

Daniel Newman: I don't. We'll see.

‍

Patrick Moorhead: All right, hey, let's go into the next topic here. And, and Dan, this is kind of your, this is kind of your favorite new model called Kimmy 2 from Moonshot AI over in China cranked out a pretty good, really good Open source reasoning model and I hear that they trained it for $4.6 million, which means this CapEx, the trillion dollars in CapEx is pretty much just because we suck at software and China is better at it.

‍

Daniel Newman: Yeah, China's, you know, MacGyver. They're the MacGyver of AI over there. And that's why I said they don't need any Nvidia chips because they don't need any chips at all. They're going to use, you know, a couple of Mac Minis and they're going to run the world with it. Pat, I, I've said, God, all I can say about this, this is just total nonsense. It's, yeah, there might be something that was trained for 4.6 million. I'm not saying they didn't do something, I'm saying like, but what I want to understand is just like this is, does this not feel like deep seq v2, 3, 4, 5? Like you know, there was a lot of stuff done and then maybe like the test time optimizations that were done that allowed for, you know, inference in, in, in the reasoning strand as opposed to inference, you know, all done in, in having to be developed in the pre training. But I think what they're doing is taking models that were trained for hundreds of millions of dollars, if not more and then they're doing little additional, you know, functions to them to make them work. And I think that is part of the evolution of it. But that initial spend is kind of being overlooked. I don't understand why this narrative just keeps getting shared. Like if this was the case, this means that Satya is wrong, Sundar's wrong, it means that Mark is wrong, it means that Jensen's wrong, it means that the, you know, the entire strategy of our US government is wrong. It means that everything right now that we're doing is wrong. And that there was some little lab over in China that figured out what all these companies and all these researchers could not figure out. And by the way, there's an incentive for Google, Microsoft, Amazon, Other, you know, Anthropic, even OpenAI to figure these things out. It's actually to their incentive because spending all the money they spend with Nvidia is not to their benefit. They do not make money on that. Like if OpenAI pad didn't have to spend all this money on infrastructure, they could be a lot more profitable a lot sooner. So like I said, it's so absurd to me. Like this isn't like, oh, they figured out how to go from a billion a model to 400 million and cut 60%. They're like, you know, I figured out if I swallowed two Tylenol that I could turn into the Red Hulk. I mean, that's like what they, you know, they figured out right here. So I. Only because of my own sardonic nature do I even give it this much oxygen. But this is a horse.

‍

Patrick Moorhead: I like that. I heard you say that before.

‍

Daniel Newman: It's a new one.

‍

Patrick Moorhead: It is a new one coming out.

‍

Patrick Moorhead: So shareholders approved an unprecedented plan potentially worth up to $1 trillion over 10 years, tied to performance milestones. You got market cap, unit, units. I almost said unix, robotaxian robots and, and retention. And guess what? It. It got. Got approved. Which again I'm seeing on Twitter versus. So we're going to flip the coin here. Is this fair, is this good or is this bad? Let's flip the coin here.

‍

Daniel Newman: So you're for the pay pack. You are the for.

‍

Patrick Moorhead: Yeah. So at bottom line, none of this gets paid without delivering on all the milestones. Capex, market share, robots everything and Robotaxi. Right. So, you know, we debated his pay package up to this point, which by the way was never approved. A court in Delaware nixed it and he didn't get all that money for doing obscene valuations here. But it's true. So that's the first thing he doesn't get if he doesn't actually do all the moonshots. Very fair. And I think it's a good package too. It gets, you know, keeps Musk's focus on Tesla across all the different companies like SpaceX, XAI, Neuralink and X. A lot of people refer to those as distractions. And if you look at, you know, the 56 billion dollar plan that he hit on it aligned incentives and you know, coincided with an amazing stock price. So this is kind of the sequel to that. I think the pivot, like you look at how difficult this is, the pivot to robotics here is, is absolutely huge. And that's not only for bipeds, but also for Robo Taxi. And I think the shareholders chose, I mean 75% approval said that they for the most part agree there were some major shareholders that disagreed. But I think that in itself, I mean we are a democracy. It's pay for performance. I see absolutely no reason why this would be anything but a thumbs up. Daniel, try to argue against this thing. Good luck.

‍

Daniel Newman: I think that paying any person a trillion dollars is just absolutely an absurdity. You know, the pay package itself, you know, Pat, you can have a trillion dollar pay package too. If you take more insights from, you know, your size and you make it, you know, 10,000 times larger, then you can have that as well. This is a gimmick. It doesn't even matter if he ever gets paid or not. It's such an unreasonable, unlikely and improbable thing that, you know, making a package like this is great for headlines, it's great for creating retail sentiment with investors, maybe get you a little near term share price gain, but you know, there's nothing in it that he's going to be able to accomplish. So you know, China takes the robots, you know, EVs, nobody even wants them anyways. Once the tax credit went away, the unit volume has sunk. You know, there's other companies doing energy, there's other companies doing robots, there's other companies. So, you know, the bottom line is that, you know, a trillion dollar pay package as a PR stunt is a lot of fun. But the bottom line is, is that in your lifetime and in my lifetime, they're not getting anywhere close to these numbers anyway. So I'm also going to announce a trillion dollar pay package for myself as CEO of Futurum on Monday. And I'm going to outline the absurd and unrealistic things I'm. And then maybe I can get a bunch of people to, you know, like my tweet. So that's all it is. This is a PR stunt. This is silliness. These numbers aren't even real. And anybody can make a nice pay package for themselves when they're willing to be completely unrealistic about what the business is capable of doing. So good luck. You know, this is just a mockery of executive compensation anyway, so I think.

‍

Patrick Moorhead: That they should give you a trillion dollar package.

‍

Daniel Newman: I think if I hit certain numbers, if I grow the business, you know, by 10,000 times, I, maybe, maybe a hundred thousand times. But I mean, look, I'm actually like, yeah, sure, why not? You know, fine, give it to him. Like if he can hit these numbers, pay the guy, like, pardon my language, this guy. You, you, you. Because you know, I did this, an assessment last night. I'm like, he's going to take a one and a half trillion dollar company and make it worth more than Apple and Nvidia combined. If he does that, like pay the man, like, I don't know, he's worth 400 billion now. What's the difference of like another 600 billion of, of, of net worth at that point? Like, sure. I mean, if he, you know, does that and he's willing to basically say, like, I'm gonna hold myself to that, yeah, great, Pam.  I'm all for it, man. I mean, live the dream. I do think it's insane times now. Like a billion is nothing. Like a trillion. A thousand billion. Yeah.

‍

Patrick Moorhead: I need a package, need me a pay package like that.

‍

Daniel Newman: Good gosh, good gosh. I mean, like I said, but if he does that, you know, because it would be bigger, by the way, he would have to add meta, Amazon and Google in market cap. He needs to take it from where he is and add a meta, an Amazon and a Google in market cap.

‍

Patrick Moorhead: Crazy.

‍

Daniel Newman: I mean, I'm just saying like, if he does that, pay the guy, like heck, you know, if you and I did that much, I think we'd both want to be paid. Well, I don't know. I have no issue with it, but I just, I do think it's been, I do think it's a bit of a PR thing. Like, you know, because I would guess that if Satya or if Jensen grows there, if, if, because essentially this would be, if, like, Jensen grows Nvidia to 20 + trillion market cap, how much would he make? Because it's like the same thing. You know what I mean? Like, it's kind of one of those things that's like all the guys basically saying is, I'm a moonshot. I can do something massive, and if I do, I want to be paid. Like, I want to be paid extremely well if I accomplish this, you know, but seems very mean anyway. I don't know. I don't know. This Sunday. I don't even know if I have enough energy to fight this, this level of kind of silliness, absurdity. I'm gonna let it go. I'm like, I don't know. What do you want to do next? What do you want to do next?

‍

Patrick Moorhead: Yeah, so let's do bulls and bears. We definitely had a market meltdown and a bunch of earnings, but let's, let's dive in. Daniel. All right.

‍

Patrick Moorhead: Market, Keep the bed. Daniel. We're definitely in a bubble that just had a little, a micro tear in there. And, yeah, the worst week since April for tech and NASDAQ valuation resets. Are we doomed? I mean, you know, when I started putting money in on Monday, market open, I, I actually caused this. Daniel.

‍

Daniel Newman: I think you did. I feel like, because it was real, like, because it didn't go up, like, for a day, and then it went, like, hard down, though. Like, the first day you were in, you're like, oh, this is good. I like this. This works.

‍

Patrick Moorhead: And then, no, it didn't even do that because, you know, I bought it market open.

‍

Daniel Newman: Oh, I thought Tuesday was, like, scary. Scaremageddon was on Tuesday when it really kind of had that hard fall. And then on Wednesday, everything came back hard. Then Thursday, it sold hard. Friday, it sold hard then. But actually, by the end of Friday, it was up. It was up. I mean, Pat, I actually attribute most of this to government shutdown. I think right now what we're getting is, you know, there's a little uncertainty. Consumers are going to start to get hit. We have a massive federal government. People are going to start to get nervous about spending money. We have limited data because the government is being shut down. So we can't actually see everything that's going on in terms of employment, in terms of inflation. And I think that it's a good excuse right now with this uncertainty for a little bit of selling. I still think it's a. I still think it's going to rip higher, but I think we could see a lot of volatility and turbulence until the government reopens. I mean, we're at a, what, one of the longest shutdowns ever right now. And there does not feel, it does not feel like there is a lot of desire on either side of the aisle. We had some big Democratic wins in some races this week. I think that threw some more cold water on the idea that maybe this current administration and its very, very business friendly policies are going to sustain that. We may see a little bit of a pendulum swing coming back at the midterm. And right now it seems that the Democrats are willing to dig in and basically wait. I mean, even if people snap, benefits aren't getting paid right now. But when all of a sudden Thanksgiving comes, Pat, and you know, people's travel, I mean, if the ATC is smart, they'll walk out in Thanksgiving week. They'll just, that'll end this thing. Unfortunately, it'll ruin some of our lives in terms of, you know, our travel and destination. But, you know, when you start to really punish the average consumer and then that starts to get associated with a different party. But then again, both parties are doing a pretty good job of blaming the other for all of this. So it's not even really evident to me who is at fault. But I'll say one thing before we actually talk about the earnings, Pat, is tech earnings have been good. There's not been, I can't really think of a company that hasn't had good earnings yet. And not just the earnings, the guidance. So what I'm saying is that our valuations stretched. Yeah, you could argue that some valuations are stretched. Has there been an incredible face melting rally upwards that you decided to get in on at the very, very tail after like 300% runs for a lot of names? I mean, that part is, is, is real. And so I was laughing like, you know, let's not even talk about just pure tech. We'll talk about speculative names like Robin Hood. It was up 300%, so it went down by 10 or 15. Now it's up 260 for the year. And everyone's like crying or, you know, you look at, you know, you look at a Broadcom. I mean, it was getting close to 400, but I remember we were looking at this thing for 250 like four months ago. It was under 200 in April. Companies like this don't double in a year. Like that's not normal. So is it a, you know, so going back to the whole, I actually think this is very real. I think the forecasts are very real. You look at things like memory storage controllers, you look at energy turbines, thermals, all the things that are indicative of how long this builds out. And I think I shared something yesterday. I'll give credit where it's due. I saw it on Ben Baharan's Twitter but it was about like a Microsoft exec that was basically talking about seeing literally no slowdown and build out of data centers before 2020. So that basically means if this was the Internet bubble, if you want to use that analogy, we're in 1996 right now. We're not in, you know, we're in 97, we're not in 2000 or 2001. We are literally not even close to the part where we start to feel like demand and supply are out of, out of whack. We are still building and the earnings to me are indicative that it is. My biggest worry though is like I said, shut down. So I know I went on a little long. Here is where consumers start to get hurt at some point. Like where you know when jobs aren't strong, when the economy is not strong, when you have 10 + percent of your workforce is the federal government. Those people aren't getting paid like some point the consumer gets hurt and whenever that happens you're going to see the market react.

Patrick Moorhead: Yeah, well let's dive into those amazing earnings. ARM had an incredible print right. They had a triple beat revenue EPS Q3 guide and most part got upward revisions. And their story is very straightforward and continues. Not only are they getting into new markets but the markets they're in, they're growing and they're adding even more value with programs and products like, like css. And this was one of the first ones you Daniel where like you know it seemed perfect, perfect, perfect. And then the stock was dumped and I believe it was a positive reaction. It was hard to separate the doomers out there for the week and, and how they performed. But you know the upward revisions from, from all these, I mean Evercore, KeyBank, TD, Cowan, Mizu, JPM, UBS, all you know increased price expectations on there. So it's a great story. And oh by the way we haven't even seen details on Asics chip chipsets or SoCs that we keep hearing hints of. I think a great case study here is Google. I mean not only overall hyperscaler Data shares around 50% on, on the CPU side but Google 30,000 internal apps have been migrated to ARM. Axion 22 you know, million developers out there on, on ARM, others gaining share out. Every time Apple gains share, they gain share in PCs regardless of how ARM based AI PCs are going. And yeah, there were questions. I mean of course it's not just a love fest here, right? There were questions and concerns around arms, China exposure, licensing pipeline deal cycle timing. But I feel like those are just noise. There was nothing new for, for anything, you know, and the scrutiny over SoftBank related revenue and its impact on the overall results, I look at that as reaching for finding some challenge. And you know, financial analysts, it's the reason they get, they get paid but solid earnings.

‍

Daniel Newman: Yeah, I mean we, I don't, I don't know if you had the chance, had the chance to speak to Renee usually we both do. You know, this one was just good. I mean the thing I love too is there's a story still evolving, the expansion. But like the CSS expansion margin expansion, record royalties strength in terms of each of the cloud providers growing. What I love about it is it's not so much a, it was always more of a handset low power story and it's really finding its way more and more into the data center. We talked about Axioms earlier. You know, of course the Graviton business is growing very, very well and there's an AI play here, there's an AI play that's coming and, and I just don't, don't underestimate. I think ARM has a role to play and I think this will be a bit thematic again because people are gonna be like, oh, it's, it's arm's gonna compete with Nvidia. It's like, stop it. You know, we'll have that talk I'm sure when we hit Qualcomm too. Same thing, but great result for, for arm. Another, you know, indicator again looking for the weakness in the earnings and, and this wasn't it.

Patrick Moorhead: Yep. So,let's dive into AMD. Basically a triple beat. Daniel?

‍

Daniel Newman: Yeah, same thing. I mean, you know, data center is up, you know, 22% strong there. You know, the client and PC are even better. Epic, very, very strong. You know, the nit on this one, and I said this in public, it got picked up a lot was the 22% data center growth. We know EPIC is crushing it. We know AMD is doing amazingly well with Epic. I think there's a lot of worries about what that meant year over year on instant growth growth. And I think that's probably after the company had this massive melt up. I mean it just absolutely. I don't, I don't know off the top of my head, but it was up like about 50. I mean it's a huge number. It was up after the open AI deal, after the Oracle announcements. I mean AMD has just been on absolute fire. But I think people are just over estimating how fast that sort of transition is. I think AMD's big AI GPU inflection is Rack scale. I think it has to be this Mi 450. I think that's where we're going to really get a sense of how powerful and how successful and how competitive AMD's portfolio is. But I think that also was the number everyone had eyes on. So in terms of performance against expectations, the street, I think AMD did great, continues to do. So I think where you're going to get the net is that people want these huge growths in mi. They want huge growth in the data center GPU side. And we're just not getting enough clarity there to be 100 certain where it stands. But I think that's actually going to come more next year. That's where I sit on that one.

‍

Patrick Moorhead: Yeah. I mean 100. I mean people were expecting an Mi450 parabolic takeoff here and it's the second half of 2025. We're literally nine months to a year away from this acceleration. And the one thing that just people don't understand is AMD just currently does not have a Rack scale, Rack scale solution where you can address more than eight GPUs. Blackwell, you can address 177 moving to 144 coming up here. And what that does is that essentially gives you workloads, particularly training workloads, slow down because you have to go across a network to string all these things together.

‍

Daniel Newman: So you said 177 going to 144. Did you mean 72 heading to 144?

‍

Patrick Moorhead: Yeah, sorry about that.

‍

Daniel Newman: No, I just want to make sure. Just because I. Yeah, you're starting to question my own. Okay.

‍

Patrick Moorhead: Just I, I am jet lag and I'm definitely not on this morning.

‍

Daniel Newman: You're doing okay. I'd give you like a D minus.

‍

Patrick Moorhead: Yeah, no, I appreciate that. It's better than enough. Let me go hang myself right now. I did want to give the client folks over at AMD some credit. I mean 70, 73% improvement here. So they've kind of lumped clients and gaming together and I think that's to potentially not shield, but the gaming number is, is a lot smaller. They do bring it out. I mean gaming did almost triple year over year. The consoles have finally kicked in here. So yeah, you have to give credit but that's just not what the street is looking for. It's not what the street cares about because it's not a direct comparison. And you know, when, when Nvidia is given a 500 billion dollar five to six quarter read, I mean if you're not putting up, you know, $100 billion forecast or, or floating at everybody, people will not be happy. Let's move on to Qualcomm. I mean, oh my gosh, I mean I'm going to call this a quadruple beat, right? Revenue EPS, Qt QCT handsets and, and Q1 guide got mostly upward revisions. You know it's funny Stacy Raskin calls you know, Qualcomm the company, the, the Rodney Dangerfield company, the company respect and they're starting to get some respect as well. There are some things that are weighing on it, right. China losing Apple. Although Qualcomm can't be clearer on Apple and what that means. In fact they have a slide that shows non Apple like literally the word Apple on, on a slide. I think that investors also got a little bit spooked even though Qualcomm said hey we're going to hit 75 premium shares for Samsung versus 100. They even question well what do you mean 75, why is it not 65, why is it not 55? So it was just, it's just weird at the same time, you know auto over $1 billion this notion of data center and, and the numbers behind that I think the company once they start to give more details I would say technical details on their data center products. I think that the market and the street will actually lock the gear. The other thing that the company did is they, they essentially said hey, here's what we told you at investor day and here's how we did and we're ahead of everything we told you at, at an investor day. And I thought that was again a very high degree of disclosures out there. I think the two that are going to get people very excited is where the market is rocking, which is the data center. And I think the next one and I think you agree with me, will be. Will be Robotics.

‍

Daniel Newman: Yeah I think you hit the current. Well I mean they did have that little tax adjustment. They made the same, did the same thing Meta did. Locked in long term lower tax rates, good for cash flow, non cash item on this quarter the AI play the silliness is everybody saying oh it's, they're trying to compare it to like Blackwell Ultra. It's like I think, you know, this is a next gen grok play. Like let's go build high volume inference, you know, LPDDR memory. They're doing their own little techniques and strategy. It's inference centric. It'll probably be super successful in the Middle East. It'll probably, you know, some neo cloud scale and then of course enterprise cloud scale opportunities. And I keep saying like 1 or 2% is a 5 or $10 billion swing in their incremental business. So really, really meaningful. I think they've got all the right components. Been saying my kind of long term bull thesis on Qualcomm is they've got all the components to play a role in lower end humanoid robotics and do that well. And then you know I, I mean my, my overall read is just this company has ex executed pretty well against its multi year Apple fade. You know, the Apple fade. And what are they going to do and how do they diversify out and pivot away from being a handset company? I mean they have the handset business but that's, you know, they want to be valued differently and part of why they're so interesting is they're valued so frankly they just valued so low. They're valued worse than a PC company, like a pure PC company. I mean 13, 14, 15 forward. You know, I think that's a third of what AMD is, you know and that varies from time to time. But like if they can get some credit for some of these adjacencies, they have a lot of upside if you're an investor. Something to, something to keep an eye on. But they have to execute. That's the thing I'll say is they, they have to execute. You know they got that first deal in AI Cloud with their AI 200 series. I think one or two more announcements will go a really long way to proving that they can, they can win meaningful business in the space. And I expect they will. Will. Yeah.

‍

Patrick Moorhead: Good, good print there. Qualcomm. Hey, let's go into Coherent. Daniel, I think you and I got briefed on that at the same time. I was barely awake.

‍

Daniel Newman: Laser beams. Laser beams, yeah. I mean there's a lot going on there. But this is really interesting. I call them, I call these like the breadcrumbs companies. The ones that like, you look at their business and their outlooks and you know, because you basically have, you know, data centers connecting to other data centers. You've got Coppery, but of course then you got light, you got co packaged optics, you've got a company like Coherent that's building a lot of the receiver transceivers that allow this data to move, scale out, scale across. And what I'm hearing, and what we're hearing is basically the insatiable demand. This is a company that does other things too, like thermal, a lot of material science that they do. And so when you look at things like copper plates and GPUs and cooling that they're building new next generation technologies that should help. But long story short is they're increasing margin, they're increasing their earnings, they're increasing revenue substantially and they're basically sold out. And what I think you and I both have agreed and said in the past that we really like about is besides the really strong demand, this is also a company that has built a very resilient supply chain with minimal dependence on China. Was able to get out of that sort of issue early on in this trade, in the trade conflict. And so their capacity and their ability to scale is really more in their own destiny with less risk related to outside forces. So I think this is one of those companies you got to kind of watch Pat. And this demand, especially around these transceivers to move data across data centers and within data centers is going to be a massive boondoggle for a long period of time. And they're supplying and they supply to a lot of the companies that we've talked about on the show. So that's the thing is like they don't say who they supply to but if you start to insinuate these big name companies that are building these big rack scale data centers and connecting these data centers, this is the clients of Coherent. So you have to like this as a play upstream if you believe in this data center build out and this continued growth and need to move data faster from, from compute to compute tile.

‍

Patrick Moorhead: Yeah, I mean this, this company does a lot of different things down to you know, being in the OLED and LED display business where it uses its lasers to cut and you know, make the proper sized display for different devices. But 69% of the business is data center and comms and that's where it's rocking. So as goes hyperscaler, Capex goes Coherent and that is up, right? You pretty much had everybody from Rosenblatt, Barclays, Benchmark, Stifel, Needham, Morgan Stanley who all raised their price targets. The only Downgrade was Northland who, you know, moved it, downgraded it to market perform, which just is, is, is, is, is insane to me. A stock that's 159, they're going to put it 125 with all this gigantic growth, right? And they're, you know, doing the 800 gig to 1.6, you know, Tera ramps. They've already sampled, you know, the CPO scaredy cats. Cpo, npo. Lasers are sampling margins, you know, should raise, but they're going to have to get, get capacity. And you know, they're nice people. And I think Jim in particular is not going to come right out and say we are the biggest manufacturer in the United States and, and others do it in China. But I wish he would, but, but he doesn't. And that's, that's okay. But if the hits the fan and we're cutting off people, they're going to go just up, up and up. But then again, the other side of my mouth, you know, can they, can they hit that capacity with the manufacturing that they've stood up here? So. Hey, Daniel, one more to go. We're almost there, bestie. Lattice, you know, after hours for the company was pretty tough, but it's a tough market. And they're cranking where you would expect them to crank, which is in the data center and networking. I wish they didn't call it comms and compute. It's really data center networking and compute. And that is absolutely rocking. Their servers up 85% and networking is up 60, 63%. It's just that this industrial and automotive business is just in an absolute funk. And I mean, the entire industry, I'm waiting for somebody to say, well, a lot of people have said we've reached bottom, right? Like Texas Instruments on semiconductors. And then it's like they lay an egg in those, in those markets. So it's like, at what point do we actually believe that we've hit a bottom? I mean, there were quarterly growth numbers. Lattice, you know, moved up quarterly growth and that's a good sign. But still the discussion of bleeding off inventory. When does the bleeding off inventory stop? Lattice did say, hey, we are under supplying to demand. Okay. And that's code for, yeah, we're still, we're still burning off. They had some raises. Stifel, Raymond James, benchmark, Rosenblatt, Baird all moved. It moved up their price target. So, you know, here we go. Daniel. Great earnings, but not necessarily a, you know, a completely positive reception out there. In the marketplace. It's AI doom all over the place.

‍

Daniel Newman: Yeah. So I'll add just one thing on this one then. I've got one other name I just want to say a really quick remark about before we close out today. But the data center business is great. I mean this is another company that's just making that pivot. It had some legacy parts of its business, consumer and stuff that have been dragging it. It's been a longer term but pat 85 growth in server, you know, 60 up in their data center business. This is one a little bit like that process that Marvell had to go through it at one time to sort of pivot its business. I think you know, they're there and the demand for FPGA is very strong. And so this is a good one. But like I said, this is going to be data center revenue and server revenue that's going to offset consumer and some client type revenues that are going to continue to drag a bit. They're just not going to have the same kind of growth. And by the way, AI, you know, and some of the things that Surface compute can do inherently within, you know, within Copilot might take out some of the need for certain FPGAs. So they're going to get the growth but it's going to come on that in the industrial and server side. I did just want to give a shout out before we close out the show to our friend Nicolo Damasi. He had IonQ at earnings this week. You know, he's, you know, Quantum is controversial and a lot of people have a lot of questions about what it's going to be and what it's going to become. But you can't argue with revenue. I mean the company did, you know, hit 40 million this quarter against a 27 million target. So I had a big beat, you know, came in a narrower loss than expected on adjusted and also this has been a lot of the culmination of all these acquisitions. The buying spree raised multiple billions of dollars. They basically have enough cash on the balance sheet now to run this thing till the end of the, you know, the end of the decade without raising more money. So very smart, you know, admire, admire the work that's been done there. And by the way there's this gap loss that a lot of the sort of, you know, the, the bears, the, the Quantum never will be anything bears. And you know, I spent time with IBM last couple weeks at their Quantum event and others like I don't believe this is nothing I Believe this is an industry that's got some real legs in different areas. But the most interesting thing is all that loss that everyone's throwing around there, it all had to relate to warrants that were issued in relation to a run. It's a non cash item. It's usually like, oh, that's cash. But it's actually a non cash item. Actually did, did really, really well. But I just thought they had a huge raise on their annual revenue guide and they're going to be over 100 million in revenue next year. Now I know for a company that's, you know, worth 20 billion, people might throw that out and say that's too high. And you know, that's a debate for the, for the market. And the market is efficient and will find its own path. But of all these other quantum names, other than IBM, who said they have a billion dollars of revenue attached to it over its existence, 100 million in a year, they're, they're showing some, some great trajectory. So I just felt in the spirit that that was worth getting a call out. Pretty good performance there as well.

‍

Patrick Moorhead: Appreciate that, Daniel. So hey, we made it. Sorry I couldn't make it earlier in the week. I was literally on an airplane coming back. And I will be completely recovered when uh, this is airing. But I'm a little jet laggy right now. I gotta, I gotta tell you, my brain is, and my mouth is not working at, at full capacity here. But hey, appreciate all of you showing up. We do appreciate you hit that subscribe button and hope you have an amazing week. Hang out. Hang. Take care.

‍