On this episode of The Six Five Webcast, hosts Patrick Moorhead and Daniel Newman discuss the tech news stories that made headlines this week. The handpicked topics for this week are:
- Latest NVIDIA DOJ Antitrust Conversations
- Intel Lunar Lake Copilot+ Chip Announcement
- Qualcomm 8-Core Copilot+ Chip Announcement
- Latest Intel Conversations (20A, Arrow Lake, 18A, I-Bankers, IFS Broadcom, Lip-Bu Tan)
- Broadcom Q3FY24 Earnings
- HPE Q3FY24 Earnings
Be sure to subscribe to The Six Five Webcast so you never miss an episode.
Watch the episode here:
Listen to the episode on your favorite streaming platform:
Disclaimer: The Six Five Webcast 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.
Transcript:
Daniel Newman: Hey, everybody. Welcome back to another episode of The Six Five podcast. It’s Friday, it’s episode 231. We got a banger today, banger, Mr. Moorhead. First and foremost, I’ve been halfway across the world. I’ve been there, I’ve been back, and I’m energetic. I’m not jetlagged at all. Don’t ask me how I do it. Otherwise, you’re going to get me to share my supplements online along with my health journey. I would never do that. Just kidding. How’s the handsomest guy at Moor Insights and Strategy doing today?
Patrick Moorhead: Doing great. It’s so good to be here. It’s the calm before the storm, Dan, and I really like my home time. Before that, I cracked open my calendar and I think I’m on the road seven out of eight weeks and somebody’s trying to get me for that seventh week, but it’s not going to be cheap. That’s all. That’s all I have to say.
Daniel Newman: Manager.
Patrick Moorhead: But no, listen. Yeah, people who share their health journey online, what a bunch of exhibitionists.
Daniel Newman: Man, I don’t get it. No, I’m just kidding, dude. I’m so proud of you. I know we talk about this a lot, but there’s different things and different moments in our lives that come up and you’re a changed man. I know we got some really sad news. I won’t share the details of it about someone that passed a little bit too young. And it’s those moments that kind of help you reflect and say, “Hey, am I doing everything I can?” You’re working hard. You’re driving revenue. You’re on the TVs. You’re advising some of the most important companies on the planet, but you want to come home and see your kids grow up, maybe become a grandfather at some point in time. And it’s those moments that kind of bring us back to earth and realize that, heck, yeah, helping that next node get out or driving that next new launch of an instance, it’s all fun stuff, but it’s a lot more fun when you’re moving and healthy. And by the way, I hiked a mountain with you and I know you’re like 67 or whatever you are, but man, you don’t act a year older than 60. I’m telling you. You are humming buddy.
Patrick Moorhead: No, I appreciate that. And I’ll add to your reflection statement. I mean, sometimes you just can’t do anything about it. And I’m a man of faith and I don’t care what religion it is. I think faith is important. Because you know what, here’s what I can tell you with 100% degree of certainty, you will die and you will pay taxes. So, anyways.
Daniel Newman: Genetics are, they are a problem. I mean for sure, you can do a lot of things, but I always say it’s like a 10 percenter. It’s like you can make up about maybe five or 10% by doing everything, but there are certain genetic markers that it doesn’t matter what you do, you’re up against it. But you know what, Pat, you’re doing the right things. Proud of you, buddy. Call that out. But speaking of proud of you buddy, I mean some bangers this week out there, you were real busy talking about the DOJ or gosh, did they file a probe, did they not? Seems like they didn’t. But man, the source of that story, geez, I mean it was a good source. It was not the kind of source you would expect to be wrong. So, we’ll have to talk about that.
But man, we got just a great show this week to cover. You got DOJ stuff. We’ve got massive EFA launches, EFA, the Computex of Europe. You’ve got some pretty big announcements that came out of there. We’ve got some, just a vast number of rumors around Intel. I think you and I are going to try to dispel. I tried to do it yesterday and I got like 85% right. I joke. I get confused sometimes where the lakes meet the rivers, meet the falls, you just start to conflate things in your brain. So, listen, all you out there correcting me, I really appreciate that. I look forward to seeing your research when you figure out how to put this all together.
And then of course we got some earnings. Pat. You and I, we’re both covering, we talked to some of these CEOs this week, last week, and so lots to cover this week. For everyone out there, just know this show is for information entertainment purposes only. So, while Patrick and I will be sharing some great insights, we are not providing financial advice. Pat always like to say do the opposite. I just say don’t do anything of what we say, just do nothing. And that will be the best bet you can. Pat, because we have so much content, I’m not going to BS anymore. We’re going to get right into it here. I don’t know about you, but starting at about 2:00 AM the other night, my phone, I was in Europe, started blowing up about this DOJ filing or not filing. Give us the rundown.
Patrick Moorhead: Yeah, so first off, the Wall Street Journal probably a month ago picked this story up, but Bloomberg came in with some new information. It was super, super popular. And I think the reason is they used the word “subpoena” in it and typically the word “subpoena” is more correlated to crime than something different. But what happened is the stock tanked. And what a lot of people said was, “Oh, my gosh, there’s this correlation.” Reality was the stock didn’t tank because of this article. It tanked because of worries about the economy. So, I just had to put that there out on the table. So, it was funny, I got a call, “Hey, can you be on CNBC in five minutes?” And I did a phone in on John Ford’s show After the Bell. And then the morning after I did a Yahoo Finance talk about it. So, let me air this out. So, first of all, a little background. I’ve been involved in tech antitrust for a long time. I was a key witness in FTC versus Intel, AMD versus Intel, and very active in the stuff that happened with World versus Qualcomm, Apple versus Qualcomm. And not that that makes me a lawyer, I’m not, but gives me at least enough-
Daniel Newman: Glad you clarify that, Pat, not a lawyer.
Patrick Moorhead: … enough experience to talk about this. I was interviewed or grilled by Intel’s lawyers for the Texas state maximum of 24 hours. But anyways, first of all, it’s not illegal to hold a monopoly. It’s legal to use monopolistic powers to harm consumers. And typically, the bar for market share market definition is important is 50% and NVIDIA has 90 plus percent of the data center GPU market and its AI, which is huge for society. So, yeah, it just sends off red flares. No surprise here at all that the DOJ is investigating NVIDIA. Now, what are some of the illegal things that I’m speaking hypothetically that could be time, which could be a, “Hey, if you buy GPU plus my networking, you can get better allocation or better price.” Retaliation, again, hypothetically, if you buy AMD, I won’t get you the samples early or ship the volumes that you want. Exclusionary rebates are also illegal. Again, hypothetically, if you go with AMD, I’m not going to give you the same pricing. Pricing has to be volume-based. That is the legal way to do it or the way you segment that.
And why don’t investors care? Because these things take 10 to 15 years. I mean, check out Microsoft, AT&T, IBM, Google, heck, Intel. Intel, literally, this thing kicked off in 2005, FTC in ’09, EU in ’09 is still fighting the penalty payment, 15 years later. So, these are infectious from a global standpoint, meaning I fully expect the EU, Japan, Taiwan, probably Australia, they get emboldened by the EU. Oh, yeah. And the CMA in the UK.
Daniel Newman: There’s a filing this week, Pat, in Texas too.
Patrick Moorhead: Oh, yeah, yeah.
Daniel Newman: Not the same but kind of related.
Patrick Moorhead: Yeah, so final points here, these do slow companies down, right? You have to have a lawyer in important meetings about pricing, about strategy, about allocation. Correspondences need to be reviewed or lawyers are copied on it. What happens is you copy a lawyer, have a lawyer in a room, it goes under attorney-client privilege, and it can’t be shared and it slows stuff down. Now, it slowed Microsoft down big time. They missed mobile. They missed social. They missed a lot of inflection points. I just don’t see that happening with NVIDIA. Jensen has command control versus a distributed management style. It’s going to slow the company down, but I don’t know, we want to argue, I did a fortune year view. It’s going to slow it down 5%, 1%. When you’re going twice the speed of the rest of the industry, I don’t know if it matters. Final comment, I was asked about the competitors AMD and its Broadcom and Intel in 25, 26, but I literally, I don’t think this has an impact on short-term particularly.
Daniel Newman: Yeah, I think the first clarification point you made was really important. You have a day when the market’s selling hard. You have a company that’s maybe seen as running too far too fast on its price. Some catch-up in terms of the overall digestion period of AI and when that’s going to really hit. And then you had it sell. It did sell faster than the Mag-Seven names. So, there was an opportunity when they were selling 2%, 3%, one and a half percent off. NVIDIA was off nine, that maybe this DOJ probe had created a bit of a selling pressure. I spent more time on that than I spent at EFA because I had so many calls.
Patrick Moorhead: Crazy, man.
Daniel Newman: Well, listen, I think what happened was back in April, I wrote a market watch op ed. I started this process. You had to see, Pat, you and I have been involved in so many of these things even just the last few years, whether it’s been Apple, Qualcomm, Samsung, Huawei, of course, the various DOJ and FTC probes across different M&A deals. And when a company’s doing this well, you just have to expect it to be scrutinized. You have a policy position right now in the government that big tech is really bad. You’re seeing they’re getting emboldened by the wins they got with Google that’s at least pushing that forward. Apple’s App Store has been consistently under pressure, probably the most realistic antitrust case that exists, although they’ve made much less progress than I think they should. One of the great quotes I heard on the All in Pod was that’s because Apple’s popular. People like Apple, so they don’t want to push as hard as say a Google where they’re pushing harder.
But here’s the thing about NVIDIA, Pat is, and I’m being candid, you and I, we like to say the word “channel checks.” We have real channel checks. We have real conversations through the channel all the way up and down from the OEMs to the distributors, service providers, and even to the enterprises that are consuming this stuff. And listen, it’s a lot of hot soup is what I’ll call it. Kind of just comments of things that you’re like, well, that could be. We don’t put a lot of things in writing or deals. You hear things about deals not being described in writing. You hear things about, “Oh, company A is going to potentially work with company B, and company A was mad about that, so they gave more product to company C.” Yeah. Nothing’s proven. I mean, this is all just wild speculation. Bottom line is, is it happening? And this was my comment in my tweet, my extended tweet was like, it could be. It absolutely could be. And now it’s on the FTC, the DOJ to actually, well first of all, Pat, do they even have a probe? Is this even going on? Because we’ve heard.
Patrick Moorhead: Well, Dan, there is a prove It always starts with investigations, whether we will call it a subpoena or a forced and everybody’s on document retention.
Daniel Newman: Yeah. You’re right. I agree with you how far this is and where it’s at, how official this has become versus sort of semi-official versus like you said, document retention versus them coming into offices and having sit down meetings, I mean, I think they’re talking to the channel. That’s my guess. I think where a lot of this is coming from is they’re talking to people who are buying it, consuming it, implementing it, and they’re trying to figure out what’s going on. And you talked about consumer hard, Pat, but the other part of antitrust that’s so important is competition. And so, you obviously, because I’ve always said along the Apple Store for instance, is the consumers aren’t harmed. They feel it’s safer, it’s more secure and it’s easy experience, but the competition has gotten smashed by it. And that’s why I’ve always said that that’s such a problem. In this particular case, it’s like, look, people that are implementing and utilizing NVIDIA, they built on it. They’ve developed on it. They like it. It’s integrated. It’s a very closed sort of architecture where everything works together.
But when you have an AMD Broadcom or an Intel or you have these homegrown options, they’re making entry really hard. And you mentioned the Intel thing, I see a lot of parallels to the Intel thing. They’re going to have 90 plus percent of the market. They’re going to lock this up in the very earliest stages of AI. And then when people say, “Hey, I want to write programming to a different piece of hardware,” they’re going to be like, “Well, it’s unfortunately it’s in a container that’s running on a piece of hardware that’s running on a thing of software that’s using a proprietary library and you can’t move this.” Or it’s going to be really expensive to move this. But Pat, I do see this continuing. It’s going to be an ongoing conversation, something that will entertain us for many, many months to go and I like this topic. It’s fun. So, we’ll keep talking about it. We’ll see where it ends up, but we got to keep moving here, Pat. So, let’s talk about Intel. I’ll kick this off because I was in the room. Intel had a big moment. This next core ultra launch, Lunar Lake. This is the newest node. This is their breakthrough product that really made that competitive in AIPC. They’ll talk about the 20 million Meteor Lake units shift.
But that part in terms of the NPU and the competitiveness of the part was somewhat subdued, but it was the first in the market. So, it’s one of those things where I always say there’s a left and a right to the story. The left is they did shift 20 million. It was the highest volume AIPC in the market. But from a competitiveness standpoint, it really didn’t meet, it certainly didn’t meet the Microsoft Copilot+ PC metrics that were required. But that’s all been done now. Those sins are in the past. And now officially, Intel was able to parade the executives from every major OEM onto the stage to reaffirm their commitment to a design-heavy, volume-heavy cycle with this new Lunar Lake chip. And then of course, Pavan Davuluri from Microsoft came on stage to confirm that this will be an official Copilot+ PC announced at EFA.
This was a good moment for Intel during a really bad time. And I won’t dig into the rest of the bad time here because that’s a whole nother topic that we’re going to talk about here in a few minutes. But they talked about a 20-hour battery life. Listen, I’m going to say this. Signal65 has done some work, our firm, yours and mine’s, beautiful baby. Signal65 that tests the performance on this. We did some work on the matrix of compatibility. You actually saw that flashed up on the stage in the presentation that their compatibility across AI workloads and software was pretty impressive. The 20-hour life, I want to see it. I would hope Intel in this particular moment in time would be really highly scrutinizing of any claim like this they make, because if they’re wrong, they’re going to get absolutely crushed. I’m hoping it’s right because from a competitive standpoint, that’s great. Making big progress, X86 becomes more viable to be able to deliver that kind of performance, low power.
That was the biggest thing that I got out of was that number, that metric. Of course, they have a really strong gaming story right now. They’ve been able to really lean into the gaming story, the GPU story, which again is part of it. I know we love talking NPU, but GPUs really matter with AI, especially if you can have that with the low power. And this AI hub, this performance thing that they’re doing for being able to do LLMs, do graphics and do code, I thought that was pretty cool too. That was kind of my first seeing it in real time, seeing it perform. And so, it was a really positive announcement, Pat. And by the way, I talked about the parade, but I mean you had Dell, you had Lenovo, you had Microsoft, you had, I think it was Microsoft HP, Alex Cho who we ran into, and several others, also Google, which by the way, Google came on and for Intel, don’t think Google showed up this week with Qualcomm, which was interesting as well. So, among all the chaos at Intel, there’s still a lot of proof points to be made. There’s still volume proof to be made. There’s still execution that has to be done. Intel is under much greater scrutiny than anyone else, but it was a positive moment.
Patrick Moorhead: Dan, that was a great breakdown. I did get a little bit of FOMO watching you.
Daniel Newman: It was these, right? It wasn’t the actual announcements because you were watching. It was just all the selfies you couldn’t be in with me. I missed.
Patrick Moorhead: Yeah, not being in those selfies is a big hit. So, listen, you did good macro. I’m going to do a little bit more micro here. Lunar Lake has exceptional single-threaded CPU performance. Part of that is the newly architected cores that they have. These are single-threaded and not multi-threaded. They’re very, very efficient. So, kudos to the team for that. There’s embedded memory, which is going to give you a pop as well because you can more tightly couple the system to the memory. Apple did it and Intel’s doing it here. I’m expecting the best graphics performance in this class of processors. And you can either take that to gaming or you can take that to AI. Less efficient AI than the NPU, but much more efficient AI than the CPU. You get 100% app compatibility as well. Microsoft Qualcomm and ARM have done a stellar job getting ARM native compatibility out there, and there are roadmaps for the more difficult ones that hit like the .sys that aren’t just a loader issue but require native and you can’t do them emulated like VPNs and basically malware detection, stuff like that.
If, and a big if, this is the key part here, Danny, you hit this. If the battery life claims or bore out across different use cases, this is a very good thing for Intel. And I will be impressed and I will be surprised that once we get these into our Signal65 labs, that this bears it out. And here was the claim, 20 hours on Intel versus 18 on Qualcomm and 10 on AMD using Procyon productivity, which is independent third party. I just want to see it myself. I want to see exactly the conditions that AMD and Qualcomm were tested. Now, we’ll see. Here is where Lunar Lake will get beaten. It will get beaten on most multithreaded applications because of core count. It’s very simple. Even if Intel has more efficient cores on ST, you throw more cores at it. Eight cores on Lunar Lake versus 12, 12 and 10 cores. I, with 100% degree of certainty, will tell you that on MT CPU scores, Qualcomm with 12 cores will beat Intel with eight. Windows AI in November, Copilot+ ready. There’s a lot of historical examples of being ready. There was Windows 8 ready. There was Windows 7 ready and now there’s Copilot+ ready. I don’t know if that’s exactly what we’re going to see on the sticker, but it’s the way it’s going to be positioned. And having Pavin up on stage from Microsoft talking through this almost just makes that go away.
Daniel Newman: Hey, that’s a great dive. That’s why Perplexity tells the world that you’re smarter at the deep dives and I’m the high-level macro guy. That’s why we’re so good together. We should get married.
Patrick Moorhead: We really should, Dan. We really should. And I think-
Daniel Newman: I mean we have two beautiful children, The Six Five-
Patrick Moorhead: Well, that’s so good. I was not expecting that. That’s really good.
Daniel Newman: But I don’t know. We also brought, we have our children from our previous marriages, a Futurum Group and Moor Insights. But anyways, that was a good dive, man. Listen, one quick question for you while I change the topic. What’s at stake in your opinion, if that battery life thing backfires? Because I mean, again, you mentioned third party. That’s the thing I keep raising my hand is what’s at stake here because it seems like it was so bold.
Patrick Moorhead: I mean, they lose credibility in a time that they’ve lost a lot of credibility. I am comforted by the fact that they use the Procyon benchmark, which is hard to game, but until I see it, it’s going to be hard. And then there’s the, with the Qualcomm laptops, I can close that thing, stick it in a drawer for two weeks and bring it out. And maybe the leakage was like 5%.
Daniel Newman: It’s amazing. It’s amazing.
Patrick Moorhead: I had a Mac ARM-based Mac sitting in a drawer for a year. I opened that thing up, it had 30% battery life. I have never ever had an X86 laptop that did that.
Daniel Newman: It was so appropriate that you did that, it’s almost like you’re the host. It’s almost like you’ve been doing this for a minute that you know how to change the topics. Let’s talk about Qualcomm. They have a eight core part. They’re coming down market while we’re talking ultra premium and top of the market. Qualcomm’s like, “Look, we’re going to attack the whole market. We’re going to come out with an eight core part. We’re going to put in the $700.” What do you think, Pat? What’s going on there?
Patrick Moorhead: Yeah, so giving Qualcomm credit here, Qualcomm completely disrupted the Windows Notebook market. Some people can say that, “Oh, it was Apple.” I mean, Apple really initiated this, by dropping the M-bomb on the entire industry. Hats off to them in doing that because then that inspired gave an opportunity to Qualcomm and AMD and Intel had to react. So, hats off to them in not only bringing this amazing NPU to the table and enabling Copilot+ experiences, but even as impressive, bringing the performance and battery life CPU and GPU to the table. It’s pretty amazing. So, the first announcement was that they did a while back and announced to Computex was a 12 and a 10 core variant. And this latest out is an eight core Copilot+ plus PC processor. And you might be like, “Well, so what? It’s down core.” It’s very important. First and foremost, 799 price point. There are Copilot+ Qualcomm. PCs as expensive as 1699 out there. I would say most of them range at about the 1199 price point. And from a unit volume standpoint, there’s two, 3x the volume at that 799 price point and volume is very important. It’s important for scale to Qualcomm. More importantly, it gives scale to the partners who need to be able to write off below the line expenses and amortize those over a larger volume base. Qualcomm put it in the same socket, which means all the platforms that are out there that were already out there to drop in a piece.
This puts Intel in a very difficult position. And gosh, it reminds me of the good old days of AMD versus Intel that I was knee-deep in. Literally, Qualcomm drops an eight-core part that has as many cores as the highest end Intel part. And yeah, I get it. Intel cores are more efficient. Check, I get it. Higher performance, we’ll see about the battery life. But it puts it in a very difficult position in the consumer space. And here we go. It’s going to make benchmarks. The benchmarks wars have begun between Intel, Qualcomm and AMD. It puts Signal65 at the forefront of this. I’m recommending, I think Intel should drop some more detailed benchmarks within two weeks until the systems come out on September 24th. It’s sad that a non-tech pulled the plug, but that just shows the state of affairs today. And quite frankly, it doesn’t limit the need for third-party independent benchmarks. I think it actually increases the need for that. I don’t think the YouTubers are going to fill that up. Super exciting, man. Stepping back, more competition means more innovation and lower price points for everybody. And you need three healthy markets have three somewhat equally sized competitors.
Daniel Newman: Yeah, absolutely. And I mean while the size is kind of somewhat still asymmetrical in Intel, it might feel like they’re the smallest right now because of market cap. They’re actually by quite a large amount revenue the biggest. Having said that, Pat, I agree with you. First of all, great takedown on the overall. And by the way, we need to get to AMD. We’ll probably have to come back maybe next week and talk about that, because they had a bunch of stuff. Their stuff started popping yesterday, so we didn’t have as much time to assess. But they’ve got a Copilot plus thing going on too. Pat, just because you hit a lot of Qualcomm. I was in the room. Cristiano Amon basically said, “Let’s have a benchmark off.” He’s ready. He’s basically saying, “I’ll take your numbers.” He’s kind of looking at it like the comparisons were across different parts. It was inconsistent. You put your best part against our mid part. That’s kind of the tenor and tone I got from the stage. And then the questions Ryan Shrout, our president of Signal, asked the first question, I asked the last question last. We open and closed that baby down.
And I got to spend a bunch more time with the executive team while I was there. There’s a lot of confidence at Qualcomm that they’re getting it right and the numbers won’t lie. Of course, there’s two sets of numbers though, Pat. There’s the numbers of what the benchmarks say, which we’re here to work on, and then there’s going to be the numbers of what the market says, which is what sells out and what people buy and what people use and what becomes more popular. And does Qualcomm really begin to become a household name in the laptop business? Pat, I mean, you and I have been through iterations and iterations of some of these first-generation. This Qualcomm part for the first time I’m using this device regularly and I’m saying, I could use this device regularly. This is good stuff. So, addressing sort of mid part of the market, Pat, is also super important. That’s where companies exist. They’re going to make this big shift in this big wave and this AIPC thing is just getting started. We knew it would only start this year. We were trying to temper the expectations. We’re starting to see this come to life. We’re seeing volume starting to ramp. I think it’s great that Intel is healthier and I think it’s great that Qualcomm’s going all in on making more and more competitive products. This is good stuff and like I said, we’ll come back to AMD later.
Patrick Moorhead: Yeah. Hey, just one quick comment. This bodes well for the AIPC market.
Daniel Newman: It does.
Patrick Moorhead: Having Intel come here, guns of blazing makes an absolute huge difference. And yes. “Hey, Eric, thank you so much. I appreciate it. Thank you for dialing in.”
Daniel Newman: That’s always nice to have someone tell you you’re pretty. I think he was talking about your tweet about your VO2 max. That was great stuff.
Patrick Moorhead: What a surprise. How did that happen?
Daniel Newman: Yeah.
Patrick Moorhead: Anyway.
Daniel Newman: You’re looking and feeling great. I think you deal like a fight. What do you do? Fighter have a boxing match. Now I’m starting to wonder if you’re actually metabolically and biologically younger than me.
Patrick Moorhead: Dude, if I got hit with one of your punches, it would probably just go through my body. Not interested in any of that.
Daniel Newman: Yeah. But it would be funny, like a fight club moment. The world would enjoy you and I just, there’s so many other people I think we would enjoy anyway. All right, so let’s talk a little more about Intel because I don’t think we’ve talked enough. Interesting. 36 hours actually. Interesting. 10 years we could say. But a really interesting week and you kind of come across the top here. I’m going to talk about pieces of this. You can talk about some other pieces of this. There’s a whole bunch of 18A stuff. There was a deprioritization of 20A, a prioritization of 18A, a TSMC node. I knew that. I tweeted something a little bit inaccurately. I apologize for it. Thank you. By the way, all of you that corrected me. It was really important. That was the takeaway of my 2000 word dissertation. What do you say about credibility, Pat? I guess I blew it early.
Patrick Moorhead: Dude, dude, anybody with a brain know … Yeah.
Daniel Newman: And it’s fine. I just laughed because like I said, did you read any of the rest of it? Is that all we got out of it? It’s like being at home, dude. It’s like I give the whole story and then someone tells me your shoelaces untied. Thank you. All right, I digress. So, I’ll hit a couple of these things. Maybe you want to hit a couple and then we can go back and forth on what you think is worthwhile of talking more about. So, just really quickly, Lip-Bu Tan, board member at Intel, left the board, did it amicably in terms of his public note. But this guy, a legend in semiconductors, a cadence CEO clearly disenfranchised with the direction of the company. Not a good look. And of course we heard after that, risks about activists, risks about starting to do further breakups of the company. We saw that there was basically a board meeting coming up. Everything is on the table. We’re now hearing, by the way, I read a new note in Reuters today about Qualcomm circling Intel to potentially acquire parts of its business. If they end up having to spin them off, check the Reuters report, Pat, if you didn’t see that one.
I could see Mobileye, for instance, being a really interesting play for Qualcomm at this moment. Other things too, potentially. But that caught my attention. Not a good look right now to have really, really credible board members leaving in this particular moment in time. Going to be interesting how they respond to that. I’ll kind of tie a couple of 18A threads, because you mentioned it says Broadcom 18A, 20A. First of all, there was of this narrative about Broadcom in a sample, a pre 1 dot OPDK not getting the performance out of Intel. Interesting that these rumors have dropped. There was a SoftBank rumor. There was a confirmed, or confirmed not rumor, story related to SoftBanks and others that have been sort of testing on Intel. I think there’s an enthusiasm and excitement from Intel to get things out, but clearly there’s this weird leaking that’s going on, Pat. By the way, you’ve been around this industry even longer than I have. So, I’m doing this more from hearsay of my experience in learning over the last decade. But this stuff doesn’t typically leak. Things like this. When you sample and you’re letting customers touch this stuff, the fact that this is leaking is very odd.
But at the same time in my long tweet, I’ll put it in the show notes, I kind of talk about how there’s a bit of this feeling. And I don’t want to use the word “conspiracy,” but I’m going to use it because I don’t know what else to use right now that Intel’s like the prize fighter that’s on the ground and it’s on the four count and it’s like you have the chance to punch it while it’s on the ground and keep it from getting back up there. Feels like there’s a little bit of this circling around Intel right now that maybe it’s because they were determined that the government-backed national champion of Foundry. Maybe it’s because of the ongoing channel and feedback we’ve gotten from the market that says, “We love TSMC. We don’t care. We’ll just work with them forever.” Maybe it’s all the risks associated with that. But having all of its woes, all of its woes leaked out these days. But then again, we had Blackwell leaks too, so who knows. And then of course there was the deprioritization right now, where they’re basically going to double down on 18A. So, right in the heels of Broadcom supposedly getting low performance on a sample pre 1.0, they also came out and said, “Look, we’re going to skip to 20, we’re going to go straight to 18. We’re going to compromise our five and four.” But we all sort of knew, Pat, that not all five were going to be serious nodes, not serious. That’s not the right word. Like full nodes maybe was the right word I’d use. Serious.
Patrick Moorhead: Yeah.
Daniel Newman: You’re not serious people, Moorhead. But in the process it kind of was like a flip of indication to me though that they actually feel really confident in 18A and they feel they can get capacity and they feel they can deliver a really strong part. So, that was a really interesting conflict or maybe rebuttal in that moment after the Broadcom story leaked. And then what it also brings some question to its own foundry and what it’s doing with 20 and what it’s going to be able to do going forward. So, there’s just all this kind of coming together at one time, Pat, and I know this is where I start to try to take four or five threads and not conflate them. But man, are they a new cycle right now. And it just feels like there’s so much conviction that nobody quite wants to see it go well. And I think that’s why I think you and I are trying to just be balanced about what this is. There’s some good things happening, some good indicators in this and some really concerning ones.
Patrick Moorhead: Yeah, I’ve never babe-ruthed Intel. I always said, “Here are the conditions for success that Intel would have to do to get back on top again.” But Daniel, it is knives out on Intel. And those who don’t have history don’t understand why knives might be out. I mean, Intel was the NVIDIA of today. Okay. Let’s just be very clear.
Daniel Newman: Memories.
Patrick Moorhead: They are a convicted monopolist in the EU. That’s just a fact. I’m not making stuff up here. And they’re still debating the fine 15 years later. But if you go and read the complaints from AMD and even when the FTC took a swing at it, I mean it was ugly, exclusionary pricing, tying, even selling below cost at certain, it was crazy. So, yeah, there are people who want Intel dead. I’m not surprised at all. Just look at the people who would benefit if Intel were dead and they want Intel dead. Now that doesn’t separate-
Daniel Newman: I said that in my tweet.
Patrick Moorhead: Oh, you did? Okay. Intel has shot themselves in the foot many, many years ago and it’s hard to get back. I mean, a semi-turnaround takes five years. But let me try to fill in not blanks or some oxygen you left for me. I was wondering about Lip-Bu Tan and by the way, you’re right, absolute icon on the board of HPE, Schneider, Softbank, the guy won the freaking Robert Noyce Award in 2022 from the SIA. Robert Noyce, an ex-Intel guy. And the rumor said he was frustrated with the large workforce, risk-averse culture, lagging strategy in AI. Now the interesting thing is the thing that Pat rolled out last earnings seemed to address the large workforce course. We know what their strategy is in AI. It looks like that’ll probably be a GPU that will make an impact as early as 26 and risk-averse culture, I don’t have the spreadsheets or how they make decisions in there.
So, yeah, I first thought that maybe Lip-Bu was putting together a group of people to maybe go and put pressure on Intel, maybe get Pat fired or something like that. I’m not thinking that anymore. There was a lot of conversation. It’s so funny. Talk about knives coming out. I think I had 60,000 people look at my post on LinkedIn about spinning out IFS. Let me be crystal freaking clear. You spin IFS out before the design company is healthy and IFS is healthy, it will fail. Now, my only caveat on there would be to take it private and don’t put this thing out on the market and then you have to find private investors and it’s negotiation of how you fund that, cap equity, debt, things like that. What could make sense is you take it private and its biggest investors, let’s say you do a third debt and you do a third capital allocation and then a third of it are investments by its potential customers. You get Apple to pony up money. You get NVIDIA to pony up money. You get Broadcom to pony up money and then you wait until 2027 to where Intel Foundry will be driving significant profits and then I think 2028 before you start seeing profits. Okay? So, that’s where it is. Well, why do you say that? Why wouldn’t that work? Listen, I ran corporate marketing at AMD when we spun off GlobalFoundries. I know how this works, man. Talk about two at the time, two sick puppies. You had AMD that was locked into long-term wafer agreements that they never kept. But they had to do that to keep GlobalFoundries going. And the only reason that didn’t fail is you had Mubadala who is bankrolling GlobalFoundry and internalized all the losses. Final comment on here. God, I love this could be a show itself. We just have the Intel show.
Daniel Newman: Actually, we probably should just do the Intel show.
Patrick Moorhead: Yeah, final comment, had a little bit of a side eye on, okay, 18A is going so well. And by the way, their defect density does look really good before going into mass production. That’s super, super helpful. And by the way, that’s true. I have no reason to believe that’s not true, is a very positive sign for 18A. If 18A is going so well, then why are you getting rid of 20A? It seems like you could do both. Well, first-
Daniel Newman: Okay, go. I had an answer. You have an answer.
Patrick Moorhead: Yeah. First off, 20A. 20A was designed to be a pipe cleaner for 18A, but moving resources from 20A to 18A, it just gets better. And you reduce risk by doing it. You reduce likely design risk on Panther and I know the resources are different, but you also, you can apply to make those transistors, let’s say, even more power efficient. You have a thousand engineers, where are you going to put them? Going to keep them on 20A, you’re going to move into 18A. So, that foots for me. I had a little side eye in the beginning, but I totally get it and it foots with me. What were we going to say, Dan?
Daniel Newman: Yeah, listen, great take. This was a show. We need to do the Intel run-up. Pat, they have to focus. I mean that was my biggest take was in this moment in time, they have to focus resources. They’re promising the world that they’re going to be intelligent with their spend. They’re going to lean out and get out of businesses that aren’t driving profit for the company. More nodes mean more cost. It’s more setup. It’s more planning. There’s more engineers. All those things, Pat. If they think they can get on 18A more quickly and ramp it get to, and by the way, focus can mean more profitability, on a per wafer basis.
Patrick Moorhead: Yeah, man, that was a great comment. We didn’t even talk about potentially Brandenburg not happening. Not to be confused with Hindenburg.
Daniel Newman: That’s coming next. I bet.
Patrick Moorhead: Listen, the Germans didn’t come up with as much money. I could see them getting rid of that. I don’t understand dishing off Altera completely unless they have a buyer in mind. I think what that could do is that brings them cash to be able to do something else. Pat did talk about getting out of businesses that were upside down or zero revenue. I think Quantum is on the table. I think neuromorphic is on the table. And one thing that never gets talked about, but I know is a reality is Intel has backup teams. They used to have three teams that they would put to attack a problem. But you look at the AMB numbers, number of people they have and it’s like, “Wait a second, how does that work?” But I’m sure there are a lot of multi-hundred million dollar projects that add up to billions that are going to get whacked.
Daniel Newman: Yeah, I look at three things, Pat. This is PC Copilot AI thing, nail the Gaudi. Then of course to the GPU transition and become an actual player in AI. The market absolutely hates that they messed this up and it’s unforgivable in many ways. It’s not a Pat thing by the way. This happened well before Pat. He just inherited.
Patrick Moorhead: Picked the wrong architecture. That’s what they did. They picked an HPC architecture that was really good at 64- and 32-bit flops and they did really well out there and they had an ASIC. But the ASIC didn’t save the hyperscalers enough money and they just did their own as Broadcom.
Daniel Newman: And so, we’ll be doing some testing on Gaudi. We’ll get more out there in the future because there is some level of compellingness there.
Patrick Moorhead: Gaudi is compelling. And as we’re going to talk about in Broadcom, Broadcom’s not even focused on the enterprise.
Daniel Newman: Nope, nope, nope. And then of course, the third is Foundry. And I think it’s part of the differentiation is that it’s under the umbrella. Spin-off would be pretty devastating in my opinion. I just don’t know that we need another fabless and I just don’t know how successful Intel will be. Being the IDM was always sort of what made them special. So, spinning off another fabless in this era of ARM and risk and everything else that’s going on, I don’t know. I don’t love it. I don’t love it. I understand why they might have to do it. I just don’t love it. All right, listen, let’s talk some earnings, Pat. This one, you and I both have had a few opinions about this Broadcom. So, good result market hated it. What’s going on?
Patrick Moorhead: Yeah, so Broadcom beat on nearly everything, except the revenue guy that was off expectations by 1%. That’s $135 million on $14 billion. I digress. I just want to say CEO and CFOs, take note. This is the way to do the call. Hock Tan, it was a freaking masterpiece. Insights into the overall market, how Broadcom was playing, detail after detail, splitting the market between very crisply between AI and non-AI, Dan, the Q&A was 39 minutes. Okay, I think everybody got a whack at at least one or two questions. And Hock very patiently, sometimes they asked the same question. He didn’t get necessarily irritated. But that’s the way to do that. Okay, strong results, three factors. AI revenues continue to grow and they go strongly. Second thing, VMware bookings are accelerating. And three, non-AI semis have stabilized, which is crazy because the broadband market is still in the toilet based on the telcos and the SP. So, let’s start off with VMware.
What I loved is Hock just came right out and said, “We said pre-acquisition, we’re going to deliver 8.5 billion in EBITDA within three years of the acquisition. Where are they? Well on the path to achieving or exceeding this goal in the next fiscal ’25.” Okay, so I don’t know what is that, two years early, one year early. It’s early. Knock it out of the park. ABV analyze booking values up 32%. And yeah, I mean VMware is going well, but Dan, I think the meme was VMware is just like, “Ah, crazy. Customers are canceling their contracts and rolling their own KVM or going to Nutanix or Red Hat.”
Daniel Newman: Hypervisor, baby.
Patrick Moorhead: We want to do that. Well, it’s not in the cards right now. Listen, am I babe-ruthing this one? I mean there will be customers that will leave. I get that. Okay.
Daniel Newman: I think he’s actually ushering them out. He’s carrying their suitcases. Hock himself is like, “Let’s go.”
Patrick Moorhead: Yeah. This is something that companies like SAP should have done years ago. And by the way, VMware on its own would never have done this. Okay, let’s snap into semis. First of all, 11 plus billion forecast for AI going to 12. Two thirds of that is XPU or accelerator. And one third by the way, that’s $8 billion in XPUs. That’s almost twice as big as AMD’s data center business. So, factual it goes NVIDIA, Broadcom and AMD. And inside AI semis, custom AI accelerators grew 3.5x. Fabric ethernet switching, that’s Jericho, sorry, Tom Hock five and Jericho three AI, 4x growth. Lasers and pin diodes, 3x. PCI express switches, doubled. But hey, there’s non-AI two, which was just very eloquently up sequentially. Hock believes they hit the bottom in Q2 and Q3. Non-AI networking was up 17% sequentially even though it was down 41% year-on-year. And then bookings, demand is up 20% sequentially and all of this, even though broadband is still in the toilet. CEOs, CFOs, take note. Watch how Hock does it. Hock and Charlie show, just do what they do.
Daniel Newman: Hockonomics, baby. Hockonomics.
Patrick Moorhead: Totally.
Daniel Newman: Did I trademark that? I don’t know. Maybe someone else.
Patrick Moorhead: Are we going to debate on who? No, you said that first.
Daniel Newman: I said that first.
Patrick Moorhead: I know. I know.
Daniel Newman: I don’t get a lot of firsts, but I occasionally say so. Like I said, my historic rise to relevance isn’t all about one-liners. It’s never about death. That’s what you’re for, buddy. I’m kidding. I’m kidding. I love you all.
Patrick Moorhead: I’m so much older. I’m like 60 years old, Dan. You’re like 30.
Daniel Newman: We won’t hold that against you.
Patrick Moorhead: I’m twice your age. I never hold your inexperience against you.
Daniel Newman: Yeah, except when you do. Listen, I went on and I did an interview this morning on Yahoo Finance on this topic. I think you missed the email. I appreciate everyone for having me. Listen, first of all, Broadcom’s fine, 60x percent EBITDA top line. I mean look, only NVIDIA returns more EBITDA on revenue than Broadcom does. I mean, Hock runs an amazing business and the distribution and diversification. This isn’t some new cutting-edge future part. I mean of course those XPUs are, but he’s also selling wireless controllers, PCIe cards, and he’s still making this kind of money. That’s what Hock does. That’s why when people get all disenfranchised, I’m like, “Why? This is a great business. It just puts off tons and tons of cash.” The software, the move. I mean, look, CA performs well, the LACO-run mainframe business, the brocade side of the house, the semantic side of the house, all good. VMware’s going to be good. Pat, and I tell you, I’m being sincere when I say he’s carrying the bags of customers out the door. Any sort of deal, midterm, long-term agreements that he had that he was able to shed, he shed it. He’s okay with the 20-80 rule. He’ll take the 20, 30% biggest group of customers and then he’ll let the long tail do what the long tail does. He’ll make more money. He’s chopped up the heck out of the cost structure of that business and he will return capital to both the business and to shareholders.
That’s what he does well, Pat. The XPU part’s really interesting because sell in, sell out, the design part, how much is going in going out, it’s going out a lot faster. Meaning these companies are spending lot more on this stuff than maybe it’s being consumed. And that’s according to our data. But I think the consumption is going to start to ramp really fast as this killer inference workload starts to ramp. And XPUs are really, really in a good position, Pat, we’re showing the XPU accelerator to be like 20% faster growth than the GPU. Now having said that, I actually think it’s going to be faster than that. I think that’s a conservative outlook based upon sort of sentiment. And I don’t think the market fully appreciates the sentiment about what an ASIC, a well-designed ASIC for drug discovery recommendation engines, filtering can do in terms of lower power, performance and cost. And of course, vertical integration and control. You’ll never hear the cloud companies talk about vertical integration. They’re all going to go down the path of vertical integration.
And Broadcoms become really astute at building these XPUs. So, they know the plan, they know the cost, they know the structure. And Hock talked ad nauseum about how this kind of market pivot is going on. Pat, I actually think this is a killer business. And then by the way, all this, we’re barely even talking about the fact that networking. Look, all these companies that don’t want to be beholden, don’t want to be locked to the NVIDIA and Finnegan in rack, out the rack and be link. Who are they going to go to? Broadcom is super well positioned in this particular space, that 12 billion number … Here’s my take on why the market kind of crapped all over it is they wanted it to be like 15. It was the same as that whole kind of whisper number with NVIDIA like, “Oh, they only are going to be by two and a half billion next quarter.” It’s like, “That’s not good enough.” Look, they’re trading at something like 26 times forward. It’s not a super high multiple or premium for a company that’s delivering 60x percent EBITDA. They’ve diversified. They’ve got different businesses. Hock always makes them profitable.
And one other thing, Pat, the whole wireless and connectivity part of the PC, the stuff they do with Apple and smart devices, the ramp on that is just starting, meaning that old kind of part of their business, they’re not really on the AI side of Apple as much in the devices, but as this ramps, it becomes what I think will be one of the most popular parts. This iPhone 16, 17, Broadcom’s got the most content out of any silicon provider to Apple. That partnership is deep and we’re not even seeing that in the numbers. That’s the part that’s lagging right now. So, all of a sudden next year with these new PCs, the new Macs, the new iPads and the new phones, Broadcom will deliver. Pat, I’m super bullish on this one. I know we’re late. We said we knew we were going to be late. The show is too good. We still got another topic. I have to do this in five minutes or less. I’m going to move pretty quickly. You talked to Antonio, right?
Patrick Moorhead: I did. I did. And I’m glad you didn’t fall asleep for it.
Daniel Newman: I stayed awake this time, but this time I was only, it was like 10:00 at night. I was at 4:00 in the morning I was trying to do that call when I was in Taiwan. That was idiotic. Blame me. So, HPE, Pat, look, you and I will sometimes talk about kind of these OEM businesses being single digit, not double-digit growth this quarter. Very excited for HPE. They saw 10% growth. I made a joke. I said they grew faster than one of the largest CRM SaaS companies on the planet this last quarter. How fun is that. Margin integrity in place. Antonio was very optimistic about not just AI, but actually seeing growth in their traditional server business. Who’s done that? I mean, nobody’s had good numbers in servers.
And he really talked a lot about discipline and rigor in the way they’re running the business right now. We’re seeing sort of this chasing of these mega AI deals. Big margin erosion, high revenue growth rates, exciting AI. I think HPE is kind of taking the other, I was a little critical of the all-in on NVIDIA thing, not because NVIDIA is not great, but because I worry about the hedging that that’s going to create risk and maybe you could differentiate, but look, it seems to be working that 1.7 billion number, the margin looked good on it. He sort of said they’re not taking deals that they’re not doing well on. So, they’re being a bit discriminant in kind of how they’re doing that business. At the same time, reliance performing.
And Pat, the other number I really liked is GreenLake. 3000 new customers, and I want to do a little compounding for you here. They got into close to 40,000 customers at the growth rate. They’re doing 10% per quarter compounded by the end of the 2025, ’26, their run rates have 100,000 customers running on GreenLake. Now as a company that we kind of call a big iron old hardware company, that sure looks like a software model to me. A hundred thousand subscription customers driving billions of dollars in revenue. That’s really positive. Pat, look, I don’t always get excited about these businesses, but I came away from the conversation with Antonio for the first time I told them this, this must be probably the cleanest, one of the best looking quarters I’ve seen from HPE in a really long time. I’m optimistic about it. They seem to be playing their cards right. They’re not chasing the bigger, the companies that kind of can go that way, they’re playing their game, they’re running their race, they’re sticking with GreenLake. There’s probably some flaws in it, Pat, but I want to give credit where credit’s due. And it was a good quarter.
Patrick Moorhead: Yeah, and my tweet was like, “Double digit baby.” It just makes a difference.
Daniel Newman: Feels good.
Patrick Moorhead: At 10% plus you should be considered a growth company for sure. I was really enthused by the split between AI and non AI. Servers were 4.3 billion, up 35%, and operating margin increased from 10.1 to 10.8%. And that was AI and non-AI spooked about AI profitability because gross margin was down 420 bits. Edge and hybrid cloud revenue and operating margin was down year over year. And cash flow and FCF were down as well. But there were a lot of highlights that pushed out the low light there. And I was equally-
Daniel Newman: The margin erosion, Pat, was not nearly what others have faced in the same-
Patrick Moorhead: No, it wasn’t. And that was the Dell moment the previous quarter where people got spooked on margins. They looked at the flow through to profitability and I think they had a very similar reaction. 37,000 GreenLake customers. I mean it’s impressive. And yeah, I know there’s a lot in GreenLake, but that’s GreenLake. It’s subscription service. I love that. One comment that Antonio made with me, he said that HPE infrastructure was in hyperscalers. That was a new one for me. I followed up on who. “Hey, are we sure this isn’t tier two CSPs here, but hyperscalers?” They were like, “It’s hyperscalers.” We’ll see. I think that’s very provocative. We heard that from Pure Storage as well. And I’m guessing it’s meta there.
So, who knows? 35% improvement in ARR and that’s attributable to GreenLake. And I’ll close this out by talking about the future opportunities here. Private AI just went GA. I cannot wait to talk to customers. I’d love to do an economic study across them as well. Interview some of their customers and what are the benefits that they’re seeing. I got the demo of private AI and it is freaking impressive. It’s really good. I’m a hands-on guy. I’m not as hands-on as Keith or Matt, but I was super impressed by that. Morpheus data, build out that, how do they monetize that. And Juniper is on tap close end of ’24, start of ’25. As we’ve seen in the era of AI networking is as important as ever. There’s one 800-pound gorilla in the western world for core switches, and it’s called Cisco. Will Juniper help out in that? They’re heavy-dialed into the SP market, but wonder what they can do in the future.
Daniel Newman: So much there, Pat. What a show, man. I knew we knew this one was going to be a banger. Sometimes you just call out, they’re like, “This one’s going to be good.” And then some weeks we’re like, “Oh, let me talk about sanding there. We want to talk about sand again and laser beams,” that’s all fun. But I mean, this stuff is fun, man. The rumors, the speculation, and of course the hard data that would drop this week. And you see, Pat, the path that our newest child together, our Signal, the testing benchmarks, whether that’s data center servers and networking or that’s going to be the next generation of PCs and smartphones just exploding. And everyone out there hopefully follows us. We’re going to have the latest and greatest here.
Patrick Moorhead: Guys. Our backlog for Signal65 is gigantic.
Daniel Newman: Pat, our Signal business is bigger than Futurum was like two and a half years ago already.
Patrick Moorhead: Yeah. What’s crazy-
Daniel Newman: We make beautiful children, friend.
Patrick Moorhead: Listen, we can pat ourselves on the back.
Daniel Newman: Not all the time.
Patrick Moorhead: I mean, listen, we both saw a need in the market that was going away, and quite frankly, the independent benchmarks on the data center side that went away a long time ago. And then the huge demand with AIPCs, whether it’s processor companies, whether it’s OEMs, ODMs and stuff like that. People want to know what the numbers are.
Daniel Newman: And we’ll leave it at this, Pat, making sure that there’s a pivot from just testing and performance for being a geek and actually helping businesses and markets understand what this means and how it’s going to deliver value. I think that’s where we can really shine. That’s where we do really shine. And by the way, we just want to say thanks to everybody out there that tuned in for this show. Great comments today. Appreciate all the positive vibes, all the comments that came in. Hopefully you enjoy, subscribe, share with your friends, stay-
Patrick Moorhead: Ask questions, man. We want more questions on the socials. Hit us on LinkedIn. Hit us up on the X Twitter.
Daniel Newman: The inner community, look for long, long, long tweets from me and smart ones from Pat, and just stick with us here on the show. We promise we’re going to keep it interesting, keep it engaging. But for this show, for this episode, I got to say goodbye to everybody. Thanks for tuning in. See you next week. Adios.