The Six Five Pod | EP 275: AI's Ripple Effect: The White House, Job Market, Chip Wars, and Corporate Strategy

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. AI and Antitrust Developments: Analysis of the recent antitrust ruling regarding Google's search monopoly. Discussion on the impact of Generative AI on the competitive landscape. Examination of data sharing requirements and their implications.

‍

2. Economic Data and Market Trends: Discussion on recent unemployment data released in the latest jobs report. Pat & Dan share skepticism about the accuracy of economic indicators. Analysis of interest rates and their impact on the housing market.

‍

3. Tech Industry Dinner at Trump's: Mention of tech executives attending dinner with DJT. Brief discussion on the administration's commitment to AI leadership.

‍

4. Broadcom's Earnings and the AI Chip Market: A breakdown of Broadcom's strong financial performance. Exploration of Broadcom's position in the custom AI chip market. Predictions for Broadcom's future growth and market potential.

‍

5. Nvidia's Customer Concentration: A debate on the risks of Nvidia's revenue depending on a concentration of a few key customers. Analysis of the sustainability of Nvidia's growth in the AI chip market. Discussion on potential threats from custom chip development by major tech companies.

‍

6. Salesforce Earnings and AI Strategy: An overview of Salesforce's recent earnings report. Examination of Salesforce's AI initiatives and market reaction. Concerns about the future of SaaS in light of AI advancements.

‍

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

‍

Or listen to the audio here:

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

‍

Daniel Newman: You know, one of the things you and I love to do is have a healthy debate. By the way, we do this privately, we do this publicly. We fight in public like a couple. We love to squabble. You know, I usually say let's just take it out on the bench, but then Pat always says no for whatever reason. I don't know why you never want to do that with me, but one of these days, one of these days, right, buddy? You're gonna do it. We're gonna do it. No, we're not gonna do it because we don't want to get injured anymore. We're old men, Pat. Put the ego aside and stay healthy. Keep your shit together. Healthy, healthy. You need to be able to hold that baby.

Daniel Newman: We are back. It's another week. Tons of news. Episode 275. Pat, can you believe we've been doing this for 275 weeks? Bringing it each and every week, whether we're sick, whether we're tired, whether we're in a foreign country or we are enjoying a beautiful vacation somewhere mountainous with our family giving little snuggles to our grandbabies. Did you do anything on that list today so far, Mr. Moorhead?

‍

Patrick Moorhead: Possibly. You know, I took one of those not vacation vacations. I sometimes have these visions of grandeur which are, you know, I'm just gonna unplug and chill. I mean, I probably worked half the time. I guess if you include doom scrolling on tech in bed at 4am then probably six or seven hours a day, so.But you know what? I'm just glad that six or seven hours is better than 12 hours. So I'm going to count that a. An absolute win. No, I'm. I'm here in the snowmass, chilling. It's good, man. See my grandson. It's just so amazing.

‍

Daniel Newman: Another great pick you sent. Maybe they'll flash pick up here. Tell the producers a cute pick of Pat doing a super snuggle with the baby boy. Thank God he's pretty like your wife and daughters. It doesn't look anything like you, but you know, maybe if he gets Older, he'll get your jacked figure. He'll be, he'll be gorgeous. So I, I hear a rumor that you're sort of back to part of being on vacation and recharging generally comes with great workouts. Are you at least getting the, at least getting the, you know, the killer workouts in with a little extra time you have?

‍

Patrick Moorhead: Daniel, I'm absolutely crushing it in the gym again, feeling pretty good about things. I'm not a hundred percent on anything that has a deal with my midsection because of my surgery, I want to give it a couple more weeks. I do that, but I would say, you know, upper body and leg machines, I'm 100% and seeing gains like crazy. So. Yeah, the great thing about declining is that, you know, you have an on ramp coming back. And you know, it is funny. I published that second year health journey, third year, what's going to happen? And I got absolutely like, nobody's interested in it.

‍

Daniel Newman: Like I said, you're too healthy now. You need to get fat again and start over. People will love it. Be like, I succeeded, then I failed, then I succeeded. Because by the way, people only like you becoming so successful. Then you cross the line and they're like, nah, I don't want none of that. Moorhead, you're looking too good. You're looking too damn good. You're too successful, you're too smart. You're on TV too often. You're just too much of the man. Speaking of being the man, we. What a week. Wow. Big antitrust. I mean, I really thought coming off of the Nvidia week that there'd be like a little bit of a cooldown. I knew that Broadcom was reporting and there were some other tech names coming out this week, but I just thought that was kind of the peak. And then we get a cooldown. But we didn't get much of a cooldown this week. We had major antitrust rulings. We had Trump and the first lady having a whole dinner table full of people, many that we actually know, which is pretty cool. Like, we're like only one step removed now from being invited to the White House for dinner. Get that? Hear that? Pat and I would like to come. There were more rumors. FUD rumors. FUD rumors. I don't know. Mostly FUD these days. If it comes from the information, you got to ask a question. Sometimes they're right. I don't know. Salesforce CEO talks about cutting jobs because of the power of agents. The company also reports. This week we'll hit on that a little bit. And he actually. There was a little sniping, by the way, going back and forth with him and Alex Carp that was quite amusing. And Microsoft extends its tools. And of course, Satya came out as one of the first to support President Trump's AI agenda. We have a pro AI White House, don't we? I mean, policy aside, the White House is all in on leading the world in AI. We got that. We got a flip. Another controversial simulated argument that we're gonna make that Pat's gonna probably win. He always wins.

‍

Patrick Moorhead: I don't believe that.

‍

Daniel Newman: In case anyone whispers. And then we've got a bunch of really interesting market updates. I got a lot of victory laps to take today. I'm gonna find a way to cue those in, and hopefully they figure out a way to cue in the victory lap bell. I bet I've just been on fire lately, Pat. I, you know, batting.

‍

Patrick Moorhead: Are you tired, Daniel Newman? Are you tired of winning?

‍

Daniel Newman: Exhausted of winning. But it's kind of like, you know, if I really could break it down for everybody, it feels like there's a lot of winning and a lot of losing at the same time. Winning in categories that don't matter that much and losing in the ones that do.

‍

Patrick Moorhead: Stop talking about me like that, Daniel. That's me, it’s not you. 

‍

Daniel Newman: This is a we thing. We're in this together. This is a we thing. It's just, you know, it's like being right on X about something but then, you know, missing huge targets that I've set for myself in really important areas over and over again. So. But I did. I did. I did make a bet with somebody that I could get to 405, by the way, on the bench before they'll get to 315. So it's a big bet. And by the way, I've never gotten close to 405. I've been high three. But when you get over that weight, every pound starts to feel like. Like a kilo bestie. Be careful.

‍

Patrick Moorhead: Be careful.

‍

Daniel Newman: Well, I'm not even gonna try it till I'm doing six, seven reps at 375. Like, I'm gonna. I'm gonna get a long way before I even take that stab. Because we've both seen what an injury can do. It sets you back, but not only set you back physically, it sets you back mentally. And that's the bigger thing. You gotta stay. Stay clear in the head. And missing those workouts is an absolute killer. But, Pat, we got a long docket. We want to get through everything for the audience. Thanks everybody for joining us. Let's get to the decode. All right, Pat. I know I was, you know, pounding the table for a long time about Google. Great play, great AI name, this antitrust stuff is not that big of a deal. They're not going to get broken up. A ruling came out this week. Did they get broken up?

‍

Patrick Moorhead: It's amazing. I mean this was the biggest nothing burger I've ever seen in my life in terms of antitrust. Right. Essentially this was the remedy phase and this is where the hammer was going to be dropped. The company was going to be split and browsers would be sold off, but absolutely none of that. There was a lot of Google news this week, but first and foremost the judge ruled that Google is a search monopoly but narrowed the remedies to real data, data sharing with its competitors. It bars exclusive deals which I can't believe they got away with in the first place and there wasn't like an immediate injunction. So you know, I think the other thing that was interesting is, typically, you know, well, actually I've criticized the antitrust people for looking into the future and pretending they know what's going on. But I guess this time it probably made sense where the court actually brought up Generative AI as a competitive report, sorry, as a competitive force and reducing the appetite for any structural type of remedies. I think the biggest winner of this is probably Apple because not only can they keep getting their just incredible money from Google, but also, you know, they could, if somebody decides they want to pay more for, you know, a certain lineup on i17 type of lineup, then they could give that to OpenAI or they could give it to DuckDuckGo or they could do to Mozilla or something like that. So it's browser and it's browser and its search. One thing that really didn't get picked up a lot was that Google's ad tech fine was delayed given all the discussions between the, the different countries and all of the antitrust and tariff types of conversations. So yeah, this ended up just being a big, just big nothing burner.

‍

Daniel Newman: Well, we all like a big delicious nothing burger from time to time, don't we? There's been a few others like that rumor about Nvidia subcontracting services back to one of the clouds to use its own GPUs, making that seem like that was a big round tripping deal when in actuality they're not a cloud provider and they need capacity and and they need managed services and they need, you know, anyways, point is that this is like the media loves these kinds of stories. They love to try to take down tech. They want AI to be a failure. I don't know why, I don't know why we want that so badly right now. It's not. But you know I, here's the biggest thing. Part of what I was sort of thinking with this whole Google situation is that this is old, this is a last generation technology. This, this was like some of the things that we've seen in the past but like breaking them up now as we head into the generative era it's like who cares? I mean search isn't going to be the same and like within a few years generative and search engine optimization is going to be completely different. The advantages that Google has are going to be completely different. The companies could be completely different. And the only thing that's really, that was really critical was the data and they have to share data. So that was the one ruling that came out of it was the sharing of the data and that sharing of the data is going to give more power to creating competition in the future of generative search. So I think as a whole this remedy was pretty right on with what, what's going on. So I think I support it and the market was very pleased. I think it ran like 10% on this news. I did want to quickly touch on another Google item with you to get your take on this Google TPU thing. You know supposedly they're reaching out to some smaller NEO clouds and cloud providers potentially offering TPU in the cloud for external use cases and workloads. We've talked a lot about internal that they're doing a lot of their own training Pat. But like I think I saw you had a couple of opinions on this. What's your, what's your read on that one?

‍

Patrick Moorhead: Yeah. So with Intel moving out with Goudi and none of the other XPUs really getting a foothold in the enterprise or even in NEO clouds aside from let's say G42 with cerebras and Groq. I think it makes perfect sense and I've been, what's the right word? Lobbying all companies involved to dig deeper into that. Right. It's not the Swiss army knife meaning you can't go in and run every single application but what you can do is if you have a very, I would say open industry standard interface that you want to use, it's actually really good, really good. Solution and if we're going to do inference, I mean I've been very clear with this non stop and I know there's this whole idea of tokenomics but all things equal in the long run having an XPU will be more efficient to run inference on than a, than a GPU architecture. If you disagree with me, then give me the evidence and give me the proof on that, on that specifically. And I also think that these are smaller models that are being run, not the larger models. I'm even hearing that they're going to take this all the way directly to the largest enterprises out there as well, which I find a lot more interesting to be honest, than some of the NEO clouds or the smaller cloud providers.

‍

Daniel Newman: Look, this is going to be the big test of the moat too. You know we've talked a ton, ton, ton about the Nvidia moat. We know that the largest cloud and LLM companies, there's a lot of flexibility in the Nvidia stack. There's a lot of developers, we know that companies, companies want to build on it, we know a lot of big companies want to train on but we also know there's this kind of secondary narrative about Microsoft and Amazon and everyone doing their own chips and they're doing it in my opinion it's not because I think they want to have more control and flexibility to build their apps their way, giving them sort of unique differentiation in the future. But I also think a lot of it just has to come down with economics, the dollars of it. It's like they can build a customer, they can, once they can map it and optimize it, their costs go down substantially and that is going to be the mission because if they're going to keep spending CapEx like this, if they can do the CapEx on their platform versus a different platform, it creates higher ROI. And they also obviously can get benefits on things like R and D and the development side of this stuff that they can't get when they're using a third party. So I do think these things can succeed in parallel. I do think there's a little bit of this zero sum thinking that's problematic. But you know, this is going to sort of test the market. We're going to see if there's a scale and adoption. You know, if anything that's probably a little bit more of a challenge for Intel whether they can get Goudi in the market at any scale. And it's a little bit of a challenge for AMD because AMD has been Kind of trying to play the second card. And if Google proves that there's another alternative that has flexibility, software development, and that could end up being a little bit of a challenge for that second chair that I really have said and have thought for the longest time, that should be AMD.

‍

Patrick Moorhead: The most compelling thing to me. Can you imagine TPU plus Gemini plus Vertex AI on-prem and Enterprise? We already have Gemini running. We already have Gemini running on Prem at with Dell infrastructure, and we have bits and pieces of Vertex AI running on that same stack. But. But I think it'd be pretty freaking compelling because nobody, nobody has the entire stack. Right. You would have to have a VMware or Red Hat AI platform, then you would need to find a data platform. And with Vertex AI, it's all built in. So again, you know, the cloud guys are reluctant to bring anything on Prem, I get, because it kind of breaks their operational structural advantage. But I could also see them cleaning up in the enterprise as well.

‍

Daniel Newman: Yeah, well, you've seen them sort of heading that way, so there's definitely an opportunity that they'll continue to head that way. All right, let's. Let's jump down the docket here a little bit. So, you know, last night, first lady Melania Trump posted a bevy of executives from the tech industry, you know, from Sundar and Zuck and Lisa Sue and AMD to, you know, Sunny Madra from Groq. So we saw Shamath there. So you've definitely got this broad audience of both kind of tech executives and leaders, as well as kind of thought leaders and investors and capital allocators in the space. You know, to me, this is mostly more validation of the administration's significant commitment to, to leading the world in AI. I think that, you know, we're seeing this continued investment, this continued sort of outward narrative. Interesting to see Satya. Lisa was on, I saw her commenting on cnbc. But just kind of this overall sentiment that the tech industry and the Trump administration seem to be very aligned on the fact that they're going fast, aggressive, a lot of deregulation, pushing forward with AI is going to be critical for leading the world in space. So I don't know what this was like, this deserves a ton of time here, but I do think it's very interesting to see the White House continue to put its thumb on the scale that we are going to be the world's leading developer and creator of AI solutions and services.

‍

Patrick Moorhead: Yeah, I don't get the timing, to be honest. Other than Trump doing a victory lap. Being able to do a victory lap. Actually he asked CEOs around the room to have a comment and yeah, so he got his own built in victory lapse. But I think we'll figure out there were probably a lot of good side conversations, a lot of lobbying on both sides. The biggest irony to me is that we are still part of the AI plan. There is no talk about Foundry ip and that's the biggest miss And I'm going to start, I'm going to start dialing up the, dialing up the rhetoric on all my channels until I see something come out of that.

‍

Daniel Newman: Well, there you have it. There you have it. So Pat, another rumor that came out this week or comment was it? There was a comment that, you know, H100, 200 is not available. Nvidia I think just probably fatigued with all the FUD lately. That's why they just came head out and said this isn't true. That basically H1 hundreds, two hundreds are available and those that need them now. Love to get your comment on this. And of course some people are saying, well who's buying these things?

‍

Patrick Moorhead: No, I mean, listen there, there is a lot of usefulness. Even though Jensen wrote right before the last GTC, actually at the last GTC that you and I both attended, he essentially said there's no value in H2 hundreds. Right. But the reality is the H200s can do an incredible amount of work. I mean, quite frankly, they're still cranking out frontier models left and right. The latest models from OpenAI, the latest models from X all were trained on H series GPUs. So yeah, there's a ton of need for that. And then if you look at smaller models, if you look at actually running the inference of the applications that you've run, those are the people that are buying these. And listen, I think this has a lot to do with the stock trade, meaning people are creating rumors of anything that will drive the Nvidia stock down. But it really was good to see Nvidia just say no, this is not an issue. I wish more companies would just directly come out. I do understand the difference between they don't want to put fire on them, they don't put gasoline on the flames to make an even, even bigger issue than it is. But you know, having 200 articles speculating and the stock dropping. I think more companies should, should do this and just post stuff out there on X.

‍

Daniel Newman: Yeah, no, absolutely. So let's, you know, I'll just basically very quickly add on this one that, you know, there just seems to be a lot of desire between China's working very hard right now to create a lot of uncertainty. There were like seven different headlines this week that you know, and by the way, if I can just say to everybody out there, just cut the crap, that everything is about replacing Nvidia. This is a massively growing market that has lots of needs. Everything won't be Nvidia. And by the way, even that 5 or 10% of the TAM of total TAM that's custom is you're talking hundreds of billions by the end of the decade. You know, it's not always about replacing. In some cases there will be alternatives. In some cases there will be needs and solutions. There will be things at the edge, there will be things on device, there will be, you know, benefits of lower power accelerators and different architectures and FPGAs. The headline that everything is out to replace Nvidia. And by the way, a new chip from Alibaba that, you know, I know they've done one chip before, which if I'm not mistaken Pat, was a failure. They're not going to suddenly become, you know, software competitive. AMD's been trying this for years now. You know, Rockhound's been working at it for seven plus generations with Google to finally have something competitive. Amazon's still battling an uphill fight to make their hardware easy and usable for anthropic and other things that they're trying to do. Why do we think that every Chinese company that comes out with a first generation chip is going to instantly be able to compete? It's just, it's mind numbing to me. It needs to stop. All right, let's jump over to Salesforce. Their CEO Mark Benioff said thanks to Agentic AI, they're going to cut 4,000 service jobs. So they, you know, we'll talk a little more about their results when we get to bulls and bears. But this is interesting. When we first came out with Agent Force, he was going to hire 2,000 more salespeople to sell Agent Force, which by the way, I thought was super weird. Like if you've got this incredible agentic product, you probably are going to confuse the market with why you would need to hire so many salespeople to sell it.

‍

Patrick Moorhead: Yeah, if nothing else, it was nothing else. It was bad timing.

‍

Daniel Newman: So it was awkward to me at the very least. So now we're getting efficient. But Pat, I mean, here's the thing. It's funny, like I share some thoughts. There was a Stanford report that came out this week about, you know, young people really struggling to find work. And then of course we saw that MIT thing. You and I talked about it last week a little bit here on the pod. Now you've got companies like Salesforce cutting. Microsoft has been cutting. Yeah. You are hearing a lot across the tech space about the implementation of physical AI, the implementation of agentic AI and either hiring kind of freezes, like we're not going to hire more. I think even ServiceNow said they're going to stop adding which, you know, their CEO Bill McDermott has been the most bullish about this revolution being a growth and more employees. And they've said they're going to probably slow down if not completely stop adding. So we are seeing this. This is probably one of the most interesting challenges in the economy as we're still trying to fight temper inflation is the fact that the job market is going to get harder and AI is the reason. I share stuff about this. From time to time people get really angry with me. I do not care. I am 100% right on this. I'm 105% right that this is the case. I'm not saying there aren't going to be things that get built and developed that are going to create new roles. It's just this is happening so fast that the obvious benefits are coming more at the expense of jobs. And in the future, I'm guessing there's going to be productivity gains that we're going to start to identify, but it hasn't happened slow enough that we're able to sort of manage these two things at the same time.

‍

Patrick Moorhead: Listen, this is a big deal and Benioff brought this out because his stock is under fire for not being able to drive AI value. And what he's showing here is that no, we are in fact driving AI value. And look at this. This would n't matter if this happened outside of this. So if he got rid of 4,000 finance people. Right. He doesn't do finance software. Therefore it's not a. Not a big deal now, I expect and I think there were even conversations that he's going to reallocate many resources told toward the sales motion. Right. Which just completely makes sense. You and I have debated the speed of agentic changes and until we have machines buying from machines, I do believe that you're going to need more people to be able to do this. What I'm seeing right now from automated marketing and sales is literally a 10x increase in my spam spam phone calls, spam emails and completely out of touch pretend emails. Like they know what I actually do for. For, a business. So to me it, it, it makes, it makes sense here. But, Mark and Benioff and Salesforce are under attack from horizontal agents that can do anything and he had to flex his muscles to do this. This is classic Benioff.

‍

Daniel Newman: Did you see him kind of crapping on Palantir? Do you see the to kind of go around? He went on Kramer and sort of called out and then I think and then of course Alex Carr being Alex Carr made some comment about him being like the kid that was said I'm so nice in high school but I don't understand why girls don't want to date me or something like that. Like a funny man. He's just, he just isn't personal. No, it's gloves off with him all the time. I mean.

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: But marked kind of the same way. And so they're both kind of yammering back and forth. I mean, you know, when it comes to enterprise AI, I think this is a really critical pivot for Salesforce and I think they've obviously made a big pivot to agentic but making that pivot while starting to turn off the old way of doing things is going to be the big risk it has to take. Right now it's doing the and thing, it's doing the hey here's agents and your traditional Salesforce. But it is really hard as a Salesforce customer. I mean it is really hard to change like once you're on the platform, no matter whether you love it, hate it, you know, it, it becomes really sticky and it, it becomes this, this sort of brain center of the business and you know, they have that advantage of having tons of customers. But I do think the future to your point, these kinds of horizontal agents that are being powered by cloud providers that are being built by platform type solutions, you know, one sort of interface talking to all of your data seems like the future, not a bunch of different software. And so that's going to be the race that everybody's on. I think that's why Salesforce bought Informatic. They're trying to build those hooks into the broader data environment. So this will be one to watch. Pat, last thing here in the decode before we jump on is Microsoft announces some broader access to AI tools for education and government. What's going on there?

‍

Patrick Moorhead: Yeah, so a couple things going on at the same time. If you recall when Trump came in and we had doge, hey, we're going to make some cost cuts to technology. The first thought was, hey, we're using too much legacy technology. But second of all, we're paying consultants way too much money or sis. But generally speaking, we're spending too much money for the tech that we have. And a couple weeks back, we saw what Google did in response to this. They come in, cut deals, freebies for this, that and the other. And here we have Microsoft coming out and pretty much doing the exact same thing. And this covers the US public sector, but also education. So free365 personal for college students for one year. That includes copilot and then, you know, some seriously, some cut prices for federal access to the latest AI capabilities. A lot of conversation about, you know, typical, you know, privacy compliance, bars, education and in the government. Right, because this is cloud based. This is not, you know, some on prem type of thing. But this is really Microsoft playing the long game, right to have a lock in the public sector. And I think Microsoft so far is pretty much cleaned up in this space. And Google has been looking for some table scraps. Even though the workspace is highly competitive. I think what this does is this likely nulls or potentially limits Google's ability to go and take share here if anything gives a choice, I think to different, different agencies, which will be interesting to see how that plays out.

‍

Daniel Newman: Yeah, I think you hit this one on the head. I don't really have anything to add on this one. So Pat, let's, let's rock forward on this. You know, one of the things you and I love to do is have a healthy debate. By the way, we do this privately, we do this publicly. We fight in public like a couple. We love to squabble. You know, I usually say let's just take it out on the bench, but then Pat always says no for whatever reason. I don't know why you never want to do that with me, but one of these days, one of these days, right, buddy? You're gonna do it. We're gonna do it. No, we're not gonna do it because we don't want to get injured anymore. We're old men, Pat. Put the ego aside and stay healthy. Keep your shit together. Healthy, healthy. You need to be able to hold that baby. You know that baby. I hurt my wrist trying to kill a bug on a bleacher last week. So like, we are old dudes. Like, we've got to get that through our heads, that being healthy. But here's something that's not healthy, in my opinion, is having almost 40% of your market covered by two companies. My God, what a terrible idea. That's a big problem. Maybe. Is that really what I think? I don't know, Pat, but Nvidia, 39% of its revenue comes from two customers. And some of the market seems to think that is a bad idea. Is it a bad idea? Is it a problem that they have 39% tied up in two customers? Let's see. Maybe one of us can take that side. God. Okay, so I mean, here we go. You know, you've got a company now trading at, you know, over $4.4 trillion in market cap. And here they go reporting. And in this, in the category that was most important to the business, they put up a puny amount of growth. I know I'm being a little bit sarcastic, but a rapidly declining growth rate and an increased concentration rate with 2 of the customers now generating almost 39% of that revenue. Nvidia is in a difficult juxtaposition in terms of how they're going to go forward. They are basically now becoming incredibly dependent on companies that are building products inside of their shops to compete with them. This is a very simple one for me. You basically have a challenge where you've got an already declining growth rate in the GPU market. It's going to regulate at about 20% by the end of the decade. And yeah, sure, they were up by hundreds of percent a year or two ago, but that's not the case anymore. And now you look at Microsoft, you look at Meta, you look at their other, by the way, huge customers that get them closer to 50 plus percent of all that revenue. You've got Google, you've got Amazon, you've got, you know, you go down the list, you got Core Weave and all these companies. But of the biggest ones, Meta, Microsoft, Google and Amazon, guess what all of them are doing? They're all building their own competitive product. We already talked about this on the show. So if I were an investor, I would think that's a big problem for me if I'm already at a price where they're at a four and a half trillion with declining growth rates and with customers that are building products basically with the goal over time to replace Nvidia in their shops. And of course, here you go coming out of left field. Just yesterday, CEO of Broadcom, Hawk Tan laid the wood, knocking the ball out of the park, announcing a new open AI custom chip. That's just one of the chips they do because they're already doing it for Google and I'm pretty sure they're going to be taking Microsoft forward. Now they've got OpenAI. Oh, and they've got ByteDance and they've got tons of other big cloud companies that are all saying we want to vertically integrate, we want to increase our margins, we want to customize our own software, we want to control our fate and we do not want our destiny to be in the hand of a company that's taking 73% of the margin out of every opportunity. Pat, this isn't even a debate. There's nothing to debate here. This is a problem. And if I'm an Nvidia investor, I am nervous. I am going to be running for the hills because only five, six big customers depend on being able to keep growing at this rate. Too much pressure, too unlikely. This capex is unsustainable. These companies want to get the R and D credits, they want to increase their margins. They're going to all go the Apple route, they're going to vertically integrate and this means Nvidia's market share and growth is going to suffer.

‍

Patrick Moorhead: Daniel in a normal market this might be an issue, but we're in a completely abnormal market, Nvidia, that has all the leverage in this relationship. And the two companies we're talking about are Meta and Microsoft. And if I look at both those companies, their XPU a hedge is behind everybody. I mean, my gosh, Microsoft made 64. Excuse me, made. Yes, sorry, made 120,000 Maya's for the entire year of 2025 estimate. And that compares to literally millions of Nvidia GPUs. And even on the AMD side that's running close to a million. I mean, it's just crazy. So there's absolutely no leverage. And then if I look at what's going on at Meta, you have, if I'm looking at at least the cost, their MTIA is selling well. But let's not forget, like the first MTIA was really a recommendation engine. That's all it did, right? And there's four different versions that are in the market at the same time and this MTIA version 3 will do 190,000 units this well, I mean, you know, compared to Microsoft's, a lot higher. And then when you add MTIA v1.2 and v3, it just, you know, it's a lot bigger but still in comparison to GPU and Nvidia, it's just small. So the point I'm making is that Nvidia has all the leverage, so it's not, it's not a risk, it's not a risk at all. And then at least on the Microsoft side, if I look at Azure, Azure AI that's pretty much optimized for Nvidia GPUs only AMD GPUs are used extensively internally, a lot less for what I would call IAAS or PAAS services. So I think just by just looking at these facts, this is why.

‍

Patrick Moorhead: I think this is very manageable for the next couple of years. Long term could be very different. And the other thing I want to look at the final case I'm going to make is the market looking at Meta. Meta is absolutely an absolute juggernaut right there in fact seems to be the biggest beneficiary of AI and how they've been able to monetize this if we look at their recent earnings. So if they would decline, if Meta would decline, which they're not, that would be an issue. And let's look at Microsoft, right? I think we all agree that Microsoft is the most balanced company out there, right? They've got a smaller consumer play, but when you look at their commercial play, that's really where it's at. So are they as a juggernaut going to stop anytime soon? I don't think so. I mean, barring some economic disaster that freezes business spending. So Nvidia holds all the cards and these two companies are not going to slow down anytime soon. So that concentration risk, it's not even there, at least for the next few years.

‍

Daniel Newman: There we go.

‍

Patrick Moorhead: So what do you really think?

‍

Daniel Newman: I. I don't think the concentration risk is real, but my reason's different. My reason is, by the way, that UBS report was great, really, really great data on the chip font and expectations. I think they're underestimating it based on our forecast, but I do think the level of accuracy was really, really good. We'll have to see how that, that shapes up. Jensen's right about ASICS being canceled, delayed, but I don't think he's right about the idea that they'll never happen. I think they're gonna. I think this is an existential thing for these companies and they're gonna do it, but I do think it's gonna take longer, which is gonna be a benefit to Nvidia over the next few years. The biggest reason I don't think it's a problem is I still think the external workloads, the people building on small volumes and wanting to run AI in their enterprises are going to do it in the cloud, they're comfortable in CUDA and the developer environment of Nvidia, and that basically the five or six biggest customers support like 5 million enterprises. And that's the big thing that I was really leaning on about why concentration is not an issue, is they sell to the next 5 million companies. So everybody's using these clouds. It'd be one thing if they were buying it and only using it for their own reasons. I would be really worried about that because that's where I think the falloff comes, Pat, is those internal workloads definitely are going to get. Get chopped off and they're going to go to the ASIC, but the external workloads I think stay with a lot of the enterprises they're already building on Nvidia. They're going to stay with Nvidia. So my take, Pat, is it's really the scale is that this is really good channel distribution. They have a good channel. They have 5-7 companies that sell to 5 million more companies. And that, in my opinion, is good channel strategy. And so I think that is the reason it doesn't bother me at all.

‍

Patrick Moorhead: Very good.

‍

Daniel Newman: What do you think? How was my argument, though? Pretty good. Pretty compelling.

‍

Patrick Moorhead: Yeah, it's something that, you know, we, I think hopefully the audience recognizes and appreciates. We come, we come at these things from different angles. I know what uninformed people are like.

‍

Daniel Newman: Oh, you people are.

‍

Patrick Moorhead: You're the same person. It's like we are not the same person. And we have, well, not completely different views ultimately, but we might have a lot of different reasons why, why we think of something.

‍

Daniel Newman: Hell yeah. Hell yeah. You heard it here first. All right, buddy. There was a busy market week. You know, things, it's been hot and cold, it's been up and down. There's been some economic data, there's been some news. Some new restrictions that are changing things or a couple of earnings out there. Pat, why don't you and I break down what happened in March markets this week. Bulls and bears. Let's go. All right, Pat. Unemployment shooting up the job and labor ads are crap. You know, you heard about cuts. I mean, rate cuts coming, job cuts coming. What do you think about all the economic data that landed this week, Daniel?

‍

Patrick Moorhead: I don't believe any of it anymore. And I think that, by the way, unemployment data is probably more accurate than CPI or PPI because you've got unemployment checks, right? You've got things you have to file with a state or federal government. So I do buy into that. I mean, listen, if you want to take it at face value. Right. Interest rates are too high. We need to, you know, we need to cut interest rates immediately. And the interesting part is that the unemployment rate hits as we're deporting a lot of people out of the country. So it's kind of a double whammy here. I do think that the tariffs are too good to be true, particularly for US Manufacturing because you've gotta, you know, find a new source for your brake pads or something like that. And that takes a long time. So yeah, Daniel, it's hard for me to take any of this at face value, but if we say, if we think that this is accurate, yes, the economy is in a slowdown and we probably need an interest rate. It's as simple as that.

‍

Daniel Newman: Yeah, I mean, my take on it all is just the average household, the average household is getting squeezed a little bit. High rates, the mobility of people is down. We're seeing, it's never been a better time to rent and a worse time to own. That's because all those higher interest rates actually didn't break the pricing of the market that much. We did not see housing prices come down that much. There were a few bubbles where you saw meaningful declines. You certainly saw like here in Austin where there were like 50 site unseen buyers paying cash at 50% over asking price. That's kind of gone away, but the prices have stayed really high here. So then on top of that, like I said that 2% that people were getting, 2 and 3% mortgages are now 7 and 8%, car loans were 3, 4%, another 10% credit cards. You know, when you add that difference to the, the little bit of interest dollars that now flow in, if that interest was reduced, you put a few hundred or a few thousand dollars back in a household each month, each year, that money one gets generally probably spent back into the economy. We're not great savers. If there is savings, that's a good thing. But we're seeing people's savings decline. We're seeing increases in credit defaults and increased credit balances. People are leveraged more than ever before. So I think we've already picked an economic strategy which is growth to the hilt. And so if we're going to only go for growth on the economic, on the macro level, we have to enable people to keep investing and spending and growing and moving into bigger houses and buying better cars. You know, lower interest rates enable you to get more for your money. So it's the way you keep the market flowing. So that's what I'm going to say.

‍

Patrick Moorhead: Yeah, if we look at the data too. I mean, you know, education, health services gained, which, my gosh, how many administrators do we have to add to health care? Like, have you seen the disparity over the last 20 years between doctors and administrators? And the big losses were durable goods. Right. That's manufacturing, business services and the government, which doge came in. So it's very explainable to me. The only, I mean, listen, adding $46,000, 46,000 jobs to education health services. To me, I see that as a red flag somehow. Anyways, let me, I won't go down that rabbit hole. Durable goods and business services concern me, not declines in the government, though I think that's a positive.

‍

Daniel Newman: Yeah. Yep, yep. All right, let's, let's jump into a few earnings here, Pat. We had Salesforce, HPE, Broadcom. Let's start with Salesforce.

‍

Patrick Moorhead: Yeah. So in any normal time, I think the stock would have gone up even though the forecast was a miss, just because of the 11% year on year revenue. But the challenge that plagues Salesforce essentially is that their AI number isn't big enough. Right. When you've got companies like IBM, you've got Cisco, you've got Nvidia, you've got Broadcom just absolutely crushing it out there and then you've got this, I'll call it the meme that won't go away. And by meme, I'm not saying it's not true, is this idea that horizontal AI agents will reduce the upsell or the upcharge or over the long term reduce even the need to have certain enterprise SaaS applications. So yeah, this was the, this was the, the issue. I mean, you know, some standout stuff. I mean, data, data cloud. Right. And AI ARR up over $1.2 billion, up 120% year on year. The precursor to any of the AI magic tricks is the data cloud. So that's a very positive thing. By the way, stock reaction. Right. Down 5 to 8%. The day after that was with a 2.6 billion dollar buyback authorization. Right, sorry, $2.6 billion to shareholders. But the buyback went from 20 billion to I think 50, 50 billion. And, and the stock going down even with all of that malarkey, I think is not a good sign.

‍

Daniel Newman: Yeah, this to me stays under the lane of, you know, there's a fairly significant school of people that think SaaS is in big trouble. Yeah, you know, we think Salesforce can make the pivot, but it does have a lot of work to do to prove that it can make the pivot. It does have to be able to show that this agentic investment that it's making is going to not only show growth, because it is, but it's also going to have to replace some of the revenue that will fall off. They have to be able to basically turn it and turn this into a well oiled machine. But again, you know, they kind of came in, I think their guide was kind of in line. You know, overall, I think they reaffirm but like, you know, not upside. And this is one of those where it had kind of stabilized and run up a little bit into earnings. But this one's selling hard because it's one of those companies that people feel has a higher potential being impacted by this SaaS. This SaaS software industrial complex. You know, we continue to think they will be okay for all the reasons I mentioned up top when we were talking about stickiness, hard to move off of. But I do think that turning agents, which they're making and then of course turning to horizontal agents, is gonna be the second thing. It can't just be agents plus SaaS, it has to be agents as a enterprise wide initiative and their data cloud is gonna have to be the data cloud to support it at scale or else, you know, we're going to start to see those, those horizontal companies be more and more competitive and create more and more risk. So you know, we like it, but there's a lot of work to be done, a lot of things to be proven there. So we'll have to keep an eye on that one. Another company that's done well and actually seen, you know, quite a bit of upside this year has been HPE. Now HPE, you know, first of all, the Juniper deal was accretive in this quarter so obviously a part of it was accretive. So their numbers were a little harder to read through because you're looking at a number that includes some of Juniper in it. So I'm gonna not talk too much about that part. I'm gonna, I'm gonna keep it a little bit at the high level about what I was looking for. HPE, Lenovo is a similar one that is playing in a very important area of AI and that's the enterprise side. They have been very committed, if you listen to CEO Antonio Neri, talk to him this week, he's been very committed that they are not going to chase the bottom when it comes to servers. They're not going to be racing for the volume they're going to win higher quality deals. And if there was one sort of thing that looked really good. But also Chris, is all these companies selling AI servers are dealing with incredible margin pressure right now. It's great for the top line but at the expense of margin. They did a good job kind of holding down, but you know, they too are seeing some pressure on their margins. I like HPE broadly as a play right now because I like how they are fully stacked. I like what they've done in terms of having the network, having servers, having their on-prem services and then of course having a security angle as well. They are really thinking about the company, you know, with a much more of a breadth approach. And this Juniper deal for them really did add a lot for that kind of future of on prem hybrid enterprise AI environments. And if I'm, you know, an investor and I'm thinking about the company, that's what I'm super encouraged by is the fact that they really can start to stack the solutions. Getting that Juniper deal done was really important. The other metric that I really, really liked was I, I believe their ARR numbers are just absolutely screaming right now. They're up over 3 billion. Now remember this was a four or five year promise. Everything is 77% up from prior year. Over 3 billion and adding thousands of new customers on Greenlake. So if you kind of want to know if the hybrid AI, hybrid enterprise cloud story is still a thing, HPE is really zeroing in on this. And now like I said, with their broader portfolio, they're focused on networking and security at scale to go with what they're doing in compute. Pretty optimistic about it.

‍

Patrick Moorhead: Yeah, I thought they crushed it. You know, their stock was up two bucks. Right.

‍

Daniel Newman: Which is a huge move for them by the way.

Patrick Moorhead: Yeah, it is. And then, and then, you know, the whole market is kind of sold off based on that. But you know, they got that initial kicker which I think they deserve. And I spend most of my time really researching the computer. It was great to catch up with CEO Antonio Neri to help understand this a lot better. I really liked the AI systems revenue. I like the backlog at $3.7 billion. And I think he really articulates sovereign cloud and enterprise demand really well. So they are seeing and by the way, their brand in Western Europe is, it's really good. Right? They can really crush it there. I mean the biggest challenge in Western Europe is the amount of companies who need this and are driving there. I think the tariffs and the friction between the US and the EU actually will drive a ton of business for HPE. Yeah, the ARR of $3.1 billion on green is up 77%. I'm just absolutely blown away like this. And if I have time I might do this ARR creep percentage in whole dollars compared to what Cisco has been doing as they went through this. Like you Daniel, what I think the most interesting thing about the company is its full software stack. I haven't seen a lot of evidence that the software layer is coming through and driving big benefits. But what I can tell you is that if you put these two networking entities together they do have the potential to be the second to Cisco. I'm not saying they are there yet. I think the core switch is difficult, it's easier at the edge, it's a lot more difficult inside of the data center. But I'm very optimistic about what HPE is going to be able to do on the edge with networking.

‍

Daniel Newman: And there we have it. One last earnings before we send everybody on their merry, merry Monday or Tuesday or whatever day it is you're watching this way. Broadcom. Pat Broadcom. Were you banging the table? I'm bloody.

Patrick Moorhead: I, I see I, I was but I just, it's so funny, I run out of gas like, like I make a, you know, I'm like Broadcom is going to absolutely crush it this year and they do and then I, I lose my luster to go and say see, I told you so. But, but no, I mean listen, hey, I know that, I know that you know people are, are crap talking about the company. So actually XPUS is a category, we talked about this. But if you look at Broadcom with Google, with Meta, with ByteDance with OpenAI now likely with Apple, likely SoftBank and ARM, that's just absolutely crushing it. And what I can tell you is that Broadcom doesn't do one off deals, right? They won't do a deal that's just one quarter or sorry one product line because they know that a, they can't make money on one and they also know that typically the first go round is not going to be very good and we have seen some, I'll call it teething pains with companies like Microsoft as an example that it's been quite challenging. The other thing that they don't do is they don't do it without a ton of NRE upfront and they mentioned that on the call they did receive payment for this mystery $10 billion customers out there. And lest we forget that AI data center net requires hardcore networking in addition to the compute and they are driving on that as well. Good stuff coming out there on VMware. You and I talked about the Explore show. This company is literally firing on all cylinders. And I still don't think they're telling everybody about what they have in the bag and what they don't have in the bag they haven't talked about.

‍

Daniel Newman: Yeah, I mean, look, I just, I do the math really simply, you know, 115 billion in custom AI chips by 29. We think Broadcom gets 80% of that revenue. So, you know, they're at 5 billion a quarter now of AI chip and networking specifically. And so, you know, we think there's a massive amount of upside from here for the company. And so we're, we've been very clear. You heard my argument. It was simulated. But I meant everything I said about the growth of XPUs and what these large hyperscalers are going to do. And I think it's just getting started. And I think while we're seeing declining growth rates for, for GPUs, by the way, against a very large number, I'm not saying it's not a very large number. It is. You're going to actually see acceleration of growth rates against a smaller number. But people love a growth story. So the growth here is palpable. Broadcom is kind of wiping the floor. And, you know, you talk about a tale of two companies, you know, and I, I like Marvell, but like Marvel just couldn't get out of its own way on its earnings call. And Hawk did like a, you know, know, Hawk did a man, he delivered a manifesto on how to basically say nothing and say everything about the, the partners and companies that they provide AI chip designs for. And remember, both these companies, but especially Broadcom has full. It's not just the compute and the, in the XPU part, it's also all the networking, which is a massive growth factor. It was the bigger growth factor for Nvidia this quarter. It's the, you know, broadcast focus on open standards networking. And so if you're building an XPU or a, you know, like I keep saying, maybe you're gonna have an XPU with Broadcom partnered with arm. It's gonna have some custom IP in it. It'll be an ARM Broadcom CPU XPU with some Intel Foundry work getting done to it. I mean, I'm just throwing some ideas out there what could happen here. Probably be produced for OpenAI and maybe for Oracle down the line line. And heck, this thing might even become the. One of the Main inference chips for the Stargate project. You know, just ideas. I have no idea if this is true, but I like throwing these kinds of things out there. But there's a lot of growth out there for Broadcom, so. And by the way, Hawk is just an execution machine. He's like my hero. Like, if I could be like him, if I could hire like him. Like, you freaking make your number when you work for Hawk, Tank. There is no other way. You make your number and you execute and he comes to every earnings. Just like, here's my numbers, I hit them all. Nothing missed. By the way, here's a growth on the guide. You know, I mean, Pat, this is gonna be a two trillion dollar company calling it. All right, that's it. Let's go. What a week. Thank you everybody so much for joining us. Pat, thanks for doing this while on vacation. Go hold that baby. Go crush a workout. Try to not take too many calls. It is Friday, and I think you're coming home tomorrow or Sunday when.

‍

Daniel Newman: Yes, you're coming home. Like, try to do it one day. I mean, you've been a nightmare. You've been texting me, calling me, trying to work. This whole week I've been ignoring you. Don't take it personal. I'm trying to let you have a little vacation. But I really do appreciate you showing up today. I appreciate you, all of you showing up, being part of our community, joining us for this show. 275 episodes in the bag, waiting for that White House invite. So you all know what to do. Get us invited to the White House. We are the ones calling the balls and strikes when it comes to the future of AI. But for this show, this episode, for this Friday recording, for whatever day it is that you're listening, we have to say goodbye for now. We'll see you all later.

‍