Could AI make your job obsolete? Episode 265 of the Six Five Podcast tackles this question and other hot topics. Patrick Moorhead and Daniel Newman explore Amazon’s AI-driven hiring slowdown and the potential impact on white-collar jobs. From HPE Discover highlights to OpenAI’s legal battles with Microsoft, and a debate over the ethics of AI companies using copyrighted data for training, the boys are back with insightful commentary on the rapidly evolving tech landscape. This week’s handpicked topics include:
- Intro: Recent events, including HPE Discover and the Six Five Summit
- Amazon’s Announcement About Workforce Reduction: This aligns with broader industry trends, where AI and automation are reshaping workforce needs across various sectors & spans Amazon’s diverse operations, including physical AI and robotics in warehouses, autonomous delivery systems, and white-collar knowledge work. (The Decode)
- Microsoft and OpenAI’s Partnership: Examining the complex relationship between Microsoft and OpenAI, and the definition and implications of “AGI,” Artificial General Intelligence. (The Decode)
Intel’s Strategic Moves: Analysis of Lip Bu Tan’s recent shifts at Intel, including its automotive division shutdown. Plus, speculation on Intel’s future focus and strategy. (The Decode) - Fair Use and AI Training: A debate on the use of copyrighted material for AI training. (The Flip)
- Market Performance and Earnings: A review of Micron’s recent earnings and market performance, and a look at the overall market trends and AI-related stocks. (Bulls & Bears)
- NVIDIA’s Market Performance: NVIDIA stock climbs to all-time highs & the factors contributing to their success in the AI market. (Bulls & Bears)
- AI and Market Competition: Multiple winners in the AI chip market and an analysis of potential market share for companies like AMD, NVIDIA, Broadcom, and Marvell. (Bulls & Bears)
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.
Watch the episode here:
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.
Transcript:
Daniel Newman: And by the way, one more thing. There can be more than one winner. AMD can win because if AMD gets from 5 to 10% that’s a win. They go from where they’re at to 50 billion in AI chips. Nvidia can go 300 billion. Broadcom can go to 70, 80 billion. Marvell can do 20 or 30 billion. Like, I think there’s a lot of people that want this to be zero-sum. It’s not, it’s a dumb take. I wrote about that on Twitter. You can look at my X post about that, right? Quit it with the dumbness, people. You can actually like two companies in the same sector, especially in a secular trend.
Daniel Newman: You know, in those, in that preamble Pat. We dress up, we look like grown ups but when we get on the actual show, you know, the poor audience has to see us and we’re basically in our pajamas. Well, you look pretty good.
Patrick Moorhead: I mean we’re just, yeah, we’re just doing real life here.
Daniel Newman: That’s a great shirt. What does it say on it?
Patrick Moorhead: It says Six Five Media and there’s a couple lessons here. But the primary message is that whining does matter. Even though I pay for these shirts. I never got any of these shirts, but now I have these shirts. Dan did a dumbo drop in the airport and I now have so much merch between Six Five Media and Signal 65. I will never be able to complain again. I’m not going to tell you what size it is. It’s kind of embarrassing. But it’s not extra fat like I used to be. It’s actually a medium.
Daniel Newman: I’m going to glom onto that, that whining thing. You do whine a lot. That’s a good characteristic about you, though. That’s why you’re so irresistible. It’s the whining. I love you for it. Only when you whine. I got it. Great, great. How is that? Still tastes like crap.
Patrick Moorhead: It’s so good. And I wish this show was sponsored by them, but it’s not. So this is Only What You Need. Branch chain aminos, nine essential amino acids, zero sugar, all plant protein. That’s good stuff. 32 grams.
Daniel Newman: Yeah, but let’s be honest, it’s the peptides that are actually making a difference, that stuff.
Patrick Moorhead: Listen, I don’t know, I don’t know. First of all, I wouldn’t – couldn’t even tell you which peptide. Or maybe it’s the HCG, or maybe it’s something else. I have no idea. I’m taking so many things. I’ve gotten really good. So I take about 15 pills in the morning – supplements. I can pound them all in one, one big bite now.
Daniel Newman: Well, that’s great. There’s this new app called Prove It that you can scan your supplements, and it tells you if they actually work. It’s pretty cool. You should check it out. It’s kind of like that other one that does your food. It’s called Prove It for Supps. So you should start using it. But talk about supplement. Let’s supplement everyone’s week and tell them what happened in tech Pat. It was a busy, busy week. And we’re back with episode. This is 265. I can’t believe that long, long time. I’ve aged with you reverse age because of the peptides. But the fact is is I’m a little younger than when we started this thing. But we’re 265 episodes in and this show is going to be be a good one. Pat, you know, I would be missing an opportunity if I didn’t say What a great Six Five Summit. It was a lot of fun. I think you were over in Munich drinking big tall beers. We actually haven’t been on the pod for a couple of weeks now. He’s scanning his supplements, which is great. And you know, you were happy. And I think it’s important the audience knows that you were a miserable son of a b****. For like the last two years and you found happiness somewhere along the line. I mean, you were like calling me. It was like, “Dude, you’re like the happiest Pat that I’ve dealt with for a longest time.” Was it the bread? Was it the fact that you could eat bread and carbs? Was it the beer? What? What was it that made Pat so happy over in Munich and do we need to ship you back?
Patrick Moorhead: I literally have no idea, Daniel. In fact, my United screwed me and canceled my flight. So I got there 24 hours early. I spent the first 24 hours in the Denver airport and he was still happy. Did that thing. And then I worked two days and then, you know, I just literally hung out for two days with. With my wife. And I don’t know, I. It could be the pretzels, could be the spaetzle, could be the beer. I have no idea. But my stress was in single digits. First time I’ve ever seen that, and my sleep was. Was just off the charts.
Daniel Newman: That’s great. I gotta say, it was like, the first time you called me, I was like, I’m excited to answer this. It’s not going to be a complaint. He’s not going to be complaining about it. I’m just kidding. Anyways, this is us having our public love.
Patrick Moorhead: It’s therapy, folks.
Daniel Newman: It’s.
Patrick Moorhead: It’s therapy. Than a therapist, by the way, that.
Daniel Newman: Does occasionally cancel his therapy sessions to talk to me because I’m. So we’ve got a busy slate this week. You know, the Six Five team was all over the floor at HPE Discover. We’re going to talk a little bit about that. We talk about some news with OpenAI. We’re going to talk a little bit about Amazon. We’re going to talk a little bit about Microsoft, Deepseek. We’ve got a great flip. We’re going to talk about, you know, whether or not, you know, everything on the Internet is fair use. We’ll get after that, and I will win another debate, and then we’ll hit a little bit of Bulls and Bears at the end. So we got a good show, Pat. Excited to be back. It might only be a week. We’re heading into the Fourth of July. So, for those of you getting this during the 4th of July week, you may miss us, you may not. We’ll see what ends up happening. But we hope you have a great one. We hope you all attended the Six Five Summit. It’s awesome. But all the sessions were on demand, and if you want to catch those, you can. And we’ll put some links in the show notes, Pat, but, you know, we could actually just keep this therapy session going and BS and this and that, but maybe even more fun than talking about the week’s news. But let’s get after it. Let’s jump into The Decode. All right, Pat. Our week was consumed in the beautiful Palazzo Hotel in Las Vegas, where HPE Discover, you know, big week, Pat. Break it down. Give me the recap and the big takeaways.
Patrick Moorhead: Yeah, so first off, we. We broadcasted live there. Make sure you check out all of the action there. But it was really, you know, three macro stories. The first one was AI, the second one was Hybrid Cloud, and the third one was Networking. I did find it super interesting that when Antonio was on the stage in the sphere, the first thing he talked about was networking. Right? I was like, wait, that’s the number one thing? But I think between talking to the folks at HPE and even getting, you know, running that by you, it was really all about, hey, saving maybe the bigger, the bigger stuff for last. But you know, ironically, I think hybrid cloud is the number one thing, number ONE thing that they can be doing with their software, that’s one of the things that, that makes HPE super unique is it does have a full stack all the way from the infrastructure to the apps layer to observability, hybrid cloud. You know, I was pretty hard on HPE in prior years for them, saying they’re all in on hybrid cloud, but they didn’t have any connectors to the public cloud. Now to be fair, they had one storage connector to aws, but that was it. But they did what I think was the best acquisition I’ve seen them ever make. A company called Morpheus. And Morpheus not only enables a hybrid application stack across containers and virtualized elements, but when you connect it with something like opsramp observability, things get, get really, really powerful. The number one announcement for me was Greenlake Intelligence and that was essentially pulling every element of operations into one stack, whether it’s networking, whether it’s storage, whether it’s compute, whether it’s security, observability.
Maybe maybe the, the best thing to, to call it is is the super app. I, I got kind of, I was kind of thinking about this autonomous IT that used to be discussed that, that HPE didn’t use this term, although they, they were one of the first ones to, to talk about it. This is what it brings. And just to close this part out, Green ops, resiliency, finops support and orchestration, in addition to the networking, security, storage, provisioning and observability that I talked about before. One final note on this, and I’ll put it over to you Daniel, is I’m still uncertain as to, on the AI side whether they signed anybody up for their private AI platform. I may have missed it, but I don’t recall hearing any customers discussed in, in context there. And you know, last year I ended my analysis by saying, hey, I’m looking forward to seeing the people that are using this. And I think that question in all of the software, even though they knocked out of the park on a core value proposition that people actually start buying because in the end true value is determined on. There’s a funnel, right? You’ve got people trialing it, you have people using one piece, then you have people that might be using seven pieces of a stack. So I’m super interested in learning more about that in the future.
Daniel Newman: Yeah, I think they, they hit the big three pretty well. You know, I’ve said this a few times in some of our interviews, and we’ll drop those links. We talked to, you know, their CTO, Fidelma Russo, we talked with Antonio Neri on the Six Five as well their CEO, and had some great content. Of course, they kind of broke it up that, you know, you really were, you were a bit enamored by how they led with networking. But you sort of heard Antonio talk about networking being there, being a networking company. You know, he alluded to his confidence in the Juniper deal getting done. And maybe that’s a differentiation where they don’t want to, you know, it’s, it’s interesting though, of course, because their whole business looks a little more like Cisco’s if that’s how they start to talk about, you know, especially now that Cisco’s reentering compute more, and playing the Nvidia stack card. And of course, they’re having a bit of a, you know, renaissance right now. Cisco is coming back at all-time highs. But of course, there’s also some very different offerings in their portfolios. You know, GreenLake has been way ahead on that private on-prem “everything as a service” experience, what they’re doing. GreenLake intelligence interests me. The open data software stack, I think, is something that’s very interesting. The, the control plane of on-prem and the cloud. And of course, I think you said that well about PCAI, is like, I would have liked to get a little bit more indication of customer consumption because right now what I am really trying to figure out is really what is the differentiation from OEM to OEM, from Dell to Lenovo to HP? HPE is clearly not chasing the hyperscale deals the way Dell is. And I mean, that’s also reflected in numbers because their numbers aren’t nearly as big.
But the enterprise business seems to be consuming more and more. And I think Antonio alluded to being a quarter, a third was enterprise. And I think they’re going to do better there because what is basically being constructed for enterprises do not look like hyperscale AI factories. They are very miniaturized. It’s a handful of GPUs. You and I have been in rooms where, you know, executives of very large enterprises have talked about how they’re doing most of their AI work with just a handful of big GPU servers. It’s not racks and racks and racks of servers that are required. And so it’s starting to maybe demystify what the enterprise AI stack looks like versus what the hyperscale stack looks like, but I think that’s going to play out a little bit more over time. We’re still very early there, but overall, you know, I think the indicators were pretty good. Customers seem positive. You had more time with customers because you were part of that C suite experience. But the thing I keep getting when I do spend time with the customers of these OEMs is the enterprises just are not moving nearly as quickly as the sort of consumer and hyperscale AI narratives are. Which means that companies that are a little bit further behind in their tech aren’t necessarily in trouble yet. I think it’s going to. It’s slow at first and all at once is the way I kind of describe it. I think we’re going to see this move a lot faster in the next year. But the sovereign clouds, the, the air gapped data, things like that and security are going to be really important for AI in the enterprise because the no one can afford for their, you know, very important proprietary data to leak. And so companies are trying to be a bit more careful. So overall positive. Show a great experience and a lot of great Six Five content. So let’s make sure everyone out there will link you to it. Check it out. Worth watching. Pat, let’s keep moving here. We got a next topic. This is a fun one. Let’s talk a little bit about Microsoft and OpenAI’s news. There’s a couple pieces of news this week, so I’m going to kind of hit them both a little bit and we can wax poetic about this. But like, first of all, there’s this feud over AGI. And by the way, I don’t know about you, buddy, but I’ve not heard AGI mentioned all that much. You know, in the beginning it was like everything.
Patrick Moorhead: Yeah, it was like everybody was trying to shut it down, like if you were a D cell, right? It’s like the. And even Musk, right, was in that camp, right? It’s going to destroy humanity. And then he took that and started up Xai.
Daniel Newman: Yeah, it’s an existential threat right now. Until I built something and made a few billion dollars on it right now. “Now it’s okay.” But one of the interesting inflections for the Microsoft OpenAI deal is that their relationship runs the course up to AGI. So basically, you kind of look at the Apple narrative about what reasoning models are and aren’t and the patterns versus true intelligence and the idea that we’re going to hit this inflection, where it’s going to go beyond this sort of very sophisticated patterns based upon mega volumes of data to truly being able to reason like humans. And when it gets to that point, which by the way is completely undefined, and there’s actually no real way to define it, supposedly the partnership ends there. And Microsoft basically does not get to tap into that technology, whatever that is. Meanwhile, Satya is over them, he’s under them, he’s all the way around them. And their entire future basically, is being held up by whether or not Microsoft is going to allow them to become a public benefactor corporation that can go public. The money that they’ve been promised by Softbank, significant, you know, I think three quarters of that money. So of the 40 billion, they only got 10. The other 30 was basically they had to be able to declare the ability to go public. And this is all about an exit for all the other investors in the company. And right now Microsoft basically as, as the Donald would say, holds all the cards. And so there’s all this feuding going on. Supposedly Open AI is going to sue Microsoft for antitrust, maybe winning the court of public opinion. But the diabolical diva that is Sam Altman, I just can’t imagine there’s a lot of sympathy right now. And god, I can just imagine him testifying in front of Congress. Just the silliness that would come, you know. But anyway, I digress. Let’s, let’s pause there. Pat. My take is Microsoft has the cards. You know, artificial general intelligence (AGI) is farther away than most want to acknowledge. And OpenAI, whether they have that agreement or not is in a really difficult juxtaposition until, until Microsoft, they find a way to get Microsoft on board.
Patrick Moorhead: Yeah, I think the devil’s in the details. I mean, terminating Microsoft’s tech access, you know, what exactly does that mean? And I think that’s negotiable. And also, there’s a definition of AGI. And even, even what? That, you know, what, what, what does that actually mean? Right. Has that been established contractually? I think you had even mentioned that. So, to me, they both have a whole lot of leverage here. I think Microsoft over time, has lost leverage because if you look at what Oracle is doing with them, standing up the hardware for them to scale, I think that takes one of the elements away from Microsoft. And then the other thing I look at is, well, what are you going to do with the current customers who are, you know Azure OpenAI on there? I think if Microsoft has those, I don’t see them just running away, but, there is the risk, depending on how difficult or easy it is. You can’t just move out of Azure OpenAI. What happens is they’re tied into specific storage, specific instances, specific, maybe even Microsoft data fabric. It’s not like just pulling it out. So I think that we’ll have a negotiated settlement. I would not want to mess with Microsoft’s lawyers because you will always be outgunned probably 100 to 1. And I think that there will be a negotiated settlement on this.
Daniel Newman: Maybe we’ll have to debate this one later. But the tough part is Microsoft has a ton of paper value in their billions of investment because the company’s worth, you know, they started in with 10, they put in like another 10 and now it’s worth like 300 and they own half of it. They also by the way take, I don’t know if you knew this, but Microsoft gets 20 of their revenue off the top. I think you shared that with me, actually because I mean that would be a deal you would do.
Patrick Moorhead: A should have done.
Daniel Newman: Should have done. The net-net is, like how are they supposed to function? It’s like so handicapping and basically no major decision can be made like going public without Microsoft’s approval. And so Microsoft does not benefit for OpenAI to become too successful. And you actually see the route Google’s taking, you see the route Amazon is taking, you see the route Meta is taking. And if Microsoft, straps its, its wagon to open AI too deeply, it basically becomes the sub of the relationship, not the dominant player. And so it’s, there’s so much risk for Microsoft, they have so much control. And so I tend to believe OpenAI is in a lot of trouble right now. Except you know, the only thing that’s really going to benefit them at some point is if Microsoft really wants to get their money out, they’re going to need to allow the company to transact. And so, but at the same time, if they’re doing 500 billion a year in revenue or whatever crazy number Sam’s projecting and you know Microsoft’s gonna get a hundred billion dollar a year dividend check on cash flow. And unless they can prove they’ve reached AGI, which I’m sure that will create its own court battle later, I just think this thing could feud on for a long, long time.
Patrick Moorhead: Have you ever even thought about why? Why is the AGI? Why? I wonder where that came from.
Daniel Newman: That was part of their original negotiation.
Patrick Moorhead: No, but, but what’s the rationale for it? Can you even guess on what that might be. I mean the only thing I can come up with is by the time we hit AGI, Microsoft felt that it would be completely integrated and people would be on board and, and stuff like that. I mean I don’t understand.
Daniel Newman: It was probably more of the Alton side of the deal that was basically, that’s what they’re chasing and you know, right now they have a product that they probably felt Microsoft could commercialize better than they could at least in the beginning. The co-pilots and the way they could layer that into all their products and tools and of course the money. But I’m guessing that that was a negotiated point by opening ally that they think that’s going to be their line and basically they wanted to have some escape valve down the line because as you can see they have no, they don’t really have any leverage in the current relationship that I, that I can identify. There’s a, there’s a tangential story about this Jony Ive deal this trademark dispute on “io” there’s another company called “IYO,” that does like these in-ear pieces that apparently are sort of this pervasive AI technology. And I guess OpenAI has been using and sampling and working with this company and then all of a sudden they, they have now he, they have io and they’re going to make some device. And so I guess there’s some aggressive legal action being taken on this AI hardware device Pat, that you know, is this opportunism by IYO going after io, or do you get the sense the story. At least to me I get the sense that there probably was some real IP sharing. I’m not saying it was stolen or anything but there was some sharing and now you know, God, I mean what’s the coincidence? IYO and io and it’s a wearable, it’s a device and I don’t know. What do you think?
Patrick Moorhead: Well, what I am certain of now is what Johnny I’ve is working on, right. Which is an in the ear, in the ear device. I, you know, I’ve done a bunch of brand research and I’ve you know, created new new product brands and the process you go through, sometimes you’re using the same agencies, right. And they come in and they recommend a bunch of garbage names that nobody has ever heard before. It’s really hard to find a name now and get it and get it trademarked. What I, what I don’t understand though is how both companies can have Trademarks on it. I do think if you know, Jony is truly working on an earpiece like the IYO folks, then he should lose. Right? I think it’s as simple as that. And they need to respect the name. I really think at this point the name doesn’t even matter. But this could turn into. If IYO did share information with OpenAI and exactly what they’re doing and, and you know, Jony pops out with something that, that looks like something like that, this could escalate into from a trademark battle into a, a theft of intellectual property lawsuit which would be super duper ugly.
Daniel Newman: Yeah, Sam is pretty diabolical. I keep using that word but I mean did you see that report that came out about him like at Y Combinator? Like the stories that were like made up and now have been redacted. There’s this really interesting thread on the whole kind of coming up. A lot of people tend to think that the real brain, brain, you know, Ilya S. and the other that have left was really the real meat and brains behind open AI. I don’t know. We’ll see. But when he’s done building the $7 trillion FAB network, maybe he can get back to in ear devices or wearable pins when he’s also going to you know, take out Apple when he’s done taking out Nvidia and TSMC.
Patrick Moorhead: So wasn’t he gonna take out Meta as well, social media and, and Google and everything they do that’s advertising search is over.
Daniel Newman: They’re going to take it over. And then the last one was just recently he said they could do autonomous driving better than Waymo RX.
Patrick Moorhead: So it is amazing that, that, that, that Tesla just went to Transformers, right? They were literally doing if-then statements like, stupid. You know and they, they just moved to machine learning less than a year ago. It’s crazy.
Daniel Newman: It is crazy. Yeah. That was one of those funny takes. I don’t know if you listen to the all in where they were betting who was the best on AI and they talked about integrated stacks and how fully integrated wins. And everyone was on the show is picking Tesla because that quote unquote fully integrated stack. And all I said to myself is I’m pretty sure Tesla doesn’t have a training chip. I’m pretty sure Tesla trains on NVIDIA they use Dojo. They have, they have an in-car chip for inference that they use for sure. But I’m, pretty sure they, that it’s – Anyways, it was just interesting because like where Google actually does have a full stack, fully integrated stack and NVIDIA of course has a fully integrated stack. Tesla has done incredibly well. But like I still think, I still struggle to think that, you know, there’s not some risk with how they basically built the whole thing on PureVision. I just, I don’t, we’ll see. I mean, maybe I was too bullish on LiDAR. It’s certainly expensive.
Patrick Moorhead: Well, the farther, the farther you go out in time, I think the less that you need LiDAR. But I think the radar thing, I don’t know how you deal. You don’t use radar. At least some, something that can see in the pitch black, dark or snow or fog.
Daniel Newman: Yeah, well, we see what these robotaxis, you know, of course I’ve seen some pretty funny Waymo ones driving right into floods and other things. You know, like with all those sensors though it doesn’t stop these things from still making mistakes. But I think it’s like 90 or 95 or 99 less mistake-prone than, than “us guys,” so just remember that.
Patrick Moorhead: Well, you…
Daniel Newman: Well, I’m a terrible driver, but it’s because, you know, I’m like Ricky Bobby.
Patrick Moorhead: I don’t drive anymore. I just walk around downtown or get on planes.
Daniel Newman: I just want to go fast, you know, remember, if you’re not first, you’re last. I don’t know, I heard some nonsense about where you go to team building. Nobody wants to compete with each other. I, I can’t.
Patrick Moorhead: I mean, reality is everybody wants to compete with each other. They want to look good in front of the boss.
Daniel Newman: I can’t imagine just being okay with being okay doesn’t work for me.
Patrick Moorhead: Maybe it’s a generational thing dude.
Daniel Newman: Speaking of being okay with being okay, you know, I wrote a long dissertation on LinkedIn not that long ago about the future and AI and changing my company. And you know, over the last few weeks we’re starting to get some affirmations. One of the companies that have come out there is Amazon that’s basically saying they’re going to see their hiring volumes basically slow or even shrink. And while they’re going to grow and it’s going to be because of AI, there’s not a company that has more kind of continuum of where AI, it’s got physical AI and robots, which they’ve been doing for a long time. You got autonomy and driver and delivery and drone. And then of course now they’re talking about the white collar knowledge workers. It seems to all be coming together and they’re saying it’s going to shrink over time.
Patrick Moorhead: This was my topic, but you’re doing a fine job.
Daniel Newman: I just handed it to you. That was, it, was that too much?
Patrick Moorhead: I’ll say, look, now listen, in terms of the analysis, I think you, you brought out the facts just nice, just nicely. And if you look at CEO Andy Jassy, right? The internal email that he sent out, I think it, it’s essentially, here’s what it’s saying without saying it is we’re going to plunk an amazing amount of money into what we want to be efficiency driving technologies and there is going to be job loss. And it just opens up the Pandora’s box of this discussion. And where I sit on this is jobs will be destroyed, jobs will be created, just like we’ve seen in every generation. I think the one difference here is going to be the speed of change. And you know, even when I look at, at what happened to you know, in e-commerce, commerce, what happened in media, what happened to travel agencies, what happened to stock brokers and, and things like that, I think it’s just going to be accelerated. The one thing that’s going to slow all of this stuff down in terms of these businesses is technical debt. And I think we’ve seen this with every major enterprise that we’ve seen. Even the top 100 largest tech companies have challenges. Why? They have a mainframe, they have minicomputers, they have databases that don’t support objects and the ability to cut across those data and across those systems. But I do think AWS or Amazon as a whole, needs to be one of the quickest companies to do that, if nothing else, to prove to their enterprise customers that they’re good at AI and that they actually know they are customer-zero for cutting edge. We’ve seen Microsoft do this. And while Satya didn’t directly relate job cuts to AI efficiency, you have had – I think the first CEO that came out there and said it out loud was IBM’s Arvind Krishna when he talked about how 40% of his tasks inside of human resources, particularly the, the asking questions part that are inside of a knowledge base and their policy could be, could be answered by basically an LLM chatbot. And I’m not even talking about an agent, right? And I’ll close with saying I do think the biggest areas of job loss will be in areas that we didn’t necessarily see in Internet 1.0 and 2.0. And that’s the sea of cubes that are still there. Customer Service, Accounts Payable, Accounts Receivable, Finance, Marketing. That’s where you’re going to see the big, the big job loss, and we’re going to see new things pop up. Any, you know, entry level workers out here. It’s tough to get a job right now, I hear and I read. And the best thing that you can do is get out there and get really amazing at these tools so you can be doing the work of a middle manager who people are trying to lay off in droves at this point. And, if you’re a super experienced person. Well, first of all, if you’re in the middle, if you’re in middle management, you’re probably in jeopardy. So finding a way to, you have to embrace tools so you can do what your boss does without even having the competency or skillsets to do it. I’m going to leave it there.
Daniel Newman: Yeah, I’m less optimistic. But I love your last piece of advice because I agree the people that learn to use the tools will be the winners. It’s very Pareto. And I think the 20% will always have options. The people that are the high-performers, the incrementals and exponentials, I just think the sort of, you know, the people that just sort of move bits of data from one place to another all day long – I mean, agents are just coming for you. And so you heard Mark Benioff yesterday said something like 30 to 50% of their work now is being done by AI and that it’s allowing people to upskill and augment. And that’s great. I just want to understand what these people that were doing those more mundane jobs are upskilling to? And then what are the people that they’ve now kind of come under and accelerated doing now, you know, and it’s just, it’s not clear to me. And so I don’t want to be the Debbie Downer, but I’m also seeing how productive, you know, these tools can be and how fast they can move. I mean, I can create a thousand pieces of content now while with no work, with no workers. And we used to take, you know, it would take 100 analysts to do, you know, two or three pieces a week to get that done. It’s, it’s amazing. And the content’s pretty good. And every day it gets better. That’s the thing, it just keeps getting better. And it’s like, you know, I, I’ve got a tool now that I can write things in my voice. And it’s like, it’s not perfect, but I, like, read the thing and I’m like, that’s about 80%. It’s better than I could ever outsource a writer already. So it’s crazy how fast things are moving. All right, there’s a few other things that happen. I’m going to just, I’m going to touch on them really quickly. We got to get to the Flip, Pat. You know, there’s the news about Deepseek supporting the China military. And now Deepseek’s delayed. I’m just going to do a quick I told you so. You know, they need a lot more NVIDIA. The Huawei stack is not going to get them where they need to go now. And yes, they’re going to work on that, but I’m, I’m going to stand my ground. And the whole military support thing is just more reason, I think. I just can’t believe we were so bullish on using Deepseek and dropping all our data in. When you have all these American companies stealing your data that at least, you know, are American. I don’t know if you have any big add’s on that one, Pat, but.
Patrick Moorhead: Yeah, I mean, listen, I’ll flip it. I mean, what tech companies don’t help our U.S. Military, right? Every hyperscaler, every neo-cloud. I mean, so, I mean, of course, yeah, this is the I told you so. Yeah, right.
Daniel Newman: I got no beef with that. I just mean, like, if you’re a U.S. though, use a U.S. model. I don’t know, just seems basic to me. And then of course, Microsoft’s AI chip was delayed. Pat, you know, the rumor is that Marvell is getting the design for the next generation. Maya, the first one wasn’t super good. I can see why it’s being delayed. There’s a lot of takes out there that this is a big thing, a big win for NVIDIA. Microsoft’s a huge NVIDIA customer. They’re spending hugely with NVIDIA. They will continue, continue to spend hugely because their cloud customers are going to use NVIDIA. This was an internal workload chip. I think there’s a little bit of. I don’t know how much it changes the economics for the NVIDIA relationship, but it does really go to show and continue to show that while I actually am an XPU believer, I think about 25% of the accelerators will be XPUs. It does go to – this is really a hard problem. This is a really hard problem. And this is why NVIDIA is so doing so well. And this is why AMD is a great opportunity if they can get Rackscale going is because the chip alone is only a small portion of the issue.
Patrick Moorhead: It took Google two or three generations of TPU to really dial it in. And I think that this stuff is hard. If I look at, I think, AWS was in the minority where Inferentia One just hit it. It did very well on a very narrow set of workloads. And it did them. It did them very well. All of Alexa was built – so not built – was run off of Inferentia. So they just hit it. And, and the other, the other thing is how much flexibility you put into the design of your XPU. If you dial it in for efficiency and don’t leave yourself some flexibility to make some changes, or if you put abstraction layers on top of it, you may not get the performance that you’re looking for, the performance per watt. So I don’t see this as a giant setback. I do see this as a setback. You know, Microsoft was talking about all the workloads that they were currently running internally on it. I had, you know, asked a lot of questions about growth right before Ignite. I didn’t get any answers on that. So I pretty much knew, knew what was up here.
Daniel Newman: Yeah, I think that’s a good take and we’ll see how it plays out. I just don’t think it’s going to add a significant number of NVL 72s that are going to be go out. I think Microsoft’s already fine, just about everyone they can get their hands on.
Patrick Moorhead: I mean the most successful XPU company, which is Google, increased as a percentage the amount of GPUs versus XPUs. And, and I think that this says a lot to me.
Daniel Newman: 100 percent. All right, buddy, you want to, you wanna, you know, you wanna. You ready to throw down?
Patrick Moorhead: Let’s throw it down, baby. Let’s do this.
Daniel Newman: I mean, I see you wearing that shirt and you’re flashing those biceps.
Patrick Moorhead: I mean every time you just gotta, like. I didn’t even do arms today. I did freaking cardio.
Daniel Newman: Cardio. What is cardio, that’s just dumb.
Patrick Moorhead: I man, I did zone three. Zone three on the treadmill. 30 minutes.
Daniel Newman: Yeah, I’m not gonna lie. I actually did do sprints today. I did my lift and then I’ve been doing, you know, about six or seven minutes of basically run the straight jog the curves and by the way, sprinting when you’re as old as we are, really hard. I think I thought I was going to pull my hamstring like five different times. Anyways, so Anthropic, among others, is, you know, is basically fighting the fight that they have rights to basically use all the data and information across the web that’s publicly available. You know, and we’re still in this conflicted moment. Pat. Is, you know, is, do these model companies really have the rights to just steal everything and just take it and use it and make a ton of money off of it and screw all the people that created stuff? I think, you know, one of us is gonna have to say yes, one of us is gonna have to say no. So let’s see who says yes. You’re the thief. You’re the one that says just take it. So you, you’re saying just take whatever you want. Steal it. Musicians, writers, authors, bloggers, it’s all free. F you.
Patrick Moorhead: Yeah. So, Dan, I told you I would never hold your age and inexperience against you, but I actually worked at one of the original search engines. In fact, we were the top, top in search over Yahoo at the time. And we’ve already been here before. This has already been litigated, which was “Can you go in and index content, essentially pull that content in, and make derivative works on it in the music industry?” Right? The whole and you would understand this because you know, you were a “DJ guy” in Chicago when you were younger. And you could sample, right? You could sample other artists’ content as long as it was a derivative work. We see this on YouTube today where as long as you are adding value. So for instance, I’m recently, I do game walkthroughs, Death Stranding, Death Stranding 2, and essentially it’s in 4K. A complete pirating of the walkthrough. But this guy is talking through it, and narrating through the game. So, it was of no surprise that the federal judge said, “Hey, using copyrighted books to train its AI Claude, is transforming the content and qualifies under fair use.” So we have the entire Internet Age, including search, music and video, that back this up. So yeah, I’m not, I’m not surprised at all where this landed because it’s essentially the same rules as the Internet.
Daniel Newman: I’m glad that you worked at a failed search startup. I worked at Blockbuster and therefore I am an expert in all things studios, media and Hollywood. It is a derivative versus using all the content and giving no credit issue. In my opinion. The problem, the reason search worked is the derivative product generally led people back to the origin content the origin content gave the producers of the videos, movies, music, written content, the opportunity to make that content fully available and choose whether or not to monetize that. Now what’s happening is these LLMs are able to extract that all the way. They’re able to not just derivative, but basically hack and steal entire content. And if you don’t believe me, just use a tool like Zero GPT to see how much content is actually just purely stolen. When you actually type something up using an AI platform, it’s not original. It’s not. It’s not a meaningful, I’m sorry, it is almost the original content it’s not really a derivative. It’s not changing things. And therefore, people like musicians, artists, researchers, analysts that do really difficult work, get their content basically stolen and used a lot like some of the analyst firms are today, when they just hack a market model and share it out with their own logo on the corner of the page. I guess that’s okay too. But in the end, my opinion is that basically the fair use is a nonsense trait that’s being allowed to go through because we’ve already opened Pandora’s Box and we can’t go back. We need to be able to allow this, because there’s trillions of dollars at stake here. And if someone went back and said all the data you’re training these models on that have basically built the global economy for the next two decades aren’t allowed, you can’t use the Internet, you can’t scrape things, and you can’t steal everybody’s work. We would be erasing a huge secular trend, but in my opinion, it’s still a giant F you to the creators, to the people that put in the work that create unique, differentiated content and give them basically no chance to monetize their efforts. So FU, Judge, FU, Anthropic. You suck. Fair use is all right. All right. My Blockbuster days, buddy.
Patrick Moorhead: No, that was good. That was good. By the way, just for the record, the strategic mistake that Alta Vista made was trying to go after Yahoo and not monetizing search. Although, we even had cost-per-click before Google did. When I worked at the Alta Vista shopping search along with our president, Mike Rubin. I actually worked for Mike.
Daniel Newman: Now part of Six Five.
Patrick Moorhead: Right about that. I know. No, it’s good stuff. You know, the one thing that could.
Daniel Newman: Could you guys have bought Google for like 2 million bucks? Is there like a story like that with Google, with Alta Vista? Like there has been with Blockbuster and Netflix and, you know, all these historic companies that could have?
Patrick Moorhead: I don’t know. But what I do know is, we got Elon Musk started, we bought his company called Zip2, which he took all that money and poured into what was called X, which competed with PayPal, and they merged with the PayPal mafia, which gave him enough money to start SpaceX and Tesla. So that was his big cash injection that got him going. Hey, seriously, one of the things on this content thing that I’m really going to be monitoring closely is if all of the content creators of the new content go bust or can’t monetize it, like the models are going to have nothing to pull in, and I guess it’ll kind of be a free for all of bloggers and things like that. And one side of me says, well, the trust in news sources has plummeted in the last 10 years. So you know, who cares? The second thing I’m thinking through is, “Will AI be smart enough to discern what the truth actually is, and not necessarily need all of this content to derive what it thinks is the truth?”
Daniel Newman: I don’t know. But these things are going crazy. You know, you hear now that like when they fail they’re like saying they’re going to like unplug themselves and commit suicide. And we’re definitely training these beasts to be total locos. We’ll see how this all ends up. Pat, we only got a few minutes left here. It’s been a crazy ripper week in the markets. You know, I guess the tariff thing wasn’t as big of a problem because when we went to war everyone forgot about tariffs. We fought a 12 day war. It seemingly has been resolved for now. The straight of Hormuz was not shut down. Oil keeps flowing. S&P is now ripped to a new high. Names are: NVIDIA hit an all time high this week, Pat. The market is exploding and there was a few, a few earnings and other things this week. So let’s, let’s, let’s, let’s finish up here with some Bulls and Bears. All right Pat, the big earnings of the week. It was Micron. You know, this was one of those moments, you know, I always call them the breadcrumb moments for, you know, because everybody right now cares about what’s going to happen to NVIDIA. So, so Micron had this great run up. Its stock has nearly doubled this year and it’s performing incredibly well basically as this, in this AI, you know, run up is going to require a ton of high bandwidth memory. HBM3 now one of the big concerns is that, you know, there’s some, some rumors out there that HBM4 certification may be delayed and SK is going to get all the, all the opportunity there as Samsung continues to try to get their stuff together. And now Micron’s approval is being delayed. But the overall market, they beat strongly on the top, they beat strongly on the bottom. The guide looked really, really good, but it seemed to be all baked in. Everybody sort of knew this was going to happen. The expectation that things are going to be good. The demise of AI and the bubble bears and all the people that think the infrastructure boom isn’t going to happen and Capex spend was overstated and no one’s going to use this stuff continue to be epically wrong. But at the same time, Pat, you know, there’s some of that feeling of how frothy things are right now is have we hit one of those? Like, you know, is this the top? Our buddy Dan Ives now has his old ETF on AI where over a quarter billion dollars has gone into it in two weeks. I mean clearly, people are, you know, are gambling now. What did President Trump say? No crying in the casino. But the casino right now, it’s like broken. I mean you put a quarter in the slot machine, it throws out a dollar, and you’re seeing companies doubling in not, not months, not years, but days and weeks. Absolutely incredible out there. But yeah, Micron had a good quarter.
Patrick Moorhead: I met with their chief revenue officer yesterday and had a really good conversation with him on that. I wish Micron would say more about what they’re doing because they have so many firsts, right? Whether it’s first with a certain node first with a certain technology, they’re just not telling enough people about it. And they seem to be kind of stuck in this hole of, you know, there was a time where they were cash flow negative and, they kind of pulled back on everything. But I think they have the opportunity now to really get out there and get it. And it’s not just HBM, right? It is, but, and that’s why you have to have a rocket ship into the peaks, right about every five years there’s a boom and bust cycle in memory and storage. That’s just historical, but the key is you have to power out of it. So yeah, I’ve got a lot of questions about Micron. You know, the ability, the investments that they’re making. I mean I love the $200 billion over 20-year investment in the U.S. which I do think gives them a strategic advantage. But I’m wondering from a Capex overall on a cost-basis, how do they compete with Samsung and, and Hynix? But listen, they, they ripped it, they deserved it. They are approved inside of NVIDIA and not everybody is at this point. So let’s go.
Daniel Newman: It’s a great point and by the way, worth mentioning that the HBM and the AI trend is sort of the fuel but there is a whole like cycle around devices that has been an absolute crap show for like multiple years. And if that gets the pickup, which it generally does through these cycles, it could be a kind of a double-triple whammy as we see next generation AI devices, handsets. And then of course, you know, they have businesses across other areas like automotive that continue to get more content. Speaking of automotive, quick note, Pat, I mean Intel, you know, has been eerily quiet. Not really saying much, but it looks like Lip Bu is just doing the work. He’s shutting down the automotive. This is nothing to do with Mobileye, just Intel Automotive. Seems like anything that’s not kind of critical and imperative is starting to get slashed and burned. That’s what he said was going to happen. Not a lot of public testimony going on, not a lot of public storytelling, just quietly cutting heads, cutting costs, cutting business units, and I think trying to get back to some success in those other areas.
Patrick Moorhead: Yeah, the company’s been eerily quiet and I do think there is a middle ground between the Lip Bu Tan regime right now and where Pat Gelsinger was. You know, the fact that, you know, Intel didn’t show up in the Middle East in any of those announcements to me was telling and also, also troubling to me. When Lip Bu came in and even when Pat was in, I was looking in, and I did a lot of broadcast appearances where they asked me, “Hey, what, what could be cut.?” So first off, I don’t like quick peanut butter cuts. You know, Everybody gets a 20% haircut, right? That never works. It demoralizes everybody, particularly if you don’t have the tools or the processes to get new work done. What I do like is cutting off limbs, right? And I had, I thought that Automotive, Photonics, Quantum, and other pet projects that I don’t even know about may have been there. By the way, I’m not calling the former pet project. So photonics is required if you want to do rackscale. Can’t, can’t cut that. quantum, my guess is still on the table. There’s also a Neuromorphic project as well. I don’t know how much they’re spending in there, but they’re not core to the business of client and data center. So that leaves automotive. But I do want to say that, you know, when Jack Weiss laid out the strategy a couple years back, I liked it. Right. You and I met with him multiple times and, and there was a base of x86 customers in. In the cabin. Right. So for the in cabin electronics and people who, you know, could keep the same software and, and move that forward. I liked that. The second thing I liked was that the world moving to chiplets and Intel being a full IDM, they could integrate the intellectual property of an automaker inside of their own custom chiplet. Intel would get the CPU, the GPU, likely the XPU and the networking and the memory controller would get all that IP and they would also be able to manufacture that. So yeah, I like that core value proposition. But you have to cut where you have to cut. And this is the first cut. Final thought is I’m glad we’re seeing some decisions made. Right. This is the first what I view is major decision, aside from some of the reorgs that Lip Bu did in flattening the organization.
Daniel Newman: Yeah, I agree with it. Just get narrow, get focused, get right. Win a big foundry customer, get an AI chip roadmap that’s realistic and competitive. Not after NVIDIA. Go try to win a, you know, win against some of these asics that are struggling. You know that these companies are trying to go custom and win an inference chip with Gaudi 3. We’ve seen that. We know it can be competitive, and of course don’t give up. Try to not cede any more market share in your core markets. I mean those are the three things that seem to be the focus. And so all these other things in my opinion are distraction and just delayed. I mean look, if you invested a dollar like 30 years ago in NVIDIA, you’d have or sorry I didn’t tell you to 50 cents today. So it’s, it’s just unfortunately, you know, great American. It’s rough. It’s a great American story, but very hard to convince anybody to invest right now until you show you can deliver. And so I think he wants to your window in in more different fewer wins, but big wins in a smaller bucket of categories to deliver on the promises that were made. Quietly doing the work. In my opinion, that’s what the company needs right now. Quietly do the work.
Patrick Moorhead: What do you make of them not showing up in in the Middle East deals?
Daniel Newman: I mean it’s a huge miss, but clearly they don’t have the, the juice. I mean, look, companies like Groq, you know, where I’ll admit I’m an, I’m an investor are making mega wins. They’re winning Canada’s inference platform and they’re winning major deals with like the G42s in the Middle East to do inference. I mean, Gaudi should be in those conversations, they’re not and there’s something wrong that we don’t know and hopefully he’s getting to the bottom of that because, you know, the inference quality on, on the stuff we’ve tested on like ivf, I mean, it’s not, it’s not a bad part, but it’s just, they’re not showing up. And so anyways, last note quickly before we run. I mentioned this on the way into this. Bulls and bears. NVIDIA hits all all-time high. So the demise of AI, the demise of NVIDIA, the China, the China restrictions can’t stop NVIDIA, I mean you’re talking about. They just hit almost 150 Pat. They are nearing $4 trillion in market cap. Most valuable company in the world again. You know, when you hear things about, you know, delays on asics, you hear things about Intel’s woes, you hear questions about, you know, whether or not AMD can build the software to scale. We’re seeing progress there, but we hear still rumors and doubts. Can they build rackscale? I mean, does NVIDIA just, is it just, you know, I mean, they’re not up as much as Coreweave, which makes no sense to me, but like, I mean, what an incredible run. Four trillion. I would hit three. I mean, we’re like, how do we get to four? They need another major boon, it appears. They don’t need another major boon. It appears all they need is a little patience. I want to sing like Guns N Roses. What do you think?
Patrick Moorhead: A lot of the risks, right, that were up there. I mean, if you look at the economic risk, the tariff risk that I know you mentioned, that’s off the table. And even though I think that AMD went from a 5% data center GPU play to a 10% GPU data center play, just people seeing the reality of that competitively, in addition to your smart comment about the, the accelerator, the other thing is NVIDIA de risked Grace Blackwell, right in the way that they rolled out the rack, right? They took away some of the different ways and the complexity to roll that out. Right. GB 300 had a different infrastructure versus GB 200. And what they did is they, I got to say they frank and architected it together, but they took that, they took that away. I don’t think they’re getting any credit yet for the potential industrial robotic edge. But it’s a TBD. And by the way, just as easy as as NVIDIA went up, it can go down with some sort of counter argument or some sort of event that, that happens. So anyways, NVIDIA’s rocking it reminds me of the Cisco during the Internet Age, where it could do no wrong and you know, every day trader was making money on it in the world. You just had to you know, swing a cat, you know, throw darts. My mother-in-law was day trading and making a ton of money during those times.
Daniel Newman: It’s actually gone doubled since its lows in just a handful of months from liberation day lows. It’s, it’s doubled and you know, Cisco’s actually back at all-time highs. IBM’s at all-time highs. Like, you know, it is a bit of a weird phenomenon what is going on because like you said, you can sort of throw a dart anywhere. Good news is they have a huge set of opportunities and things like robotics autonomy that aren’t really being baked into the number yet. I’m just glad you know, back in 2022 and it was about 10% of today’s price, I was telling everybody that it was the, going to be the best name to, to probably own. Even though I don’t advise people to buy stocks. I’m just saying, like, I was very confident then. I like to occasionally do a victory lap about that. Well documented, Pat. And by the way, one more thing. There can be more than one winner. AMD can win because if AMD gets from 5% or to 10% that’s a – they can go from where they’re at to 50 billion in AI chips. NVIDIA can do 300 billion. Broadcom can go to 70, 80 billion. Marvell can do 20 or 30 billion. Like I think there’s a lot of people that want this to be zero-sum. It’s not, it’s a dumb take. I wrote about that on Twitter. You can look at my X post about that, right? Quit it with the dumbness people. You can actually like two companies in the same sector, especially in a secular trend that’s this bullish. All right buddy, that’s it, man, it’s Friday. You’ve only gone one in four weeks. So, like, you know, you’re going to start to miss this, you know, with all these off trading we’ve been having lately. So hopefully, you all out there are missing us a little bit, too. But there’s a lot of other content, all the summit content, all the HPE content, AMD event content, Cisco event content, just over the last handful of weeks, there is so much Six Five out there. We love it. And by the way, 100,000 YouTube subscribers. So we’re pretty proud of that. We are, we are up, we are growing, we are going, things are happening. So subscribe, be part of our community, stick with us for all the shows. But for this week, got to say goodbye. Thanks for tuning in.
Author Information
Daniel is the CEO of The Futurum Group. Living his life at the intersection of people and technology, Daniel works with the world’s largest technology brands exploring Digital Transformation and how it is influencing the enterprise.
From the leading edge of AI to global technology policy, Daniel makes the connections between business, people and tech that are required for companies to benefit most from their technology investments. Daniel is a top 5 globally ranked industry analyst and his ideas are regularly cited or shared in television appearances by CNBC, Bloomberg, Wall Street Journal and hundreds of other sites around the world.
A 7x Best-Selling Author including his most recent book “Human/Machine.” Daniel is also a Forbes and MarketWatch (Dow Jones) contributor.
An MBA and Former Graduate Adjunct Faculty, Daniel is an Austin Texas transplant after 40 years in Chicago. His speaking takes him around the world each year as he shares his vision of the role technology will play in our future.
