Meta's Muse Moment, AMD's Trillion-Dollar Milestone, and the Agent-Driven CPU Rally

The handpicked topics for this week are:

1. Meta Connect Turns Muse Into a Cross-Device Agent Platform: Meta used its developer conference to extend Muse, its consumer AI agent, across new Ray-Ban audio glasses, a camera-equipped version, and a palm-sized Muse Charm device. Patrick Moorhead tested Muse himself, booking and updating appointments and pricing airline tickets, and called its computer-use capability the real unlock. Daniel Newman pointed to a download pace faster than ChatGPT's and roughly 700,000 daily users, and argued Meta's bigger opportunity is taking a small cut of the transactions agents complete, since only 2% to 3% of consumers pay for AI today. Newman added that Futurum had raised its 2030 forecast for AI CPU revenue from $110 billion to $250 billion days before the launch, and expects to revise it again. (The Decode)

2. Qualcomm Builds Out the Agentic-Device Stack at Snapdragon Summit: In Maui, Qualcomm moved from one flagship smartphone chip to two and introduced an audio chip that lets earbuds and glasses connect over Wi-Fi without a phone, a move Daniel Newman read as a direct shot at Apple. Newman also flagged a new seven-year patent license with Apple and the acquisition of a robotics software company built around the open-source MoveIt platform. Patrick Moorhead focused on how Qualcomm's core technologies, including HBC stacking and modular designs, scale from wearables to hyperscale data centers, arguing that volume across billions of chips could give Qualcomm a cost advantage despite a projected 1% to 2% data center share in 2029. (The Decode)

3. Opus 5.5, GPT-6 Sol and Luna, and Grok 4.7 Intensify the Model Race: Anthropic, OpenAI, and xAI all shipped new models within days of each other. Patrick Moorhead cited Signal65 research he says shows Anthropic's newest model using roughly three times the output tokens of its top rivals to reach correct answers, which he tied to Anthropic's compute constraints, and noted OpenAI cut Sol pricing by up to 50%. Daniel Newman shared that one Fortune 100 company spending $200 million a year with Anthropic plans to cut that by about 75% by shifting workloads to open-source and on-device models. Newman argued the race is far from settled, with OpenAI, Meta, Google, and xAI all showing momentum, and that the labs that locked in compute early are best positioned to stay on the frontier.
(The Decode)

4. AI Alarmism Meets Policy as Jensen Huang Rejects Extinction Risk: The push to rename AI "superintelligence" and a proposal from Sen. Bernie Sanders to ban superintelligence development kept the policy debate heated, while Jensen Huang said there is zero chance of AI causing human extinction. Daniel Newman warned that viral safety headlines rarely get corrected once they spread. Patrick Moorhead argued the tech industry lost control of the message by emphasizing job loss and danger without making the case for AI's benefits or explaining what losing the race to China would actually mean.
(The Decode)

5. U.S.-China Talks Produce Pageantry but No Chip Agreements: President Xi's U.S. visit brought parades and a high-profile dinner with technology leaders, but Patrick Moorhead noted no agreements on chips or chip exports. He pointed to a two-month extension announced by Treasury Secretary Scott Bessent and farm purchase commitments, with Taiwan at the top of China's agenda. Both hosts noted the most concrete shared outcome was agreement on what to call AI. (The Decode)

6. The Flip: Has Meta Cracked Mainstream AI Wearables Without Full AR?: Patrick Moorhead argued yes, pointing to audio-only Ray-Bans that look like ordinary glasses, Muse's connected tools, and a lighter camera version, after 14 years of failed attempts across the category. Daniel Newman countered that Meta cracked the agent and its distribution to 3 billion users, not the wearable, since Muse demand shows up in the phone's app store and people will still want to review purchases on a screen before committing money. Both hosts said they plan to buy the new glasses. Note: The Flip assigns Patrick Moorhead and Daniel Newman opposing sides of a debate, positions argued are not necessarily their own views. (The Flip)

7. Intel ($INTC) Rallies as Muse Fuels CPU Demand Optimism: Patrick Moorhead explained that Muse assigns each user a virtual machine running on AMD EPYC processors, a CPU-heavy design that lifted Intel, AMD, and Arm. Intel gained between 10% and 15% on the news and is up about 40% for the month, while Arm is up roughly 30%. Daniel Newman noted Intel can currently meet only about half of the demand for its CPUs, and tied much of the enthusiasm to its foundry business, which he said could become worth twice the company's current value as TSMC capacity stays constrained. (Bulls and Bears)

8. AMD ($AMD) Joins the Trillion-Dollar Club: AMD became the latest chipmaker to reach a $1 trillion valuation, with shares near $650 and within reach of a price target Daniel Newman had set for 2028. Newman said the market is treating AMD as having everything to gain and NVIDIA as having everything to lose, while Signal65 testing shows each wins depending on the workload and software stack. Patrick Moorhead, who spent 11 years at AMD, credited Lisa Su and the team but called the valuation rich, noting AMD has competed against NVIDIA's scale-up systems without its own scale-up design until Helios, which has not yet shipped for revenue. (Bulls and Bears)

Watch the full episode at sixfivemedia.com and subscribe to our YouTube channel so you never miss an episode.

Disclaimer: Six Five Media is for information and entertainment purposes only. Over the course of this video, we may discuss companies that are publicly traded, and we may reference their equity share prices. Nothing discussed during this webcast should be considered investment advice or a recommendation to buy or sell any security. We are not investment advisors, and you should not rely on this content as financial advice. Six Five Media collaborates with technology companies and industry leaders to produce research-driven interviews and multimedia programming for enterprise technology audiences.

Daniel Newman:

All right, we are back, everybody. Six, five is, well, we're not live live. We always take a day or so in between when we do the show and we publish it, but we are pretty much live. You and I are live, back with each other. The mustache is live. By the way, there's a poll running on the internet right now, whether this thing is working for me. I might shave it off. I don't know, thinking about maybe taking a break or maybe going a little shorter and doing like a Johnny Depp in Pirates of the Caribbean thing. You know what I'm saying? I don't know. I mean, when you're this good looking, you know I'm kidding, right? There's a lot of things I think about myself. That's not actually one of them. How you doing, buddy? How's things?

‍

Patrick Moorhead: 

Yeah, things are great. Things are all moving, man. I was so jacked last night just kind of thinking about stuff. And I don't know why I was just in the in the absolute groove. Pretty much my brain was working overtime. It must have been the eight and a half hours of sleep I got the night before, but I'm kind of paying for it this morning. Didn't get- How late were you up?

‍

Daniel Newman: 

How late were you up?

‍

Patrick Moorhead: 

Still midnight. Okay. What time did you wake? six. Did you work out? I unfortunately did not because I had a skin thing and they say do not work out for 24 hours.

‍

Daniel Newman: 

You know, Pat, everybody out there if you can't tell, Pat is he's pretty beautiful, but it's not all natural.

‍

Patrick Moorhead: 

I mean, I'm 75 years old. How do you think I look like this, Daniel?

‍

Daniel Newman: 

I agree. Being 78 myself, we are aging gracefully. I don't know.

‍

Patrick Moorhead: 

I look at the intro videos and that's different. Let me know who that other guy is. I don't know who the co-host is.

‍

Daniel Newman: 

Yeah.

‍

Patrick Moorhead: 

Some fat guy.

‍

Daniel Newman: 

Yeah, he was portly. You know, when I grew up, I had to get clothes in the husky section sometimes; you know, I'm big boned. Love that. And I was like, " Was it Cartwright from South Park? Yeah, I'm just big boned. So. And everybody's mom used to tell them that, right? That's how our moms made us feel better, right? It's just your bones are big. That's not actually true. But Pat's InBody says he's doing quite well. But his DEXA wants to argue with his InBody all the time. So who knows? But you look good, buddy. I'm glad that all that health stuff's working out. And it's been a week. I've been halfway to Australia and back. You were out in California at the GSA doing big boy shit, moderating a panel with four of the bigwigs. They weren't president G dinner bigwigs, but they were like the group right below that, right? So, you know, and I did meet with someone that was at that dinner this week. And so I felt pretty neat about that. But so I said, it's like seven degrees of Kevin Bacon, you know, I'm like this close to being invited, right? You know, you you you go for walks and have one-on-one conversations with at least a couple of people that are attending that dinner on a regular basis. We're like we're getting there. Right. We're like, so close to being invited to the President G. Trump dinner, but we just haven't gotten Alice yet.

‍

Patrick Moorhead: 

So close. So close. I mean, I figure I've eaten dinner with Lisa Sue before and she's sitting at at the table with POTUS and G. So I'm kind of like there by osmosis.

‍

Daniel Newman:

 Yeah, well, I guess once your company is worth a trillion, you get to sit at the big table. You do. People there that have multi trillions that we're actually not sitting at the big table.

‍

Patrick Moorhead: 

So that's right. Michael Dell, the second richest man on the planet. I did spot him in the crowd.

‍

Daniel Newman: 

Yeah. And, you know, Sundar was there. And, of course, Sergey and the ones that, you know, the actual ones that own most of the stock for, you know, Zuck. I'm Christiano is there too I saw yes, I met with him this week; sounds cool. Jeff Bezos was there. So poor Andy Jassy didn't get an invite, I guess, when. When your boss is ripped and jacked and actually still owns most of the stock. You don't get the CEO honorary invite.

‍

Patrick Moorhead: 

He was sitting with Cristiano. Yeah.

‍

Daniel Newman: 

Yeah. This is, like I said, we'll get there. Right. I mean, we just need a few more listeners to the pod. That's all that needs to happen. But, you know, another week. And I got to say, like, at least this wasn't a doomer week. Yes. It felt like more optimistic. Um, I was, I was actually waiting in the green room to do it. I was, I was invited on Fox News, like not business Fox News, to debate a Doomer on Will Cain show. It was like an afternoon show on Fox News. And I was on the, in the green room one minute from going on; I canceled it, and they canceled it because, um, they ended up, you know, those students at Ole Miss that passed- I guess they got the attorney general to come on very quickly to talk about rescheduling Kratom and the drug. And so it was like very, but I was super excited to debate a Doomer. It was the former Pentagon AI director and I was going to have a debate online. So I'm hoping that one will come back around. That would have been fun. But otherwise, I mean, it was a positive week. It was like this, this news thing seemed to like completely reinvigorate that AI is useful. You know, we put some new benchmarks out on Signal 65, we got a lot of news to cover. And so I realized we've been rambling on a bit, but that's kind of my thing. So in the weeks I host, you get more of this. When Pat hosts, you're going to get more down to business. It's just It's how we do it. So if you want to get on the cadence of not hearing all my nonsense, just listen next week and then skip a week. And that's OK. I'm OK with that. But we'll talk a little bit about the Muse news. We'll talk about Qualcomm. They had their big Snapdragon summit, big on agentic new models coming out from everybody as they pace the frontier, clearly. And then there's some other, you know, there's a little Doomer talk, what's going on with China. We can get around to that stuff. Not a lot has come from that yet, but we'll give a little perspective on what we've got. And then we'll go on to the flip. Everybody knows the flip. This is where, this is where, no, actually, can you hold on one second? This is my, yeah, just give me one second guys. Of course. Yeah, you got it. All right, and then we'll get to the flip. We've got an interesting one. We're actually not going to debate whether AI is a bubble today. We'll tell you more, but we will debate. But it's nice to get a little positive, right, Pat? Something that seems more optimistic. And then we'll hit a little bit of bulls and bears. No earnings this week. So we'll talk a little bit about markets. But we did have a company cross the trillion-dollar threshold. Can you guess who it is?

‍

Patrick Moorhead:

Hmm.

‍

Daniel Newman: 

If not, you can think about it between now and about 45 minutes from now when we get to that part of the show. But Pat, It is time, no more BSing, no more musings. Let's move and talk about our musings about Muse and the decode. I don't think I could have set it up better than that, Pat. I don't think I could have done better.

‍

Patrick Moorhead: 

That was a 10 out of 10.

‍

Daniel Newman: 

10 out of 10. If this whole analyst thing doesn't work, I could just be a talking head. All right, so let's talk about, let's hit the decode, Pat.

‍

Patrick Moorhead: 

Let's hit Muse. Yeah, so a ton of things positive happened for Meta. And, you know, it is funny, the roller coaster ride about just Meta as a company. Right. They're dead and, you know, their stock plummets because of some decision that Zuck made. And then it comes absolutely roaring back. Right. And I think this is one of those one of the one of those episodes here. A lot of things happened. So first and foremost, the Muse effect came out. So Muse is their version of basically an agentic consumer, agentic app that works pretty much anywhere. And it really just came out of absolute nowhere. And this was on the heels of some of their new models coming out that that they were scoring at least in the top 10 on the benchmarks. But what they announced this week was, at their developer forum, were two devices that just seemed to make sense. And we're going to be debating this on the flip, though. They brought a new version of their glasses with camera, and then they had a version just with a microphone. And if you look at all the musings on X from their leader, his name escapes me right now. He's just an absolute meme factory doing. Van, Algenor, Wayne. Exactly. Yeah, I appreciate that. Yeah, and everybody is excited about this. And as we'll talk about in Bulls and Bears, this was a huge driver of Meta, Intel, AMD, and Arm stock going up as well. And you know, I don't believe what anybody ever says until I actually use it. So I did, I did connect up Muse. And I think the real unlock there is, is, is the CUA, computer use automation, or some people call it computer use agent. which essentially uses a browser to do pretty much anything. I did go in and make an appointment, updated an appointment. I did go in and argue some airline tickets. for a trip to Germany that I have coming up. So I think that these guys are on an absolute roll at this point. It's even making Zuck look super cool in his interview. He recently did an interview. I thought he did a really good job talking about you know, does it make sense to pause or not? And he's obviously taken the tact of, no, we don't need to pause. We just need to watch product liability. On one hand, you know, I thought it was good. On the other hand, it's like, You know, Zuck was saying that social media, you know, wasn't negative for for kids. And he was staring down the double-barreled shotgun of a trillion-dollar fine. He wound up getting out of it and settling with the state state officials in the billions, not not not the trillions. So I I. I am optimistic, because this is the first consumer, I'll call it, agent that's friendly enough that does some real interesting use cases. And Apple's not doing this, Google's not doing this. So I think it's a good, it's a good move. But it's really a wait-and-see.

‍

Daniel Newman: Yeah, I mean, look, fast, what was it, the fastest growth downloads wise in the app store, faster than chat GPT. Market is dying for utility from AI. What I like about this is like we, you know, consumer adoption of paid AI is pretty small, two to three percent. What will people use and potentially pay for? And what I like most about this is he's already thinking about the business model that lives after the expectation of people to pay, right? The dollars transacted are where he's going to make his money. Meaning when someone sends an agent to go book that airline ticket, they're going to take a little piece of it. When someone orders their food, when somebody gets groceries, when somebody, like all the things that people are gonna do are gonna create this massive transaction growth. And then Meta just makes a little on the transaction. The other thing I really like about it is it doesn't cannibalize the ads per se. What it does is it creates an ad stream of enabling offers to be built for the bots instead of built for the human. We always, we know that the agentic utilization versus human utilization is gonna be an order of magnitude. 10 hundred thousand million to one agents to humans. So what if you can actually build these APIs and connectors to give benefits to the user that the agents identify? And then Meta makes a little bit for each transaction. It's a really, really brilliant thing. Meta's, you know, Google, you mentioned Apple, like there's some companies uniquely positioned to try this. I think this is what Apple should have done with Siri. It feels to me like that's the, you know, the miss. I'm not saying they can't get back to it, but it just feels like this was the, the master agent that sits on top of the application layer is the opportunity. And so I also think there's a real possibility you could see Meta come back and build a device around this.

‍

Patrick Moorhead: 

No, they announced it. They announced a charm at their developer conference.

‍

Daniel Newman: 

That's a where that's a right a pervasive where. Yeah. Yeah.

‍

Patrick Moorhead: 

Yeah.

‍

Daniel Newman:

 And I mean, like a handset. Sorry. I just mean like in addition to what. Yeah. Yeah, but yeah, they are already. And by the way, like I talk about the Homo erectus era, like of getting back to standing up vertically again and looking out at the world. I mean, this is a chance for Meta to create devices that enable us to wear and live in the physical world while having all the benefits of agents working in our digital world. It's promising, so it's pretty cool. You know, we'll talk more about like the stock impact later in the show, but like, The market's ripping on this because literally for a couple of million downloads and I think what SimpleWeb, SimilarWeb said something like 700,000 daily users right now, it's already crushing, crushing the compute. So, you know, just another turn. I already sent notes to my team and I said, well, guys, we've got to revise the forecasts again because This is a new use case in the CPU, which, by the way, our 2030 forecast went from 110 billion of CPU for AI to 250 billion in AI for CPU. That was September 18th. So that was right before this announcement. We're probably going to have to double that rev again. Keeps the team busy, keeps everybody honest. All right, speaking of next, what's coming next, Qualcomm. So I had the great pleasure of going out to Hawaii, Maui, to be in the world's absolute worst time zone. Just want to be very clear about that. Five hours back from central US time is the worst time zone on the planet. When I woke up this morning at 4.30 to work out, it was 11.30 at night back there. But I was It's just so disorienting, but you got to admit it's a beautiful place. I've got a great photo of a rainbow, like an end to end rainbow. Like I got perfect in this. It's beautiful. It's just conference rooms, by the way, look the same everywhere. So it doesn't matter where you actually go when you're inside a conference room. It looks like a conference room anywhere else. But yeah, I mean, look, Qualcomm's big message out of Maui was the phone isn't the product anymore. The agent is. And Qualcomm wants to be inside every device that agentic AI touches. So the event starts with the phone. Interestingly, they went from one to two flagship chips. This is the Xtreme Elite and the Elite. This is gonna be really interesting. The Xtreme is a five gigahertz and these phones are shipping before your end. But the big wave is expected early next year. There is a catch. Some of the early benchmarks don't necessarily edge out the A20 Pro. This is going to be something that's going to be discussed. But I don't think this strategy is necessarily an Apple strategy. I think it's a MediaTek strategy. So outside of the two chips, you know, they had a bit of a sleeper announcement, but we know the wearable is hot. They had new earbuds and new audio chip. And this new chip lets the earbuds and the glasses connect right over Wi-Fi with no phone. And so that's interesting because if your wearable doesn't need a phone, the phone stops necessarily having to be a hub. That, in my opinion, was more of a shot at Apple. It's something differentiated for people. And you saw some of the new wearables coming out at MetaConnect, and we'll talk about the event a little bit more later. But they refreshed smaller Surface Pro, Surface Laptop on Snapdragon. The cheap Windows PC is gone. And basically, you know, that, as you know, has everything to do with supply and demand right now. On a broader note, good news, and really nothing to do with Summit, they did announce a new patent license with Qualcomm. It's not a chip deal, so don't get confused. This is not a Qualcomm technology deal. This is a Qualcomm licensing deal, but seven year. And basically, Qualcomm is still getting paid on Apple Modems. So, you know, a lot of people, you know, have criticized that loss on the QTC side, but on the Qualcomm licensing side, they're still making money on every Apple phone. And then lastly, they did acquire a robotics company called Picnic. It's a MoveIt platform. MoveIt is an open-source robotics platform for physical AI in like manufacturing. So don't think humanoids; think manufacturing arms. I continue to think that's a nice extended TAM for them, and it's interesting that they keep picking up open-source software. They did the Arduino thing, they did the modular thing, now they're doing this. So, I don't know, I guess Pat I'll turn it back to you, but what do you think of, you know, I know you weren't there, counted on me to, but you know, extreme, is this a big thing in your mind?

‍

Patrick Moorhead: 

Yeah, so I pulled the transcripts, created a summary, and also was paying attention to Anshul Saag, who was there as well. Ryan was there as well, so paying attention to that. Yeah, what I was struck with was if we look at Qualcomm's overall edge to cloud portfolio, what they did provide more details was on how each of these core technologies work across all the way from the smallest device in your ear or as on a pendant, all the way to the hyperscaler data center. And one of those was HBC, right? There was more details on how HBC shows up in client devices, which HBC was, you know, a brand new stacking technology for, you know, for CPUs and accelerators, but how that scales across Another thing is modular, right? They showed more detail about how modular spans and, you know, how it spans the gamut of all the devices and everything in between. And I think the You know, I think that to me was something that got me super interested in what they were doing. Because here's what this means. People might say, oh, Qualcomm is the small player in data center, which based on their forecast, puts them at about 1% or 2% market share, I think, in 2029. Is that accurate, Daniel? Say that again, just to make sure I got it. Qualcomm 2029 data center share 1 to 2%.

‍

Daniel Newman: 

Yeah, at, you know, they were talking about, what was it, their number 15 billion in 29, fiscal year 29. Yeah, I mean, the number keeps getting bigger, but it's probably about two trillion of data center chips that year. So yeah, about one to two.

‍

Patrick Moorhead: 

Yeah. So people might say, oh, they're small, but what people are missing is they have the ability to leverage all that IP and all that volume and know-how across billions of chips. which could give them, think about this, potentially even a cost advantage when it comes to not only the IP, but also the chiplets. It's one thing to TSMC and say, hey, I want you to make my HBC based CPUs and and accelerators. But, you know, and I can maybe give you 100,000. But now we're talking about millions of this this this technology. So that was one of the things that I was really that struck me. You know, it was good to see the They made an announcement that I don't remember if you brought up or not with liquid context, which is really optimized for hexagon NPU, which makes super efficient AI capable. I think I even saw one or two bit. models running on smartphones, which is, by the way, anybody can do one or two bit, but to get the known good output is a completely different story. So all in all, a good show. I wasn't there. I lived through you and Anshul and Ryan. But I'm going to do a long form write up to really hit the end to end. I would say more strategic piece for investor audience. Awesome.

‍

Daniel Newman: 

Well, good stuff, buddy. Let's hit the next one, which is a another week and just pacing, pacing, pacing. We are You know, I guess pacing is a great word now, the more I think about it, because you can go at a fast pace, a medium pace. So when they say pace, it's like, maybe we were just not reading into what they meant. They might go really, really fast. But we had Opus 55, GBT-6, Soluna, more, more, more. And I think Grok dropped another model. So what's been going on on the model front, Pat?

‍

Patrick Moorhead: 

Yeah, just some great stuff. You know, competition is good. And, you know, even though OpenAI is leading the charge on slowing down and, and Sam, Sam kind of agrees, but, but kind of doesn't. You know, and Elon said he kind of agreed too. Everybody's moving forward here at breakneck speed. And Signal 65, which you and I are co-founders of, led by Ryan Shrout, really brought out some really interesting research. And I urge you to go out to the website and do it all. But it's funny, the one that struck me the most was a post that I did that I did yesterday, which essentially said, if you look at 5.5 Max and Sol Astra Max, Applaud is an absolute shit show, right? It's a lot more expensive. It hallucinates like crazy to get the right output. And the reason is, because it triples the amount of output tokens to get it correct, meaning it's spinning and it's spinning and it's going and keeps trying to get the known good output. So that's something unique that is true by facts. Check it out. But nobody is covering that part. I really think it exemplifies Claude's lack of infrastructure that they have, that they're doing every trick in the book to be able to charge the same and to be able to compete with Sam who bought more infrastructure earlier. And by the way, of note, Sam lowered his prices on Sol6 by up to 50% when Claude made this move. A final take here is on Grok. I had dinner with the CEO this week of a very successful technology company. And we spent half the time, him telling me why Elon is going to run away with everything, right? He'll be the biggest CSP and he'll be the biggest model maker driving the most revenue out there. And it's like, well, wait, what? And he said, listen, not only is he going to make his own chips with TerraFab, but he's going to be producing his own power with some of his proprietary turbine technology that he's working on. You know, I thought about it a bit and You know, Elon is truly the last industrialist in the United States, meaning he actually builds stuff. Right. He gets property. He clears land. He you know, I don't know if you if you saw the video of in South Texas, him, you know, he just bought the land and he's he's getting it ready for SpaceX. And there's just you know, there's 10 front end loaders digging holes. I mean, It reminded me of the videos that we see in China, where they come in and just go full force and are building. So something to think about. Elon running away with it all. I'm sure the investors love to agree with that. But I think there's some merit at least. You can't count him out. That's what that I know. One hundred percent probability.

‍

Daniel Newman: 

Well, look, I mean, you know, Anthropic is using SpaceX, like the other labs are funding the growth of Grok. They're funding GrokBot. They're funding, you know, Elon's doing it all. He's going to try to build terrestrial networks. He's going to, you know, He's trying to, you know, put data centers in space. He's going to obviously he's vertically integrated to do that. Like you got to you got to say like he may be right or wrong, but you can't bet you can't bet against the fact that he, to your point, is truly an industrialist. He's truly putting his money where his mouth is and building stuff. I mean, look, this week he's going to he's going to make my life more convenient. He's going to build a tunnel from Austin to San Antonio. I get to the Spurs games easier. Was that 200 miles an hour taking two, two and a half hour traffic-laden commute down to 20 minutes. And so, yeah, I mean, look, it's super cool. But yeah, the data on the models is really interesting. It just seems like OpenAI is on the comeback trail right now. For a period of time, it just absolutely looked like Claude was going to run away. And economically, and that's what I just keep hearing in all my conversations, is economically, the chickens are coming home to roost. I talked to one Fortune 100 company this week. They told me they spent $200 million a year on token spend with Anthropic, but they said they're working on a plan to reduce that spend by about 75% by offloading to open source and offloading to device, running certain, you know, N-2, 3, 4 workloads closer to the edge. So, you know, the other side of that is that the utilization keeps going up, but I think if these The trade off is always going to be the best intelligence in the easiest distribution of intelligence versus the complexity of standing up infrastructure, the complexity of running models on device, the complexity of figuring out token routing across a plethora of models to certain workloads. But if they can make the intelligence less and less expensive, I think they can scale more and more demand on the APIs. You know, the other side, that's going to be the privacy, the compliance, the zero data retention commitments. That part is increasingly becoming attended to. But I still think that's probably a small subset of the companies that have enough resources to really worry about those things. Most companies are not.

‍

Patrick Moorhead: 

Yeah, so lots going on, lots of updates, lots of new models. Yeah, it's interesting on the Grok side, we haven't tested 4.7. I'm sure Ryan is working on that. But even 4.6 comes in at number six, right? With GPT-Astro, Gemini 3.8, Claudia, you know, 5.5 on top of it, so.

‍

Daniel Newman: 

Yeah, I mean, you know, one of my themes, you know, has been like, people are just wrongly ruling out companies well too soon. You know, we all, we basically anointed Claude the King.

‍

Patrick Moorhead: 

Yeah.

‍

Daniel Newman: 

And we've seen Meta's making a comeback. I mean, Flash 3.8 rated pretty well on a number of different things. And everyone keeps talking about how behind Google is, but like, I don't, I keep saying like just the wrong company to rule out. in so many ways, just so much resource there to build. And we've seen OpenAI come back, now we're seeing Meta come back, we're seeing Grok come back, like this race is not over. And I think you and I have said this, we both can claim fame on this one, but we've said over and over again that he, ye with the compute, ye with the compute.

‍

Patrick Moorhead: 

Yeah.

‍

Daniel Newman: 

So where, I think most companies did better than Anthropic was actually securing compute capacity. Anthropic has more recently been on a buying rampage, but people criticized Sam for going early and making a lot of big commitments. People have criticized the amount that Zuck is spending because he's not even a hyperscaler. People criticize, you know, the overall compute spend that's going on, but you can't train these best of breed models. If you want to be on the frontier, if you want to be on the leading edge of intelligence, You got to spend big on CapEx. Wrapping up this section, Pat, there's a couple of things. We did have a little bit of carryover from last week on the whole alarmism. I think, first of all, President Trump wants to basically get rid of AI as the term. kind of interesting in what you think there. But, you know, he wants to change the name to superintelligence. And he actually just tweeted that him and President Xi, you know, new besties are agreeing. It's super intelligence. And by the way, on the other side of that, I think Bernie, the feel in the burn, put out an act to ban the development of superintelligence and including 20 years in prison for companies that continue working on the development of it. He's up there, I think, on the fritz of some of the dumbest ideas that have been brought to, but he is playing on the fear. And the only thing I'll add to that is, you know, another interesting topic this week, Jensen came out and said 0% chance of extinction, zero. Zuck said something similar, basically zero as long as humans, act right, which just means it's not zero, because there are humans that will not act right. So there was sort of that, and then the whole Jacob Coxon thing that sort of started the whole Doomerism movement, it looks like the proof has come to fruition now that he's scared everyone and no one's going to read any of these details, but it was all a big PR play. Well planned, well orchestrated, working with a firm, that leans into PR doomerism and basically advises doomers on getting their doomer content out. So be careful what you read out there, folks. There's a lot of nonsense and bullshit, a lot of misleading. And because most of us hear the first headline and never hear the retribution, we continue to believe inaccurate information in a post-fact era. And so there's a lot of post-factuals here that are coming to the surface.

‍

Patrick Moorhead: 

Yeah, that was a that was an excellent breakdown. Jensen's really been investing a ton, doing every circuit imaginable, right? I think 60 Minutes or CBS was Sunday. And, you know, he did a New York Times podcast that, you know, the commenters seemed like all a bunch of doomers and oh, this guy makes GPUs and It was interesting to see how everybody is talking on both sides. Uh, the 20 years in prison thing. I mean, this just is an absolute travesty and it's funny. Uh, unfortunately Greg says are, by the way, he used to call himself Gregorio. Uh, but he's not even, um, he's literally not even Spanish. Um, but he just made up the name, uh, when he ran for office here. of the D. S. A. As well. D. S. A. Member and I do Sanders in the same room t times leads me to believe for socialism in the future. In the end, I think this is a money grab, right? It's unfortunate that it's become so political at this point. The industry as a whole, you know, at my dinners this week at the GSA, we did talk a lot about Um, you know, not only what needs to be done in the future, but just how, how much the technology industry lost the message, right? Uh, and weren't able to counterbalance. And in fact, the high tech industry. perpetuated the first myth that there was going to be massive job loss. And then, you know, these multiple hits on it. And then this whole AI being dangerous without, you know, without effectively correlating the benefits of AI. So, Tech industry made its bed. Now it's got to sleep in it. And hopefully it doesn't end up suffocating it in the end. We've also done a piss poor job of explaining, well, what are the ramifications if we, quote unquote, lose the race to China? right? First of all, we n what is the finish line o finish lines look like? A that race to China. It's something I've been trying to do personally, and I think the leaders that I talked with, the GSA, are going to try to do a better job at getting the positive vibes out.

‍

Daniel Newman: 

You know, one of the things that is shocking to me, because I can play back in my head, is just the number of what I would consider well-plugged-in, very senior-level execs that just absolutely bit on the coxswain story hook, line, and sinker. Yeah. And, you know, it was just interesting, like, that people have so much faith in kind of what comes out in news and media without necessarily stress testing it. And then, you know, like I said, the post-fact issue is just that we never walk back anything. You know, like you and I see it because we're like in the 1% of people looking at Twitter all day or X all day and seeing the stories materialize and then mature. But like most people saw the story, they saw the headline, that was it. Like AI is going to kill us. 10% chance the AI is going to kill us. The fact is, how irresponsible is it that the media literally ran with a headline that said there's a 10% chance that this technology, the one that's booking our airline flights and helping us find our Uber rides, and right now is writing a bunch of slop grenades that we're all passing as strategic work, is going to become so powerful that it's going to end humanity. I mean, yeah, and humanity, dude, that's like apocalyptic. And we were like spreading that 10 percent is not negligible. This is not like, you know, that we have a one in one millionth chance of an asteroid ending the world like that could happen. I get it. But we don't report that 10 percent. I mean, gosh, I get that it makes good TV, but. You know, like I said, we just scare the shit out of people that have no ability to discern for themselves and even understand how any of this stuff works, believing that, you know, that an agent is going to get the access to the nuclear football and just start, you know, you know, just bombing the shit out of major populaces. And, you know, we're going to we're going to all die from a radioactive wave anyway. All right. Rant over my dad's good rant, though. Yeah, thanks. So last topic, I want to be getting on it. China, we kind of even hit on it in the last one, but like we haven't heard much of what's come of it besides superintelligence. But President Xi was here. Big dinner. Lots of our friends were were were there. We were not. That's my biggest complaint so far. But anything on that that you want to you want to share to our friends here?

‍

Patrick Moorhead: 

Yeah. So there's a lot of pageantry, a lot of parades and a lot of great stuff like that. But there's absolutely nothing that had been agreed to or even inferred that give us any better visibility into technology. Bessant did announce a two month extension of the tariffs. And I know that the G and his company were pressing for the United States not to get involved in Taiwan. That was seemed to be number one on their on their agenda. I think there was a farm guarantees, farm purchases that who knows this might help help for the midterms. But at this point, I think there is one day left no agreement on chips, chip exports, and really no agreements on anything else other than what we should call AI, which you noticed, and I noticed you pumped out an X post while you were on the show.

‍

Daniel Newman:

Deep thoughts with Dan Newman. My average tweet is like seven words. I think hard before I post. I always enjoy when people are like, there's not a lot of substance. It's like, dude, you're not paying me. Like, don't, don't fucking follow me. If you don't like my, my shit posts, like it's not my job to give you context. Anyways, I'm very ranty today. I should work on that.

‍

Patrick Moorhead: 

Yeah. Your brain's working overtime today or something.

‍

Daniel Newman: 

Yeah, I'm very ranty. It's very whiny, though. I don't like that. Sorry, everybody. I'll try to tweet hard. I'll try to tweet more thoughtful for you, for those of you that don't like my short shitposty tweets. I just don't have time. I'm usually walking to the bathroom trying to get a full thought out. Just a lot. All right, buddy. This is it. This is the big moment. We're going to move on from the decode here. We're going to have a little debate. And it's not going to be about whether AI is going to kill everyone, and it's not going to be whether or not AI is a bubble. So we're going to make a, you know, we're going to do something a little bit more positive today. We're going to talk about Metamuse. And, you know, we are going to basically look at whether or not Meta and this Muse agent is going to take wearables from the doldrums of Reality Labs that has shed billions of dollars in trust in Mark Zuckerberg all the way back to being king again. So, Pat. Let's see who's gonna argue for whether or not Meta has cracked the mainstream with AI wearables and Zuck is back, baby. Skinny Pat's gonna say it's four, Fat Dan's not going. All right, so Skinny Pat, beat me up, let's go.

‍

Patrick Moorhead: 

So I have been, watching the wearables market for forever, probably when you were in high school. Recent history, Google Glass kind of kicked it off. That was literally 14 years ago. We've been working on this problem for 14 years. Apple brought out a stinker. Multiple companies tried to get involved. Many of them went out of business. Two companies that just kept leaning into it, and it's funny, Meta, the name, right? Remember, it used to be called Facebook. And then it was renamed Meta. And then everybody thought this was going to be the time. And Suck was literally investing tens of billions of R&D dollars that was deriving absolutely zero value out of the company. And people pretty much thought, OK, well, this sector might be dead. And then we saw a couple of startups, one of the names was Rabbit, of these tiny little devices that were AI enabled, and they seemed to make sense, right? They were audio only, and you talk to it, and it gives you this amazing result. And unfortunately for those early wearables, their response was absolutely atrocious. They were slow. The AI was bad. So a lot of sordid history. And the final historical note or context setting I'll do here is that OpenAI bought Johnny Ives' business and then hired, I think, 250 Apple engineers to create an AI wearable. So now we enter Meta, again, who's been hitting this for years. And by the way, their Silicon partner is Qualcomm who's been making like investments to finally crack it. And I think they absolutely have cracked it. So first of all it was always the they have to look like normal glasses. And if you look at the brand new Ray-Bans with audio only, they literally look like the glasses that I'm wearing right now. And I think that's a huge piece. The second thing is utility, right? There has to be incremental utility. And when you match those audio meta Ray-Bans with Muse and the capability that it can do, it really is the unlock. You can literally ask it to do anything that you would the standard Muse just sitting there. All the tools are connected. By the way, ChatGPT just added tool support into its audio. version of its agent. So, and with audio, it delivers, again, useful functions without having the visuals, right? And the visual version with the camera, they're about twice the size, and they weigh about a deck of cards. So I think the wearability was one of the biggest unlocks here. And I think the little charm, I always thought those things were silly. This is the Muse charm. It's a little, it looks like a silver dollar that is kind of the audio only. You don't have to wear it. 5G enabled gives you Muse capability pretty much anywhere that you want to go. And if I look at the utility, what it does, primarily backed by how much better Muse is than the predecessor, Lama, it absolutely has a chance. So I believe that this meta has cracked the mainstream AI formula without needing the full AR.

‍

Daniel Newman: 

Well, Pat, it's good to dream. And actually, I'm going to agree with you, but I'm going to I'm going to beat you on technicality here, because I always do have to win. Meta did crack the agent. It just didn't crack the wearable. And every hard number on AI demand is going to indicate that all the demand for Muse comes from the phone. And I love Muse. I mean, I'm playing with it. I think it's an incredible demand signal. We're gonna need more CPUs, but look at where it showed up. It showed up in the app store. iPhone. The best evidence Meta has that people want AI beyond the phone is a download chart on a phone. Cracking the formula means we need to change habits. And what we have is a launch. We have a keychain that ships in the future, which we don't even know what people are going to adopt because we all thought that people were going to wear that spaceship that Apple put on their head. And look, only you and like three other people I know bought one and none of you are using it. So, you know, As a whole, like, yeah, the demand's real. We've got 700,000 people or so daily using, again, but on their phones. The app that people actually use the glasses for is not actually this, it's for audio. So sure, he put an interesting demo of what could happen in the future if glasses did work the right way, but people aren't gonna use it that way. They're wearing the glasses mostly just to listen to music in their own little headspace. And Charm, I mean, really, it isn't even done yet. We don't even know. It's just an idea right now. It's going to ship sometime in the future. So same with Crack the Formula, saying that people are all in on this wearable thing. Look, they showed an audio glass. They showed a camera glass. they're showing a possible display glass, the VR thing, people aren't going to be doing this there. It's a portfolio of bets. It's not a formula. And finally, let's be realistic. When people actually use agents to start doing things in their bank accounts, start spending real money, are they going to do it without a trusted platform in which they can look and review before they say, go ahead? Sure, there's the you know, the futurists, the ones that were wearing the glasses in the shower 20 years ago. But the people today, they're going to use it. They're going to wear it. They're going to talk to it. But when it comes time to basically say time to buy, they're going to pull out this little thing in their pocket. They're going to look at a screen and they're going to press a button. That's how people actually use technology. They're not going to say, hey, pay my mortgage or go sign me up for a new credit card and book a vacation without seeing what they just committed to. first. So look, Meta, they built something cool. You and I 100% agree. But they didn't crack anything on a formula for wearables. They cracked distribution, 3 billion users, and they're getting a tiny, tiny fraction of them to rapidly sign up for a really cool product. But if you want to say they've cracked the wearable and they're heading into the future, give me a call when people start leaving their phones at home.

‍

Patrick Moorhead: 

That's it. You're done? So you're going to buy one? Probably. I am, too.

‍

Daniel Newman:

I'm going to buy the pendant, but I'm definitely getting the new specs.

‍

Patrick Moorhead: 

Yeah.

‍

Daniel Newman: 

Plus, I need to cover up my lack of eyebrows. So I think a pair of specs could create an image of me having eyebrows.

‍

Patrick Moorhead: 

You should just start painting them in. I try. I forget.

‍

Daniel Newman: 

I don't know why they're so thin. My thyroid's fine. I'm not sure. Anyway, we'll have to dig deeper into that. I just can't grow hair above my eyelids. I can't grow any hair above my eyelids. All right. That's the flip. I think we both agree the chances of the whole wearable thing is better with agents, right? I think we would both agree. But I would also say that there's a lot to prove before we can say that the form factor is evolving from people doing most things on their phones to doing most things on some other device. That was the technicality in which I had to play, because otherwise your debate would have been too good to win. Light, weak, and bulls and bears, but there's still some stuff to talk about. So let's head over and talk a little markets.

‍

Daniel Newman: 

All right, Pat, we've talked a lot about muse, but what was the big impact implication of muse?

‍

Daniel Newman:

 I mean, Intel was ripping. It felt at least a little bit tied to that, no?

‍

Patrick Moorhead: 

Yeah, you know, the Muse data came out and of how many signups, we talked about this before, but what people did once they realized that there were so much CPU involved in the processing of all of these agents, and it's funny, you do the double click on how Muse actually works, they're actually giving a VM per person, and it happens to be sitting on AMD EPYC processors, which I thought was pretty cool. And Alexander Wang, I keep forgetting his name. He actually came from he he was famous for an open claw fork. Right. And so essentially and I pinged a couple very more technical people than me on this on this thesis says this is essentially open claw in the cloud. OK. and which requires a lot of CPU processing. So that got all of the CPU numbers up, right? Intel, AMD, as well as ARM. I mean, ARM is absolutely screaming. It's up 30% this month. I pity the fools who sold that too early. And Intel, by the way, Intel for the month is up 40%. But what happened is they went up between 10 and 15% just based on this pop. And it strikes me as yet another proof point for CPU and the requirement for CPUs. And I think you're gonna have to upgrade your forecast based on this.

‍

Daniel Newman:

Um, yeah, I'm in a hundred percent. Look, it was really the whole category ripped. Um, Intel. of course helped lead the charge. I think Intel is moving on some other things because again, this isn't just their own CPUs, which I think a lot of people would argue their actual CPUs are not as competitive as they need to be. And there's work to be done on the Intel designed CPUs, but they right now are selling everything they can build. And I think what they've said is right, what they can only meet about 50% of the demand that's even there. So that gives some really promise to the Intel on the design side, but I think a lot of the enthusiasm about Intel has more to do with the foundry side. Like, I mean, can you imagine had they sparked that up and sold that off or spun that off for parts? And I mean, you and I, you know, we did not dismiss that possibility, but we did fight I think harder than most that breaking that up and spinning that off would be a detriment long term to the company. And I mean, look right now, even though like most of the external customers are still very secret, like no one really knows, like people are starting to put the pieces together of which external customers are going to be building on Intel. And Intel's foundry business could easily become worth twice what the company itself is worth right now. So even still, like there is so much upside here. And I think people are putting that together because TSMC can only grow so fast.

‍

Patrick Moorhead: 

Yes, that's right.

‍

Daniel Newman: 

And so where else does growth come from? And that's also very bullish for Samsung. So we need more capacity, more foundry, and Intel is going to deliver a chunk of it. And we also know we want American-made chips. And Intel is the only company that can do that for the leading edge nodes. So that's a lot in my mind of what the Intel enthusiasm is coming from. But Pat, I mean, I know we can't not finish the show talking about the newest entrant to the trillion dollar club. Yeah. I actually think it's important that we start this off by talking about what your stock would be worth now. Do we have to? No. I mean, like it wasn't that long ago when we had this conversation, it was like worth half a trillion. I mean, the divergence in this one versus NVIDIA has been crazy. And I posted something about this because, you know, we had a target at about $800 a share in like 2028. They might hit that in 26. I mean, they're like 650 a share now. And it's just for AMD, it's just running rampant. I mean, that multiple expansion is crazy. It's probably two to three X now NVIDIA's forward. The way I'm positioning this and look, you know, I like AMD. We've been very bullish about AMD's prospects, but literally the way I feel the market is treating this is that AMD has everything to gain and NVIDIA has everything to lose. And there's some truth in that, like that NVIDIA has all the share. And then of course, AMD has a bigger CPU business right now, but then on the GPU side, and you know, we're doing a lot of testing there, Pat, you know, and it's been really interesting because what we are finding out, I think maybe the simplest way to explain it is in certain specific cases, both are really good. And when people are like, one is better than the other, and the answer really keeps coming back to, well, it depends what you're doing. It depends what software you're running, what stack you're running, which hardware you're running it on, which model you want, which outcome you're looking for. And in some cases, the Blackwells are better. In some cases, the MIs are better. It isn't cut and dry, but overall, Pat, I mean, what an amazing milestone trillion dollar business. And you remember back when the stock was a dollar and the market cap was probably in the single digit billions by that point.

‍

Patrick Moorhead: 

Yeah, I keep just stewing on because not all about me, but it's kind of all about me.

‍

Daniel Newman: 

Probably been a 500x probably from three.

‍

Patrick Moorhead: 

Yeah, let me let me put in numbers. So I left nine figures on the table, based on based on the valuation. But Daniel, I never- You're going to make it other ways, buddy. I never would have met you. I probably wouldn't. I probably would have jumped off the 39th floor of my condo. I had 11 years at AMD, three years, we really did run up the score on on Intel. And I made some good, I made some good money while I was there. But I think I was laying people off every 30 days. And for like, two or three years, it was not a fun, not a fun environment. But I am infinitely happier. Now. I don't know if that's worth nine figures. But I will. We'll see. No, see, let me just some matters here. I liked your I liked your description of Venice and I think people are. finally sinking in that the six derivatives of that is going to give AMD a leg up where people are really concerned about competition from ARM and Intel. By the way, they should. Right. But AMD, I think it's, you know, AMD and ARM have significant market share in the hyperscalers right now. And on the ARM side, Microsoft hasn't even really taken off. at this point, it has really taken off primarily at AWS and then and then Google. I think the other thing that's sinking in, Daniel, and, you know, I like to keep it simple, which is AMD did not have a scale up design until Helios and they still haven't shipped. Helios is for revenue yet. So this entire time AMD has been competing with NVIDIA scale up system, Blackwell with a scale out system. And you can't connect as many GPUs together at the high performance that you need to make a difference. I do think the valuation is super rich, and it's similar to Intel. It's banking on a lot of future goodness out there that might not even be in the valuation spreadsheets of the analysts. But right now, they're on an absolute tear, and Lisa Su and the team deserve a ton of kudos.

‍

Daniel Newman:

 Yeah, like I said, the sentiment is the sentiment. the valuation is value and the value is what people are willing to pay. People are willing to pay a lot to own AMD stock right now and give a lot of credit to them. I think that the positives are people are, the vision's been well understood, the diversification of its portfolio, the quality of its CPUs, the potential of its scale up. I do think that people are very comfortable with execution risk there. Whereas like on the Nvidia side, people are, they have less execution risk and people are treating it like it's higher risk. It's an interesting kind of juxtaposition that it's in right now, but a lot of times, you know, we've all seen how fast you can go from one trillion to five, but I think a lot of people don't see the potential for Nvidia to go from five to 25, whereas they probably see the potential for AMD to go to one to five. And I think that they're betting on the doubles and all those doubles. You would have had four to 500 doubles from where you exited. I just wanted to say that again so you can stew a little more. But yeah, I'm glad you screwed that up and we got to hang out and become friends and we'll find a different way to do a nine figure win for you. How about that?

‍

Patrick Moorhead: 

Yeah. By the way, the other script I want to remind you on is during the internet days, I joined the wrong search engine. I joined AltaVista. I did have the opportunity to interview with Google when I was at AltaVista. I did make $32,000 during .com. It's pretty good, pretty good.

‍

Daniel Newman: 

Did you hold the stock all the way to the bottom and then sell it when it was worthless, pretty much? I mean, that's my strategy, dude. You know that. That's why you are one of the best product advisors on the planet, but nobody's turning to you for stock picks, right?

‍

Patrick Moorhead: 

Definitely, definitely not.

‍

Daniel Newman:

 All right, that was a fun show. you know, make sure I drink my baby's blood every morning so that I have all kinds of energy for the show. Appreciate everybody tuning in. Thanks for being part of Six Five. Pat, we done it again. And was this episode like 847? Three, two, one episode. Three, two, one. This is episode 321. Thanks, everybody, for joining. See you all later.