Claudeforce, AWS's 2 Million GPU Bet, and the Earnings Week That Buried the SaaSpocalypse

The handpicked topics for this week are:

1. Claudeforce Turns Salesforce Into Anthropic's Enterprise Front End. Salesforce and Anthropic launched Claudeforce, positioning Claude as the interface across roughly 27 Salesforce services while Anthropic supplies the underlying AI engine. Marc Benioff and Dario Amodei appeared together on CNBC to make the case for the partnership, and both stocks rallied on the news. Moorhead flags one open question: how Anthropic protects Salesforce customer data without collecting the usage traces that AI systems typically retain. (The Decode)
2. AWS Adds 2 Million NVIDIA GPUs on Top of Its GTC Commitment. AWS committed to another 2 million NVIDIA GPUs, layered onto the 1 million it pledged at GTC five months earlier, alongside its own Trainium and Graviton silicon build-out. Moorhead estimates the deal at 6 to 7 gigawatts and $80 billion to $120 billion in NVIDIA revenue, and reads the pairing of NVIDIA's Vera CPU with Graviton as evidence that agentic workloads need capabilities Amazon's own silicon doesn't yet cover. The commitment reinforces Moorhead's argument that wafer, packaging, and memory supply set the ceiling on AI infrastructure buildout, regardless of how many custom silicon projects come online. (The Decode)
3. Hot Chips 2026 Draws Mainstream Attention as Custom and Merchant Silicon Both Scale. What was once an academic gathering turned into a press and social media event, with OpenAI's Jalapeño inference chip drawing the most attention for its bandwidth-heavy first-generation performance. IBM and Arm detailed a joint development agreement enabling IBM Z mainframes to run Arm code natively at sub-nanosecond switching latency, and AMD showed a full Helios rack while Arm walked through its AGI chip architecture. Moorhead notes that Jalapeño almost certainly relied on Synopsys or Cadence EDA tools rather than in-house design tooling. (The Decode)
4. The NVIDIA-Hugging Face Acquisition Rumor Raises the Stakes on Model Distribution. Reports from The Information, Reuters, and Bloomberg point to a roughly $12.9 billion deal between NVIDIA and Hugging Face, though neither company has confirmed it. Moorhead and Newman question whether NVIDIA would treat Hugging Face as a neutral, GitHub-style repository to keep the open source community on side, or use it as a route into inference services without building out its own datacenter business directly. Newman raises the regulatory exposure of a chipmaker owning the leading open model distribution layer. (The Decode)
5. The Flip: Dario's CNBC Appearance and Claudeforce Put Anthropic's SaaS Intentions to the Test. In this simulated debate, Daniel makes the case for Dario Amodei, pointing to Claudeforce's integration with Salesforce's identity graph, field-level permissions, and audit trail as evidence that Anthropic wants to operate as the intelligence layer sitting on top of enterprise systems of record. Patrick makes the case against this, framing the CNBC appearance as a Trojan horse and citing Amodei's past comments about a small number of AI companies eventually controlling the market as evidence that owning the UI, and the pricing power that comes with it, remains the actual goal. (The Flip)
6. NVIDIA Posts a Quadruple Beat and Commits to a $700 Billion Revenue Target. NVIDIA reported $96.22 billion in quarterly revenue, with $89 billion from data center; it guided Q3 total revenue toward $108 billion and pointed to a $700 billion annualized revenue target Moorhead calls increasingly credible. The company's new AI Clouds, Industrial, and Enterprise reporting category grew 138%, outpacing the roughly 100% growth of the rest of the data center business and easing the concentration risk bears have flagged. Goldman Sachs, Morgan Stanley, Bernstein, and Raymond James all raised price targets on the print. (Bulls and Bears)
7. Salesforce Raises Guidance as Agentforce ARR Grows 240%. Salesforce raised full-year guidance to $46.1 billion to $46.4 billion, with Agentforce ARR reaching $1.5 billion on 240% growth and non-GAAP EPS of $5.90. Moorhead points to premium SKU bookings more than doubling quarter over quarter and half of AI bookings coming from existing customer expansion as the durability signal Agentforce needed. Newman reads the results, paired with the CNBC appearance, as the market correcting its overreaction to the SaaSpocalypse narrative. (Bulls and Bears)
8. Synopsys Beats Across the Board Despite Investor Confusion Over IP Revenue. Synopsys reported $2.48 billion in revenue, up 42%, with $711 million from Ansys, and raised guidance, with operating expense control around the Ansys integration paying off ahead of an expected 2027 revenue lift from the combined businesses. The stock still declined on investor confusion over IP segment reporting differences between FactSet and LSEG data. Newman highlights the company's unit-based royalty model and its early visibility into custom AI chip demand through both its EDA and Ansys simulation businesses. (Bulls and Bears)
9. HP Beats on Revenue and Earnings But the Market Wants an Edge AI Story. HP reported $15.68 billion in revenue, up 12.5%, and EPS of $0.83, both ahead of consensus, with strong personal systems growth as supply constraints give the company pricing power on premium devices. Moorhead reads the sell-off as a margin trust discount tied to tariffs and lingering uncertainty over the company's interim CEO search, with device demand holding up. Newman is looking for HP to show how it monetizes distributed AI and on-device token economics, from lower-cost workstations up through devices like the $100,000 DGX Station. (Bulls and Bears)
10. Everpure Grows Revenue 38% and Raises Guidance on a Second Hyperscaler Win. Everpure, formerly Pure Storage, grew revenue 38% to $1.19 billion, beat EPS at $0.70 versus $0.58 expected, and raised full-year guidance by more than $500 million on a second top-five hyperscaler design win landing in fiscal 2028. Newman points to eight straight quarters of accelerating revenue growth and expanding gross margin dollars even as pricing holds steady. The stock fell roughly 15% despite the beat, which Moorhead and Newman attribute to elevated investor expectations. (Bulls and Bears)
11. Marvell Meets Expectations as the Market Waits for Google Deal Detail. Marvell posted data center revenue up 46% to $2.17 billion, total revenue up 37% to a record $2.739 billion, and Q3 guidance of $3.15 billion, essentially matching estimates, with its full-year outlook raised to roughly $18 billion. Shares fell more than 10% on investor appetite for a guidance raise tied to the Google custom silicon deal, which Moorhead expects Marvell to detail further at its October Financial Analyst Day. Newman frames the quarter as steady execution on socket wins, with the stock up 187% year to date, without the guidance drama some investors wanted. (Bulls and Bears)

Watch the full video 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:

Oh, hey, hey, hey, everybody. We are back.

‍

Daniel Newman: 

It is Friday and it is the Six Five Podcast. Pat, I don't know what episode it is. It's like 3526 or 317. But it's felt like, you know what? Doing this pod with you. What are we, seven years in now? It's felt like five minutes. 

‍

Patrick Moorhead:

Right. 

‍

Daniel Newman: 

Underwater.

‍

Patrick Moorhead: 

Underwater. Sorry. 

‍

Daniel new man

Good morning, Pat. 

‍

Patrick Moorhead: 

I was focusing on, uh, popping another, uh, gummy creatine gummy.

‍

Daniel Newman: 

So on the story of creating gummy, I have a picture just to prove I'm not, I popped open a new container of my, my crush creating gummies or the beetle in there. Fucking ruined it for me. Absolutely. Just put the lid back on, put it on the shelf.

‍

Patrick Moorhead: 

I'm like, I'm going to find a new creatine today to get that out and get it out there, you know, with your. pump community, right? Oh, yeah. Because I'm the only one in the fitness game here. Exactly. Yeah. So I forget the brand. It's a pure creeper.

‍

Daniel Newman: 

You're creeper. And I actually think those are really good. I'm just saying I'm going to find a brand because even if there's this is just an anomaly, one in a billion, there's just something about opening it up and seeing a beetle inside your shoe, your edible routine and just think, I don't think I want to eat that. So. It's been a weird week, man. You know, yesterday I didn't have any water. So, you know, I had to not shower for a couple of days. It was pretty gross. But I mean, these AI data centers are really the problem here in Lakeway. Exactly. Using all the water. Oh, wait, no, it's not. It's the irrigation systems of all the A-holes in the 115 degree weather that water every day when you're only supposed to water like twice a week. Yeah, let's not take any self responsibility here. Let's go ahead, join the Chinese coordinated bots and all AI data centers. And by the way, I'm really glad this is also important that Bernie Sanders made the time AI 100. It's great all the things he's doing and adding, you know, to the AI community, his moratoriums and his desire to immediately stop the progress of our economic engine. I mean, Pat, how did we you know, this goes right there with the CBC today, the Canadian broadcasts giving new parameters on what we can and can't say about the terrible events of September 11th. Okay, so I've had my news diatribe in the morning, but this is not a podcast to talk about big political issues, except when it is. But we do have a lot going on this week. Before I break down the list of everything, everybody's gonna wanna know, how's Pat doing? Why is the mustache gone? Weak sauce. Did you get a lift in? Were you accused of being a firefighter? What happened? Why are we not mustache bros anymore?

‍

Patrick Moorhead: 

Yeah, sorry to let you down. Two things going on. I won't give the percentages of which drove my decision. But the first thing is just started itching like crazy. I was pretty uncomfortable and also looked at a picture of my grandfather and literally we looked exactly the same. And I looked at your stash and kind of how you kept it up and mine was just ugly. And then the final reason, which may or may not be the real reason, is someone in the family very, very close to just every time I'd walk in the room, there would be a comment. So the harassment may have gotten to me.

‍

Daniel Newman: 

That's a tough one. That's a tough one. Yeah, I've been accused so far of this becoming my personality. Like I was at the grocery store the other day and I was like high-fiving random dudes with mustaches. Like, at least I was like, is this going to become your whole personality now? I'm like, well, this is working out pretty much. And I guess AI and rocket ships. So, you know, we're very simple creatures, you know, feed us, be nice to us. That's all we really need, right? Yeah. Thanks. All right. So anyways, we have a great show this week. Thanks for giving us a little insight into your mustache meltdown over there, Morehead. At least we had it for that one moment in time together.

‍

Patrick Moorhead: 

That was a special comments on that. By the way, even Ben Baharan texted me and asked me, you know, you love the stash.

‍

Daniel Newman: 

Yeah, you look great. I actually I don't know what the itching to figure that out. But I thought it actually was, it was quite becoming of you made you look younger, you look younger. But yeah, I mean, look, what a busy week. It was the Super Bowl of earnings. Once again, NVIDIA, it was a big week for software. Some smart analysts have been saying for a while that the SaaSpocalypse was stupid. And now we're starting to maybe get some validation. There's hot chips this week. The jalapeno came out during hot chips. That was pretty clever. Wasn't it? Yeah. You know, NVIDIA is rumored, according to the information. Great source, by the way, one of those that tend to truly trust high quality journalism. I also like the National Enquirer. It's another source I go to often that NVIDIA is doing a deal with Hugging Face. I mean, would they even buy a company at this point or would they just license something? And then, yeah, so, and then we've got Google. What happened with Google? We're gonna possibly get to, they had some updates to their, their offerings and I think they went vertical. I think that's the topic there. If we have time, we'll get to that one. And then of course, we'll do our regular thing. We got the flip, we'll hit the earnings hard. And there was quite a bit of earnings news this week. But again, all of it rolled around NVIDIA. And I would be remiss to not mention there was the best summit of the year, the summit that turned SaaS around, turned it into a growth market again. I got a text from somebody that it was our session with Benioff that actually turned the SaaS narrative. The investment community was very busy covering. Almost every one of our sessions got written up by various financial outlets, which was very cool to see that. We had a great headliners with Mark Benioff, with Matt Murphy, with Sridhar Ramaswamy, with Gary Dickerson from Applied Materials. And then of course, a litany of other CEOs and top executives from across the tech space. And the beauty for all you out there is if you missed it live, Everything's available on demand. That's the good thing about the 6.5 Summit. We crushed it, Pat. We absolutely, for the seventh time, crushed it. All right. All right. So enough set up, enough rolling into this great slate of topics.

‍

Let's go ahead and hit the decode. All right, Pat. What can I say? Pat force. I mean, Claude force. We have entered the air. I mean, I set this up a few times, right? Dario and Benioff on CNBC together. What? What happened? I thought Dario said it's over. I thought he said it's over like six months ago.

‍

Patrick Moorhead: 

Yeah, it was interesting. I was on CNBC. I came in right after Mark Benioff and Dario Amadei got on there. So I had a good chance to comment here. But net-net, this is just a reinforcement of the headless capabilities that are being integrated that Mark Benioff discussed. Quite frankly, if you will go and watch the interview, he didn't say what he was going to announce, but I can certainly connect the dots between what he said and what they announced. And essentially, AI is the UI, which is essentially using Cloud as a front end to all of the, I think it's 27 different services that Salesforce has. And then it's using Anthropic as the go to AI engine. So I think it made a huge difference. And, you know, if you want to connect the dots between, uh, cloud force and, and, and earnings, you know, they had a, like a 20% pop, uh, out there, which, you know, you said in the, in the run up and, and we're gonna, you know, hit this hit salesforce earnings. So we don't need to go into it, but what, what I think, um, So I'm positive on it, right? I think it was good for Salesforce and it was good for Anthropic. And I think essentially because the companies are doing what they do best, right? Salesforce provides the tools in a way that users want. and adds the anthropic technology goodness while protecting the customer's data. And that's a dot, dot, dot. I need more information on it. I don't know how you cloak data without the traces. And there's a lot of stuff that you can do with traces to be able to figure out what people are needing. And don't confuse that with data retention, right? Data retention just says you're going to keep it. It doesn't mean you can't learn from it and stow it away like a squirrel with nuts in its cheek, right? So I need to do the double click on that. And that to me is the only dangling chad that needs to be viewed.

‍

Daniel Newman: 

Yeah, I mean, look, these kinds of partnerships have been discussed, but it was all drowned out over the last couple of quarters. I mean, ServiceNow has partnerships with these. Oracle has been doing stuff with these. For Salesforce, this isn't the first time. Salesforce was doing stuff with OpenAI as well. But I do think there's a cementing here. And I think the big thing was that Dario coming out and saying, you know, it just was a change. It was a change of tune. You know, and I guess the question is, is like, did he capture enough market cap or valuation since he's not public yet that he no longer needs to say these things? You know, remember last week we had the conversation about like AI needs new spokespeople. And like, is there like a PR person in his ear? Like you should do this partnership and you should maybe come out and realize that it doesn't benefit for Anthropic to be the only company on the planet that's gonna survive other than the US government or whatever nonsense he wrote in his manifesto. Like, because the idea here is really just putting into practice that you have probabilistic and you have deterministic. Models are probabilistic. And for anyone out there that doesn't really understand what that is, ask a model the same thing twice. you will get two answers that'll be similar, but they will not be identical. And that's because, again, what is required from an enterprise software system is gonna be deterministic, meaning it needs to be able to extrapolate exact data and deliver an invoice or a response or a workflow in a way that is completely within the guardrails, the governance of a system. So you put those two things together, though, a probabilistic system with a deterministic system, and then you put all the rules and rails around it, it can be really powerful. People are not going to necessarily vibe code their solutions, and they're not going to point all of their proprietary data, going back to the CARP comments about their alpha, they're not going to point all their data to a to a model. I just don't think that happens. And I think that's why this industry survives. But your point about headless is like, look, it's also a rerating. It's also means the industry's value. It probably isn't gonna be the UI anymore. It probably isn't gonna be the destination it once was. So people will be working in APIs. They'll be working in chat interfaces that'll look like cloud or Slack or teams. and they will be connected to all of these different systems. So this is the move. And I think it's good that Benioff recognizes that he needs to do this now. And he's aggressively moving in this direction. And we'll talk more about the business itself and the earnings. So that's the readout there. I think it's a good step. And by the way, like I said, we'll talk more about earnings, but these stocks absolutely ripped on this news. So, I mean, now they're almost back to not being down for the year. I mean, that's the crazy thing. They're ripping just to get back even close to where they were before. But I think we can agree the hyperbole was fun, but it was overdone. For sure. So, all right. So AWS, Pat, next topic, committed to another 2 million NVIDIA GPUs. This comes as part of the announcement slate from NVIDIA's earnings. This comes following OpenAI's dropping a bomb, dropping the hot sauce, dropping the jalapeno the day before, which I don't think we have this in our slate, but we'll talk about it at hot chips a little bit. We can talk about it a little bit in hot chips, but like, you know, where again, our sort of zero-sum brains cannot comprehend the idea that maybe OpenAI is gonna build a good first-generation ASIC for AI that's going to do. very well for inference. And at the same time that NVIDIA can still grow. This is like, but AWS, I think kind of came out and said something that the market needs to appreciate. That despite Tranium's progress and despite the improvement and the demand and the fact that it's sold out for years to come and that they're improving with each generation, the outsized demand for AI is requiring the hyperscalers to build as much capacity as possible and 2 million GPUs. And then of course, that is not including that some of them will be Vera with mean CPUs, and then they're going to be deploying some of the Vera CPUs separately. So they're actually going to deploy a mountain of additional NVIDIA, despite the fact they're doing their own thing. And this is, like I said, one of the most frustrating things for me, but the net of it is, is this is where we're at based on our infrastructure forecasts. This is what it's going to look like for the next few years. And this is a reason that NVIDIA, you know, upped their guidance, which again, we'll talk more about later, from 45 to 70% is because no matter how many custom AI projects are going on, There's only so many wafers. There's only so much packaging. There's only so much memory available. And all these companies to deliver the demand for AI in their clouds need to stand up all the infrastructure that they can. So it's a big win for all parties, but it just continues to be the era of abundance. And I'm going to say that a hundred more times because I'm a quote guy. I'm a talking points guy. Abundance is here to stay.

‍

Patrick Moorhead: 

Now those were good points and let me fill in some of the some of the cracks here. So first of all, this was 2 million on top of the million that they committed to at GTC. five months ago. And I do think that's significant. Also, my back of the envelope deal size on this. So let's say it's six to seven gigawatts, right 28,000 NBL 72 racks, probably 80 to $120 billion in in overall revenue for for NVIDIA. And, you know, we're going to talk about earnings later, but that trillion dollar annualized number, and I recognize this is more like 18 months, is looking more and more real. more and more real every day. And it also fills in the multi-year visibility that I keep hearing from the bears that NVIDIA doesn't have. The other call out for me was Vera next to Graviton. I mean, Graviton is religion as it should be at AWS. And the only thing I can come up with is that architecturally, Vera does some very unique things related to agentic workflows, whether it's sandboxes, whether it's handoffs, whether it's RL, regardless, it must be doing something that's very, very unique. And NVIDIA has been saying this for a very long time. And this, to me, I'm going to take as proof positive that this is the reality. A couple of things that I had initially thought that people would get wrong, that they didn't end up getting wrong. So this whole idea of NVHBM, it's not NVIDIA getting into the memory business. It is them doing a base die. that is built with the memory vendors. All I can conclude is that this is very much an optimization for either yield or an optimization for performance and heat.

‍

Daniel Newman: 

Do you not see it as any sort of reducing the dependence on any one of the three? So standardizing so that they have, go ahead.

‍

Patrick Moorhead: 

No, I'm going to answer no, just because, and even though I know that TSMC can do base dyes, which they will continue, but the architecture, what TSMC doesn't do is the overall architecture. So I'm not walking away with that yet, but if I see some evidence that this is, But I believe 95% this is done in alignment with the memory vendors here. Got it. Okay. Yeah. And it's not what I would call custom HBM, but it's, it, it strikes me as custom HBM for a specific NVIDIA, um, uh, design and moving logic to the base dive free is like 25% more area for XPU for compute about 30% more memory bandwidth. Um, So I think it's more about performance capabilities than the availability. Finally, I think the whole discussion on NVLink, the other religion at AWS is EFA, Elastic Fabric Adapter, which is their version of scale up and I believe scale across. that the company has. There was an NVLink announcement made a few months ago, but this is proof positive that it's gone from announcement to reality. So EFA and NVLink will work together. And I think that's a big deal. given, you know, I mean, I've spent so much time around the Silicon people at AWS. And I just think it shows the value of the integrated capability of NVIDIA.

‍

Daniel Newman: 

I don't know. I heard Jalapeno is going to take them out. It's over. Yeah. Yeah. He is taking them out. And fortunately, he's not a problem because their middle managers are distracted. Yeah. overusing resources is what I've heard.

‍

Patrick Moorhead: 

Yeah. Yeah. I've heard that too. I mean, why would it, why would AMD want to come up with something unique and novel like Adam? So I don't know. That's. I don't know, we should become the six five tabloid. Yeah, I've been I've been I've been floating that around. And I do know so much. I think it'd be super spicy. What does your stuff up though, just to see how the market reacts? Well, I'll be part of it. Right. And then time it. So I would either crash a stock or accelerate a stock that that maybe I'm consulting on or maybe is in my holdings. And so now we're talking Pat. Especially when it's your book, that's when it's really smart to do it. All right, good, good. Inverse moorhead trade is the best. I do like that trade.

‍

Daniel Newman: 

I do want your buy signals, so I know when to sell it. All right, hot chips, hot chips. Anybody will get your hot chips, Pat. Look, we had, all kinds of stuff, a whole day on memory. NVIDIA announced a bunch of stuff and showed the same slide every time. By the way, they were right. IBM arm in the mainframe. I mean, we go all the way from the cutting edge to the mainframe and as the mainframe maybe is cutting edge. But anyways, what was your kind of big takeaways from Hotchips?

‍

Patrick Moorhead: 

Yeah, so just kind of an overall piece, overall thing. Hot chips literally used to be just a geek fest. Nobody, press didn't care about it aside from E.E. Times. People weren't talking about it on social media. It was very much an academic, you know, hence it being held at Stanford, right? But this absolutely turned into a social media, fin twit and press extravaganza, right? And I think if I look at the kind of the headliners, which we've been saying forever is you don't pick a winner between custom silicon and NVIDIA and AMD, right? It just reinforces the strength of different architectures and also ecosystems. So yeah, I mean, the big news was jalapeno, which again, I'm really interested to see. By the way, I'm really impressed with what they put up on the board. It looks like they used a custom version of inference max. And they put up some amazing numbers. It's a bandwidth machine, which probably means that teraflops per square millimeter isn't great, but that's okay, right? As long as it's delivering what you need it to deliver. I've never seen a V1 piece of silicon ever come out at scale at performance with the yield that you need. And quite frankly, if, if jalapeno can do this, it will get unicorn status in, in my mind. Let me go through for the other announcements. IBM announced a collaboration, a joint development agreement with IBM. IBM and ARM came together. And essentially, IBM Z chips will run ARM code natively. And I think the one big takeaway here was it does it at sub-nanosecond latency. And that was always the question. They never answered the question, but they did. So if ZOS and ARM or a Z-compatible software comes in. It's less than a one nanosecond switching penalty, which by the way, is a lot less than what goes on in the actual VM. So AMD highlighted full Helios rack, gave more details of that. And then ARM detailed its AGI chip, right? It went through the entire architecture, which I thought was good. So I feel like I need to go next year. Either myself, I know you had Brendan there. Yep. To cover it. But again, Infrastructure is king right now, and every double click in the entire infrastructure stack is now being scrutinized and covered in details that is awesome.

‍

Daniel Newman: 

Yeah, I don't have a lot to add here, having not been there, but it's definitely You know, this is the state of the industry, right? Supercomputing was a nerd fest. Now supercomputing is a huge event. Hot chips was a nerd fest. Now hot chips is a really mainstream event. You know, we are in the era where silicon semiconductors eat the world. And so it's a good thing some people said this well before it happened and other people got around to understanding it after it happened. But look, what we're seeing here, like I said, is just that there's… Both things are true at the same time. These general purpose architectures are scaling at an incredible rate and a lot of work is being done to optimize these systems. And at the same time, you know, these custom chip projects are also scaling really, really quickly. And what you said about first generation, I mentioned about this earlier, it's like, look, you know, clearly we're having breakthroughs where we can do better in less terms than history has presented, right? And, you know, is this how AI is driving design? Is this AI inside of EDA and tools? Is this, you know?

‍

Patrick Moorhead: 

Glad you brought that up, the EDA part. Yeah, sorry.

‍

Daniel Newman: 

No, no, it's great. I mean, it's, you know, we are clearly seeing a, a step function innovation that we can move a lot faster. It's a combination of know-how, partnership, and of course, understanding the use case is better too. There's a lot of things happening all at one time, but Hotships seems to be a great place to get an earful.

‍

Patrick Moorhead: 

Yeah. Can you add something on that? Yeah. I'm going to be really clear, 100 percent chance that Open AI with jalapeno these guys used synopsis and cadence Or or one of the other To do this.

‍

Daniel Newman: 

So the magic they didn't come they had synopsis possibly both and This is not, this was not them uniquely having the skill set in house.

‍

Patrick Moorhead: 

Exactly. I think they did move the ball forward in, in certain elements of the Silicon design chain. But, um, again, a hundred percent chance it's using, they're using EDA tools off the shelf.

‍

Daniel Newman: 

And there you go. All right. Next topic, Pat. Um, NVIDIA going to acquire Hugging Face. Man, I had some rabbit journalists trying to get me to say something.

‍

Patrick Moorhead: 

Supposedly.

‍

Daniel Newman: 

Right, exactly. But the point is, is like, clearly, first off, rumor source was, I believe, there were several, but I don't know the original. I think Information, then Reuters, then Bloomberg. This one's really interesting. You know, both parties have denied to comment on it. And so basically, what we're really gonna have a conversation is, philosophically, if this deal was to happen, is it a good deal? Why would NVIDIA want to buy a hugging face? Could this deal even get done if it was done? I think there's a ton of potential regulatory risk in this deal. It's like the arm deal now at the model layer instead of at the ISO layer. But basically, TLDR, $13 billion deal, basically, why would they do this deal? I mean, you know, it's the open source model. They basically own the, it's like GitHub for open source models. They own it, you know, and I also think that would be hard, but they have compute, they have the system layer, they have built the networking layer, the software layer increasingly. Although there was a debate this week, I think you posted about assembly code back in Vogue. Are we at the end of the era of frameworks? Going back, baby. Well, it's like the mini and then the mainframe, right? The mainframe and the mini. And now you'd add the model distribution layer. So basically, everybody would be rolling through NVIDIA now to get access to the models. Again, it would be like Microsoft when it bought GitHub. That's kind of the simplest analogy that I can use. I genuinely think it'd be an awesome move by Nvidia. And I just, I don't see how it gets done. I don't see how this seems to be, you know, I mean, Hugging Face is like the Switzerland of open source and Nvidia is a, you know, of course promotes open source. So I'm not taking that away. They've been very positive on open source and open weight, but at the same time, like, I just can't imagine the industry wouldn't be an upheaval about this move. But the other question is, would Hugging Face maintain the same value to the market if it was owned by any of the major chip players? Step back, because the thing is, Pat, is at $12 billion, AMD could afford this. Intel could afford this. I mean, Salesforce could afford this. CoreWeave could do this move. The point is, at that price tag, I just don't know if the market would allow NVIDIA to control this particular. set of economics. I don't know. What do you think?

‍

Patrick Moorhead: 

Yeah. So there's two ways I can handle this. I use a Qualcomm's modular as an example where, you know, day of, by the way, and this is just a rumor. So we haven't had a deal day announcement. Right. It's a philosophical discussion. Yeah. Yeah. So, so coming out and making it clear that we're going to do the investments to make this an industry standard. And then what that would, uh if that would drive me in the direction that says nvidia really wants to be the center of of developers like first stop right now the first stop is github and then you manage your models through uh through hugging face right it's a it's a combination deal uh hugging face very could very well could um um Eliminate the need for for github and you know it'd be hard to share stuff back and forth But maybe they adopt they adopt the same schema and also Hugging face got in they have an inference service and my thought is this could potentially be a stealth inference service that Nvidia could invest in and and where they get into the datacenter business without actually getting into the datacenter business. It's a unique way of them potentially going all vertical if they needed to pull that plug. The NeoClouds were one way to lessen the reliance on the hyperscalers. It could be a backdoor to do this. I don't think they accomplish much. By just having a repository for their own models I think in video benefits from all open models and this is why I think they're going to try and take the the Switzerland approach it's going to be really important though you know there's been a lot of talk about. You know when nvidia says that they open source stuff and then at least if you. You know believe what's on semi analysis and what they discuss on their nvidia won't take the upstream. fixes their stuff on some of the stuff that they've open sourced. So I am optimistic that open source gets a massive boost with the money that NVIDIA can put in this. I also think when it comes to inference service, it could only be NVIDIA. Right. I don't know why you would want to, by the way, and this is everybody's getting into the inference service game, you know, render fireworks, folks, folks like that. But I, I'm interested to see if it's actually true.

‍

Daniel Newman: 

Yeah. Yeah. We'll have to wait and see. And yes, I agree. I just meant philosophically this day. There's a lot of, of the, This is kind of a control layer. And it's interesting to see if anyone will be allowed to own it. But again, I also, you know, we didn't mention this Pat, but you know, there's a lot of ways nowadays that deals get done that, you know, maybe they'll just hire all the people. and not actually acquire it. I mean, you know, we've seen a lot of those kinds of moves over the last few years because nobody wants to deal with the regulatory environment. So I'm kind of being facetious by the way, because like there's a lot of IP here to actually acquire, but like, you know, how they did the Grok deal or how Microsoft did the inflection deal. There's lots of different ways to acquire IP without actually buying companies now, apparently. All right, man, we got to get to the flip. We had some more topics, but unfortunately, we have a lot of earnings to cover. I have to smoke you out first here with another grand debate. Today, we're going to talk We're gonna talk about, well, let's double down on this SaaSpocalypse thing. We're gonna talk about Dario Amadei, and is he telling the truth? So basically we're gonna debate, Dario is telling the truth, Anthropic is not trying to kill SaaS, and CloudForce is the proof. Let's see in the flip who's for, Dario Amodei, the AI SaaS good guy.

‍

Patrick Moorhead: 

Yeah, the Doomer himself, Daniel, gets the four.

‍

Daniel Newman: 

Oh, come on. I'm not the Doomer. I am the bastion of rocket ships and high fives. Pat, I mean, look, the SaaSpocalypse crowd has a problem this week. Dario basically went on CNBC, sat next to Mark Benioff and Jim Cramer and said flat out, we are not trying to destroy anyone. Hours later, Anthropic shipped Clawed Force, which puts Salesforce data, workloads, governance, all inside of Clawed. If you're plotting to kill the system of record, no matter what he's said in the past, I don't think you'd marry it on national television. You marry it because you need to. That's the whole case. Anthropic wants to be the intelligence layer on top of the enterprise data. It doesn't want to own the data. It doesn't want the permissions or the compliance liability. That's all the FUD. That's the story the market wants you to believe. Maybe it was a story for a valuation capture, but Dario understands all those things. That's Salesforce's job and CloudForce is Anthropic saying the quiet part out loud. You know, and the long story short is, you know, one is what they shipped. Watch it. You know, CloudForce binds Cloud to Salesforce's identity graph, field level permissions, audit trail. That's not a displacement. That's a codependency. It's a partnership. What coming up would be trying to replace Salesforce would build its own CRM object model. Anthropic did the opposite. It basically made Salesforce's object model the trusted substrate. You'd understand that as a chip guy. BitCloud reasons over. In every serious enterprise, that's doing AI is routing their AI through a system of record. You heard me talk about earlier, what is probabilistic versus what is deterministic. Cloud Force allows the probabilistic value of an LLM to work strategically with the deterministic value of your CRM or your various systems of record. So basically, Frontier Labs have models. SAS has 150,000 enterprise logos. And we've heard endless, you know, partnering with PE firms, partnering with VCs, partnering with middle markets, partnering with consulting firms. But in the end, who has all the business relationships? Anthropic knows what it's doing here. It knows that Salesforce, and then it will do this with ServiceNow, and then it will do this with SAP, and then it will do this with Oracle, because they have hundreds of thousands of the largest enterprise on the planet that have all their critical data. This is what Palantir CEO Alex Karp has been saying. The alpha sits in the enterprise data, the proprietary data that sits inside these systems of record. Why try to rebuild it from scratch and ask your customers to vibe code and build these things, share all their data with unknown, a company that's never done these systems and build it all from scratch. Anthropic's revenue comes from the enterprise. Enterprises are going to buy through the trusted vendors. Killing SaaS is basically killing their own channel. And the last thing is, and I said this before, the hard ROI era of AI really is in the intelligence. It is the thin layer that sits between the model and the infrastructure. And I think Claude, OpenAI, and every one of these other frontier model companies have come to realize that that data sits inside of enterprise software. It was a dream that Pat was going to vibe code his CRM system over the weekend, deploy it, utilize it, manage it, govern it. and secure it at scale with the ability to actually see all the endpoint API data in a way that would be usable and valuable. TLDRs, nobody ever left their enterprise software. It was a dream. It was a great ploy. It was a tabloid headline that created massive and unfortunate selling that hurt a lot of companies while Dario captured a lot of valuation. But Dario has come all the way around. He sees the story now. It's favorable for him to align as SAS isn't dead. It never was.

‍

Patrick Moorhead: 

So I can't believe you said all of that with a straight face, Daniel, but, uh, that was, uh, that was good. It was a difficult one, but no, listen, I mean, I'm clearly against this. Uh, you know, Dario's public reassurances is exactly what a smart displacement play looks like. And I don't know if any of you are familiar with the Trojan horse. Um, uh, Two countries were battling each other, and one gave a gift of a horse that came in that happened as a peace token. And when they rolled it in at night, the soldiers came out and basically burned down the entire city and country. So this very well could be this. And think about this. So quote, unquote, Salesforce and Claude means sellers never opened Salesforce. And I think it's pretty clear that the textbook UI displacement, right? CRM becomes headless. And quite frankly, once the UI moves, the C pricing moves. I think the other idea here is that this is really just a setup for, at some point, for Dario to buy Salesforce. right, and integrate them in, it probably would have been a better time to run the SaaSpocalypse play and keep doing that to have a lower valuation. But this very well, I mean, Dario himself did not deny that he said that there will be one private company, only one, and it would be us in the future. And that by default means that all the SaaS companies and their capabilities would be enveloped in. I think it's too hard of a poll to vibe code all of SaaS. and kind of ridiculous, but the ability to acquire them at some point, particularly when you get the UI. I mean, Apple's a great example, right? Apple owns the UI, and then they farm out the rest of the back end to everybody else. But Apple's margins are higher than pretty much everybody else, and they extract a toll uh in the in the app store um and i think that's you know you might be like oh that's the consumer version i think it applies here too once you have the the ui everything else can can move over that you can very quickly um you know exchange a sales force for I don't know, a service now when it comes to IT, ITSM, things like that, it just would become a whole lot easier and you really care less about what's on the backend. Yeah, I would be very careful with this one because you don't just turn within a year and go from, or actually three months because Dario put SAS on notice months ago and I'd love to get their pitch deck. that shows exactly how they're going to value capture. I think the number was $30 trillion is the entire TAM, which by the way, is larger than the GDP of the United States in 2025, which is pretty crazy. Ambitious. Very ambitious. $30 trillion TAM. You need to start updating your slides, Daniel, with that. I know it's crazy.

‍

Daniel Newman: 

Alright, you done? I win. Alright, thanks for the thanks for playing. You know, in some ways I think we're both right. I think there is a bit of an admission that's gone on that they can't, but I don't think that means he doesn't intend to. You know, I think any company, like you said, I think that was a really good point. Companies that are in the process of transitioning and cannibalizing their current channels and customers are often, they will often lean into their channels and customers in that period. And they will sort of eat at the bottom while they're pretending to be supportive at the top until they get enough market or clarity on their strategy. And then eventually they release the bait. So nobody wants to, you know, very few companies blow up their channel in the process of launching new things. All right, Pat, listen, we've got some earnings to cover. You get to cover the interesting ones. I get to cover the even more interesting ones. Let's get to bulls and bubble bears. All right, Pat, did the bubble bears get a serving of humble pie when NVIDIA announced its numbers? Is core what happened?

‍

Patrick Moorhead: 

Yeah. Short answer is, is, is yes. And I think the biggest, uh, bear case on NVIDIA has been you, you with all. So there's a couple, right? The competition from XPUs and AMD will, will dramatically take a market share away from you. That's one. I think the second one is, is that, um, the actual size of the market or the TAM can't grow enough because we don't have enough silicon, sorry, we don't have enough wafers and WFE and TSMC capacity to pull this through and we're not going to have enough power, right? But I think what what NVIDIA did is is they essentially committed to 700 billion dollar revenue target. OK. And that that was the net net. That was the thing that drove the stock. If you looked at the early numbers or the early after hours reaction, you know, again, even though they had a quadruple beat. Um, the, you know, the last seven quarters they had quadruple beats, their stock went down and it ripped because the line of sight and the confidence to, again, that $700 billion, uh, revenue number, uh, became a lot, a lot clearer. So I, um, They're already cranking out $100 billion quarters, which, gosh, I heard some nonsense of, that's lower than Google. Well, they do different things, my friend. But it's literally $100 billion. That was the Q, 108 billion for a Q3 guide. That's absolutely nuts. So I think NVIDIA actually has done hasn't done as well as the socks. And I think that's the big comparison that people were picking at. And it's finally getting the respect that I think both of you think it deserves. They missed the whisper number, by the way, on the revenue, which was more like 108, billion for this quarter. Or maybe it was 104. Well, I think that was also why the initial reaction was sort of negative, right? Exactly. That's exactly right. I also think another proof point that I'm glad they brought out, which was splitting data center number between hyperscalers and non-hyperscalers, which they call ACIE, AI Clouds, Industrial and Enterprise, growing at a much faster clip than the rest of the business, right? So, 138% compared to 100%. And I think that's important. That's been another drag, that concentration risk, and- That group was 40 billion, Pat.

‍

Daniel Newman: 

I mean, that's a big number.

‍

Patrick Moorhead: 

Yeah, no, absolutely. Still smaller than hyperscalers, but growing at a much faster clip. And if you look at the combination of the neoclouds, saw a piece of research that I don't agree with particularly that said that 70% of enterprises are about to run some workloads on Neo clouds. But it does, I think I might believe 40 or 50%. The whole governance piece with on-prem AI, that is getting pretty popular and, and the industrial edge. So all in all a tremendous earnings and it, you know, pretty much everybody raised Goldman, Morgan, Bernstein, Raymond James. You know, Raymond James was the biggest, the 515, right? Pretty much everybody.

‍

Daniel Newman: 

We've been all about these big numbers lately. They had some just ridiculous number on SpaceX recently too. Yeah, they've got it. They got a perma-bull analyst person. It's it's it's all.

‍

Patrick Moorhead: 

Yeah.

‍

Patrick Moorhead: 

Yeah. But everybody has a permabear permable somewhere in there. But NVIDIA has embarrassed so many people that there's no perma-Bull.

‍

Daniel Newman: I just I just don't know what the the bear case really is. at least in the next 18 months. Yeah. Like you're basically coming out, filing a regulatory, you know, report that comes out and says, we're going to grow 70%. By the way, Yeah. You know, I, I had to talk to Colette. She really reiterated about like the 70% is like, it's the floor in the sense of this is just what supply allows. if the supply was bigger, if they were actually able to make more, they could sell more. And I mean, that's the wild thing is like this, you know, it was a 45% growth now at a 70% growth that actually would could be 100 or 150. Like they won't even say how big it is. But the fact is, is the demand is just massive. It breaks all the zero summing of ASIC projects and custom set like, it's, we just can't build enough. And so it's become, I get it. Like I've become like a freaking, you know, broken record, but like, I just don't know what people need to see to eventually be like, okay, I get it. Like this is happening. Just not in my backyard, Pat. Exactly. I've got almond farms to stand up, you know, I got to use all the water for my almonds. But otherwise, you hit most of it on the head. The AC number though is almost 50% of the day center number. So it is diversifying. And one thing about the NeoClouds pad is maybe 70% if you're counting capacity deals where the hyperscalers are using NeoClouds. It's the only thing I could think of that would make that realistic to me. Well, to me, just accounting rules would say it's who's ever writing a check. No, I get it. I'm just saying like, but research reports talk about 20 engineers working on a project. They, you know, there's not all the quality is not always good.

‍

Patrick Moorhead: That's good.

‍

Daniel Newman: 

By the way, you did see the one I sent you yesterday though, the, the, the bank of America report. Did you see the headline? It was worth laughing at, but you know, the headline of a, of a bank America, Oh, this was an old one. It was funny from 2019. Dick's added, BJ's removed. A real headline from a real bank. So funny things happen in research. All right, let's hit Salesforce. Pat, we talked a lot about this. Salesforce, Had 11% growth above the high end of guidance, 590, 103% year on year earnings. Now, about half of that was EPS related to their anthropic mark to market. Just note that. But here are the numbers and why. I think the combination of the deal with Dario and the 240% year on year growth of AgentForce to a $1.5 billion AR run rate. And then, of course, 3.9 billion, 210% growth when you add in the whole data 360 ARR. So huge growth for the company. And they raised their guidance to 46.1 to 46.4 billion. Look, this happened and we'll talk a little bit, I think, you know, we're not gonna, we don't hit it today, but you know, CrowdStrike happens with security names are booming. SAS names are booming. You're seeing, you know, 20 to 30% clips. This to me is the market realizing that it had just gone wrong and oversold the SAS. You don't see parabolic moves like this until the market has realized it's made a mistake. It made a mistake. And so this was a good number, but all to me is this headless agent force and growth of 360, because this is what's indicative of the future to me, rather than the sort of RPO burndown on historic customer enterprise business and relationships. So not a lot to add there. I mean, I think it's always great when they can raise guidance amidst all this, but this is a sea change.

‍

Patrick Moorhead: 

Yeah, it really is. I sent notes to both Yahoo Finance and CNBC before my broadcast appearances. And the thing that I really pointed out was AgentForce needed a durability bar. And I think it got exactly that. You had a $1.5 billion ARR going 240%, premium SKU bookings, agent force one, agent force for apps, more than doubling quarter on quarter. And half the AI bookings came from existing customers expanding. So to me, this needed to be a show me quarter and Salesforce Salesforce showed them. I also think investors viewed Dario and Mark Benioff on stage potentially as a anthropic relents and realizes that they're a lot better off working with these SaaS companies than spouting about how they're going to replace them. And that could be a sea change. That truly would be. I don't want anybody to get confused, though, to think that enterprises still aren't trying really hard. to reduce their SAS expenses. That's been a consistent theme when I talk to CIOs and even CEOs on what they're trying to do. I've also made the case that companies are going to start with the most strategic workloads that they have. And let's say if you're a manufacturer, they're going to try to put effort into ERP systems like that first. If you are a marketing and sales-led company, don't do a lot of heavy manufacturing, you're going to put a lot of that effort into optimizing, let's say, the CRM and the service flow. So really good to see these guys got a lot of re-ratings. TD Cowan, Bernstein, Jeffries. So congrats, Mark Benioff, and we appreciate you coming on the 6.5 and giving us essentially what was a pre-look at what was announced with Clawed Force.

‍

Daniel Newman: 

Absolutely. So, all right, we've got a bunch kind of rapid fire here. We have four more earnings to do. Give me your read on Synopsys.

‍

Patrick Moorhead: 

Yeah, so listen, even though they had a triple beat, actually one, two, they had a quintuple beat on revenue EPS. Sorry, revenue EPS and the guidance, that would be a triple beat. The stock did come down. I think people got confused with what the story was with IP. You had some differences between FactSet and LSEG, or you said a few people which showed that they actually missed it when they actually hit it. Cost, really good operational expense control around the Ansys integration. I think we can clearly say that it's paying off on the expense basis. And the company said 2027 was the year when you're going to see the revenue lift from the incremental revenue lift of the two companies coming together. Yeah, all in all, I think a positive quarter. And I think it's very clear. And again, that that AI isn't going to wipe out any SAS, and it's certainly not going to wipe out the EDA toolmakers synopsis.

‍

Daniel Newman: 

Absolutely. And I'll lean in. Look, I like synopsis a couple of things. One, I like the royalty model that they moved to on a unit basis. You know, historically, they didn't always have the benefit as these projects ramp to benefit at the scale of the various ramp. I like, you know, kind of we talked about the advancements and speed that AI is driving new custom chip projects to market. This is a company that's got a very early view. You know, I did talk to their CEO on earnings day. You know, they have the first, they have a very early look, kind of like the, you know, WFE companies do at what's happening and where projects are at. It was a very bullish tone. as to the project in demand, of course, and this is on the AI chips. And then the Ansys business gives them a whole read into the physical AI robotics autonomy space. It was a very clever move that I think a lot of people are underestimating now because they went from kind of chip level design to systems level simulation innovation. So it was a good report, good earnings. I mean, you know, the market can be nitpicky, but I actually thought it was a really, really good print. HP had a good quarter, across the board, a big beat over the revenue consensus, good earnings beat, came out with a strong growth of personal systems revenue. You know, I think it dropped. There's a lot going on there. First, they still have an interim CEO that I think the market is still kind of figuring out what's the long term strategy direction there. I do think the market appreciates and understands the devices. Part of the market right now is mostly cooking on supply. So the constraint supply, which is giving a lot of ASP pricing power. And so while there was some strength to the product mix towards some of their higher end devices and SKUs, I just don't know how much that story really holds up. in terms of creating excitement. What I'm looking for from HP has been more clarity on how they are going to capitalize on the AI at the edge story, how they can deliver token economics across models to deliver workloads on device, on small workstations. Basically, that's where I see the opportunity for growth, because they're kind of fixed on how many units they can build for a while. Memory is going to be a real meaningful constraint. They're going to have that pricing power, but can they make more money by becoming the arms dealer to the edge AI narrative, which I think there's a lot of enthusiasm, and it's very early.

‍

Patrick Moorhead: 

Yeah, really good points. The sell-off to me was a margin trust discount, not the demand signal. And what I mean by that specifically is, I think it was 18 cents of, sorry, eight cents was tariff related, right? Where they actually got money back on that. I agree that a CEO needs to be put in quickly. I think that's kind of more of a, a macro weighing down, not necessarily this quarter. But investors, like you said, they need to see the demand from distributed AI computing. You and I both buy into the theory if you have heavy-duty clients, on-prem infrastructure connected to the cloud with a router, an agentic orchestrator, enterprises will be jockeying workloads that start start on the client, and my gosh, look at DGX Station, that's $100,000 a box, and its little brother is $5,000, and heavy-duty margins.

‍

Daniel Newman: 

Absolutely. All right. Another one that, I mean, Pat, the era of storage is upon us. Ever pure good result. How'd that how'd that go over?

‍

Patrick Moorhead: 

Yeah, triple beat revenue, EPS and guidance, which was good. I mean, a stock. I believe the stock rocked. Right? Actually, I've got conflicting data here. I'll pull it up because I've just- Yeah, why don't you pull it up? So- There's too many charts lately. Go ahead. Eight straight quarters of accelerating revenue growth is a really good, and by the way, that's before the second hyperscaler kicks in, which was- Actually, the stock got crushed.

‍

Daniel Newman: 

It's down 15%. So we're flipping probably among the various names we looked at, but it wasn't 15% on bad results.

‍

Patrick Moorhead: 

Yeah, it's so funny. My stock reaction said it has conflicting data. So my apologies on that, you guys. No worries. I'll have to put my AI back in the box.

‍

Daniel Newman: 

Get his head examined. It's head examined?

‍

Patrick Moorhead: Yes. So I think, you know, net net here. Charlie and the company have had the ability to better time its revenue with the cost impact. It gave its customers an initial reprieve, probably on those long-term contracts, but this is where it kicks in. This one's a slight mystery to me. They may have reacted to the cash, which by the way, The cash drain was primarily due to them buying more storage earlier, which I think is a good use of cash, particularly if you can turn it into very profitable revenue.

‍

Daniel Newman: 

Yeah. You know what, what I'll say here, Pat, is the numbers were good. The hyperscale growth is good. They've got their new data offerings, and that's really going to be a big moment for them because it's all about can they expand their TAM. Of course, you know, they're winning. We think, we philosophize that they are winning. workloads from their competition. We'll see that in some future earnings that are going to come out from NetApp and others. Our ETR data will be able to tell if that was accurate. But, you know, this is strong. And I mean, you know, they've been, you know, they've had their pricing margin stuff they've managed. They've kind of held their margins down, but their margin gross dollars is way up. So they're kind of benefiting from all of this. And I still don't think the full run on storage yet has really, the crunch has really looked anything like how memory and other parts of the supply constraints have shot up. So I still think there's an opportunity here in the numbers.

‍

Patrick Moorhead: 

One other correction, if I couldn't screw this up more. Second hyperscaler comes in calendar 27, but fiscal 28. There you go.

‍

Daniel Newman: 

Don't worry, I saw the zero hedge talking about NVIDIA's 28 revenue in 2028. Yeah. You know, these accounts, they don't have any idea what they're talking about, but they have lots of followers. So it's a great way for bad information to be spread. All right, Pat. Maybe I'll be more popular then. Yeah, we'll just make stuff up. Just make stuff up. You know, when in doubt, rocket ship. That's the thing. All right. That's a joke at myself, people. I like rocket ships, I'm very optimistic about things. Marvell, they printed what everyone expected. So this was a challenge for Marvell was they've had a nice run. They've been called the next trillion dollar company. Matt Murphy did a great job speaking at our event. They had this amazing Google deal that literally like more than doubles the size of the business if they execute it. And of course, then Google will own a piece of Marvell, which who doesn't want Google to own a piece of their business and be committed to spending billions and billions of dollars a year on, you know, custom Silicon across your entire infrastructure stack. But this is one where they literally came in like right at revenue, right at earnings. You know, the data center story is what everyone expected. And so I think it went down a little on a little bit of a build into earnings path. But like, this to me is just, they're just executing. They're just kind of executing. And you know, everybody now wants these massive, guide revisions and beats. And I think Marvell is playing a little bit more conservative. You know, it got hit, another one that got hit for a 10% plus drop this week. But I actually think nothing's broken here. This is just one where they're just doing what they said they're going to do. It's showing up in the numbers. But I think the market is kind of trying to run ahead of its growth a little bit and wants to see those, you know, 20% guidance raises because of an XPU project. That's just been Marvell's fate for some time.

‍

Patrick Moorhead: 

Yeah, they had like a 1% beat on revenue.

‍

Daniel Newman: 

Not even a beat on EBS, like at the number.

‍

Patrick Moorhead: 

Yeah, it was 94 cents versus 92. And it's very clear. right? These guys are up 187% year to date. And Wall Street wanted the Google math and they wanted it integrated in there. My hope and belief is that their October Financial Analyst Day is when they will be rolling out and giving more insight into the $120 billion of of Google money that comes in and maybe some of the details on what are the trigger points for that. I don't mean date trigger points. I mean, what does Marvell have to do by when to hit it?

‍

Daniel Newman: 

Yeah. Well, like you said, I mean, you know, when you're Matt Murphy, you're playing the long game, almost 200%. Like let's not rock the boat. Let's not create unmanageable guidance. Let's not over promise and under deliver. Let's just keep winning sockets and let's just keep winning attach and let's just keep doing what we're doing. And I mean, I think there's a lot to like about that. All right. We did it. We did it. We. God, we are long winded. We didn't make it through all our decode. We probably could have had more earnings topics, Pat, but God, we have a lot to say. And I hope everybody out there appreciates all the great things Pat has to say. And maybe even some of the things I say sometimes, but a lot of fun, good show. Money wins, Pat. Money wins. Money wins.

‍

Patrick Moorhead: 

Don't be greedy.

‍

Daniel Newman: 

Don't be greedy. And also, I just have a little advice for you and anyone out there. A little delusion goes a long way. Too much delusion doesn't pay. All right. Thanks everyone for tuning in. Appreciate you all being part of the Six Five community. More next week. See you all later.

‍