The Six Five Pod | EP 294: AI Capital, Sovereign Cloud, and the Infrastructure Arms Race

The handpicked topics for this week are:

1. OpenAI’s $110B Funding Round & Amazon’s $50B Commitment: OpenAI secured a $110B round backed by Amazon, NVIDIA, and SoftBank. Amazon committed $50B over eight years, including Tranium capacity, co-development, Bedrock integration, and custom model initiatives. Microsoft remains the exclusive API cloud provider, but the competitive cloud dynamics are shifting.
   
   
2. Anthropic, the Pentagon & the AI Safety Line: Anthropic risks a $200M DoD contract over refusing to drop safety restrictions related to mass surveillance and automated weapons. Pat and Dan explore the ethics and competitive positioning of this, and what happens if another lab steps in.
   
   
3. Model Distillation & IP Risk: Anthropic cited 24,000 fraudulent accounts generating 16 million interactions to distill model capabilities. The episode examines IP theft, enforcement gaps, and global competition.
   
   
4. DeepSeek & NVIDIA Blackwell Reports: Recent reports suggest DeepSeek leveraged NVIDIA Blackwell chips. The hosts discuss export controls, enforcement realities, and whether this was ever realistically in doubt.
   
   
5. Microsoft Sovereign Cloud Goes GA: Microsoft introduced full-stack Azure sovereign cloud capabilities with support for disconnected operations. Sovereignty, regulatory compliance, and latency management are becoming core enterprise and government requirements.
   
   
6. AMD’s $100B Meta AI Infrastructure Deal: AMD secured a massive multi-gigawatt inference-focused deal with Meta using MI450. The discussion centers on competitive dynamics with NVIDIA, scale-up architecture, and whether AMD can materially shift market share.
   
   
7. Intel & SambaNova Alignment: Intel Capital invested in SambaNova’s Series E. The hosts examine inference strategy, CPU resurgence, and how Intel rounds out its AI positioning while advancing its GPU roadmap.
   
   
8. The Flip: Is SaaS Permanently Repriced? Are enterprise SaaS multiples structurally resetting due to AI agents and consumption models, or is the market misreading enterprise AI adoption speed? Nuance emerges around consolidation, consumption pricing, and the durability of complex enterprise platforms.
   
   
9. Bulls & Bears: NVIDIA, Salesforce, Synopsys, Dell, Snowflake, IBM, Everpure, HP
   Strong earnings across several big tech companies met with mixed market reactions. Terminal value concerns, consumption transitions, stock-based compensation, and memory constraints shape sentiment more than raw performance.
   
   

For a deeper dive into each topic, subscribe to The Six Five Pod so you never miss an episode.

Listen to the audio here:
IMPORTANT: Note to Uploader: Libsyn AUDIO EMBED – added by uploader.

Disclaimer: The Six Five Pod is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and reference share prices, but nothing discussed should be taken as investment advice. We are not investment advisors.

‍

Patrick Moorhead:

‍
We are back and we are jacked this morning. I had a great back workout, in fact, and I'm feeling pretty good. Got back late last night in the gym at 7 a.m. Daniel, you're looking swole. You're back. You're jacked. You're packed.

Daniel Newman: 

That's a fact. Let's go. Let's go. Let's go. Isn't it? Aren't you the host? I am. And I already said stuff, but it wasn't like a lot of stuff up at seven.

Patrick Moorhead: 

Like that's an accomplishment. I was talking about my back workout, dude. This is just, you know, you and I are like married and you just don't pay attention to anything I say. Like, what are we going to do?

Daniel Newman: 

That's how the relationship survived. Do we need to go to marriage counseling, Danny? No man needs counseling. I think I read it. I think I saw an Instagram post on that yesterday. We just need to hang out with our friends, make fun of each other. Um, you know, talk about deep stuff, you know, maybe have a good couple good meals and, you know, go scream outside at the top of our lungs every once in a while and we'll be fine. I love that. I love that.

Patrick Moorhead: 

We had a great show for you today. We have so much. content here. But before that, we are going to be at Mobile World Congress. Daniel and I are flying out there this weekend. Hope to see you there. Tune in to all the amazing content. Daniel, we've got so much. I mean, my God, we have got OpenAI. We've got NVIDIA. We've got AMD. We've got Anthropic Fighting with the Department of Defense.

Daniel Newman:

War. Department of War.

Patrick Moorhead:

Department of War. We've got Sass-pocalypse. We've got just so many things going on. So, hey.

Daniel Newman: 

The merry-go-round of shit to talk about and he just keeps coming back, right? No, it's just, yes. And by the way, by the way, You can't even plan these shows anymore. Cause like we had a plan yesterday, right? And then like this morning, you're like, Oh crap, I guess everything that we actually had thought we're going to talk about has to be like down or moved out or added. And the docket gets longer. And, you know, cause I mean the open AI stuff, there's like three or four announcements all tied to just that deal that we didn't even have a plan to talk about yet.

Patrick Moorhead:

 I know our producer team, they threw that in there about 15 minutes ago.

Daniel Newman: 

All you need is Clawed Bot, Open Claw, constantly searching the news, populating the document. Pat, I am now a team of 50. It's me and 49 cloud code terminals.

Patrick Moorhead: 

Listen, buddy, I've moved, I've moved off a cloud completely and I'm on to a perplexity computer. I vibe coded for competitive products and services with pricing. I've got backend, I connected 14 different databases. I'm good to go, dude. Let's go.

Daniel Newman: 

Okay, let's jump into like, is your platform already? Like, have you done like everything I've spent two years doing in like a week? Well, I mean, it took me six hours. I quit, I'm done.

Patrick Moorhead: 

I just sent the URL to your site. I gave Perplexity Computer a link to your site. I threw that promotional video in there and said, vibe code this, double the revenue and make no mistakes.

Daniel Newman: 

By the way, I'm telling you though, just quickly, because I know we've got a huge docket, but that video I sent you, I don't know if you've been able to get it to work, because it's not perfect yet. But I talked to an agency about creating a video like that. Nate wanted to charge me $150,000 to do it. And literally as I'm sitting on the call with this agency, I go into Claude and I start coding out like, help me write a script for a video that does this and helps me tell this story. And then if I put all the parameters, I gave it all the strategy docs and the financial docs that I'd used and all the backend stuff. And literally it spits something out. I'm reading the script. I'm like, God, this is freaking awesome. And then I go in, give me the tools that I would need to do to create this. And dude, we did all AI, it was like 15 minutes. And I'm not saying it's gonna be as good as like getting a whole massive, but like, that's why people are panicking, dude, because this stuff, like my new question for everything is, how are you 10Xing yourself with this? Cause that's what every human should be thinking about right now. Sorry. All right. Back to the docket, back to making me talk about items. I just want to talk about myself. That's good.

Patrick Moorhead: 

Well, we got to kick off that. Otherwise we're going to be here for two hours and we're going to lose like 87% of our audience. So let's dive in. Let's hit the decode. Let's decode all the amazing stuff. Hit it. All right, Daniel, right off the presses, OpenAI gets $110 billion funding round from Amazon, NVIDIA, SoftBank. Amazon and OpenAI go super deep, deepest I've ever seen. Daniel, break it down for us, buddy.

Daniel Newman: 

So Amazon, a lot to talk about here. There's a couple of things I want to talk at a different level. $50 billion over what is it, eight years? It's a long-term agreement. The focus of this is going to be one is OpenAI is going to use a lot of Tranium. This is a, you know, this is really interesting, Pat, because I know this is what everybody got all worked up with the NVIDIA deals. At the same time, I'm the guy that's consistently and every time these deals come out and say the smartest thing corporate venture can do is invest in companies that are going to heavily use their technology. This week's been a week of smaller chip companies doing what they have to do to compete with the monolith that is Jensen Hong and NVIDIA, who has 95% share of the GPU space. And, you know, Tranium is already fighting the battle of a very successful TPU product, which just won a huge deal, which we'll talk about later. And so, in my opinion, This is a win-win. And another really interesting thing about the deal is it feels, Pat, just like as Microsoft's kind of doing a stage left on open AI, kind of moving out of the picture, slow or maybe not investing at all in this current round, Amazon's like swooping in, being like, we'll be here. We're here for you. What it makes me wonder is, you know, what's going on with Amazon Anthropic, which is obviously a huge investment that they had made earlier. Is there something happening in that particular space? Two is Amazon had been particularly active in building its own models, which we haven't heard a lot about lately. And I don't know if that's a pivot away, like, hey, we're going to just go all in and have deep, deep roots with both OpenAI and Anthropic. And we're going to kind of step back from kind of doing our own models. We're going to focus on compute. Because I don't know if you've noticed, but a couple of these hyperscalers have basically just been clearing checks like on the daily to buy more compute capacity. And as I think a year ago, it was model capabilities. And you and I debated this last week on the flip. Remember the last one I won, whether it's models or infrastructure. It feels like this is the next sort of wave of pushing more for more infrastructure, expanding capacity, kind of having a guaranteed check, meaning like we're going to expand, we're going to give you the money, you're going to give it back to us, but like that we're going to continue to invest and build out this Tranium cluster. And we're going to continue to build massive amounts of compute to be able to run whatever model ends up being the best model. But you know, and again, there's more parts to this deal. But those are kind of my big couple of takeaways off the gate.

Patrick Moorhead: 

Yeah, so what I want to hit is the AWS and how it's different. First of all, there's equity. So stepping back, there was a major November 2025 agreement that we had discussed. It was a 38 billion dollar a deal. And so there's six different ways this deal is different with that. First of all, Amazon's investing $50 billion into open AI, $15 billion up front, and $35 billion conditional on an IPO or AGI milestone. Second, Tranium is involved. There was some discussion on whether in that initial deal, Tranium was involved. It wasn't, but now it is, and they committed to two gigawatts of Tranium capacity. Third is co-development. You've got the stateful runtime environment on there, and that's not just teeing up models, it's essentially you can get these through Amazon Bedrock. Now, Microsoft is still the premier API vendor, but when it comes to this integration, it's definitely a step up here. Fourth, exclusive distribution for OpenAI Frontier. Fifth, custom models. They're going to develop custom consumer-facing applications. And I think this is a signal that Amazon's own models have been put to the side. So some interesting stuff. Microsoft still, again, is still structurally embedded. It's still the exclusive cloud provider for the OpenAI APIs. OpenAPI first-party products to host on Azure, and exclusive licensing and access to OpenAI IP. So it shouldn't be viewed as a, you know, Microsoft's getting kicked out or something. And Microsoft has been trying to diversify their way through OpenAI for forever. I would say, you know, first when they took their foot off the gas on the investment in data centers, and then you saw OpenAI going to a bunch of other places. And then when you saw the, you know, the official kind of ownership stuff, it crept up as well. So hey, let's jump into another really important topic here, and that's anthropic Pentagon and China distillation wars. Two different things going on here. You've got Pentagon threatening to cancel Anthropic's $200 million contract if it doesn't drop the safety restrictions. And Anthropic CEO basically said, no, we don't want to be involved in anything that is mass surveillance or killer robots. My words, not there. And this is going to be an interesting one because it's put all the other companies in a position of if somebody says yes, there's going to be a ton of outcry. And then President Trump hasn't weighed in yet. And when it does, it could be super, super ugly for the companies who don't support the the US military. You know, I get the mass surveillance thing. You know, we've been mass surveilled for decades. You know, the whole Snowden stuff that came out. But the whole idea of not Um, you know, you know, when we could be at war with some superpowers coming up here, uh, not having the support of, uh, of tech companies, uh, I think could be a major issue, uh, for those tech companies.

Daniel Newman: 

Yeah. I mean, look. he seems to be standing his ground right now at the risk. And the big concern I have is him stepping out doesn't necessarily ensure someone else not stepping in. I could be wrong. And like I said, you know, so what's gonna end up happening is there's gonna be some other builder of technology that will be more than willing to use their similar, maybe not exactly the same technology for the same purpose. So it could help Dario sleep at night. Of course it wasn't his, but that doesn't change the fact that, you know, I mean, Sam Altman, I'm sure, will gladly step in and take that work on. And someone's going to do it, and someone's going to build it. And then it's kind of, to your point, none of this is stopping the surveil. It's just who's going to do the surveil. So I guess for someone that's had the success he's had, I guess it's a can-sleep-at-night-ethics decision he's making. obviously shutting yourself out from the government and contracts with the U.S. will be a headwind to the business. But at the rate they're shipping and building, I think they're going to survive it. So like, you know, I don't know. I guess it's good that he's got a moral compass and he's sticking to it because I don't know that that exists all that much anymore.

Patrick Moorhead: 

Yeah. Hey, just a quick one on the distillation war. Basically, Anthropica is saying, hey, we have like 24 million bots or incidents. Sorry, 24,000 fraudulent accounts run 16 million interactions to distill the capabilities of the model. And I don't think anybody should be surprised here. What I am surprised at is with all the distillation that went on with the previous two generations of models, There's no simple way to curb this. And my gosh, if you could get Claude Bot to come in there and do stuff, you know, 24,000 accounts is not a ton. And 16 million interactions is not a ton either. So I don't have a technical answer to any of this of what could happen, but it does show that you can steal IP. And quite frankly, the Chinese, they steal IP and they don't look back. And this is an area that they don't have the capabilities of. So here we are.

Daniel Newman: 

This sort of flows nicely to the next topic, so I can touch on it when we get there. I'll leave it there.

Patrick Moorhead: 

Yeah, this does segue into the next topic, Daniel, which is apparently, according to sources, NVIDIA Blackwell was used by DeepSeek to create models.

Daniel Newman: 

Duh. Sorry, like that was always the case, guys. I'm sorry that people wanted desperately to believe this thing was trained with a couple of Pentium chips, but like it was always the case. I'm sorry. Like there was so much silly, ridiculous bullshit reporting that came out that China had built the open AI killer, you know, in the MacGyver fashion, you know, running Celeron. The net of it is that it was always going to be done on NVIDIA's best stuff. Everybody knows that the Chinese labs are using NVIDIA's best stuff, whether it's through the cloud in Singapore, whether it's being smuggled through the borders. This just isn't a secret, folks. It's not an indictment on NVIDIA. They're not doing it. It's just an indictment of the fact that this is how technology has always moved. This is the histrionics of it. It was always obvious. There was no way they were going to build a better model with less data in that little time using low powered or incapable chips. So. This is just purely victory lap for me. I don't have a lot else to say about this. I got a lot of shit from people on the Internet when I said that they were doing this. But, you know, this is the facts, like you cannot build better models than the U.S. labs are building with inferior technology. This is also why the export controls exist. This is also why we don't want China to have access to the most advanced high and EUV machines. And you can just be pretty sure that whether we give it to them or not, they're going to find a way to get it.

Patrick Moorhead: 

I want to see the smoking guns on this. Note in the same week, and I think it's coming out from DC with the Hawks, we had the resurfacing of the stuff that we already knew in 2024 about a timeframe for China to attack Taiwan was 2027. And then we have this week, I feel like the Hawks are fighting back on us shipping Blackwell all over the place, like the Middle East and areas that I know had been kind of contested. And I want to see the evidence. While I know that Deep Seek One was Bill using more than Apple watch I do think that we need some some evidence if there's evidence and somebody broke the law then Somebody should be be brought to account So am I that boring you just are looking at your phone the whole time. I'm talking No, I thought we were talking to the audience are we talking to each other I Well, you got to listen to what I say. So you don't just repeat my brilliance.

Daniel Newman:

I was brilliant. I did. This was a, this was a mic drop for me. I can't believe we're still talking about it. I, this should have been the flip. Weren't you like a China dove? Like they're, they're using, um, Celeron and, uh, Pentium two chips with three paper clips. I'm kidding. I'm kidding.

Patrick Moorhead: 

Well, no, just for the record, I supported H20 being exported. I mean, you can't do anything but meaningless inference there.

Daniel Newman: 

And quite frankly- I actually did too. So despite the fact, because I do believe three or four generation old technology, we can definitely dump that on lagging technological nations. But I never believed they weren't bringing the good stuff.

Patrick Moorhead: 

Yeah, not for a minute. Yeah. So, um, hey, we don't know. Let's move to the next topic here. We don't normally talk about when something goes GA, but given the proximity to the research and all, that I'm doing and a lot of the conversations that I'm having, I thought this one deserved it. So this whole idea of, well, let me give you the news, Microsoft Sovereign Cloud when GA adds full stack Azure server capabilities And you get you get the full stack. So for years, there was this debate on hybrid cloud, right, essentially having cloud on prem. And I know everybody told me I was crazy and cracked. And this was the dumbest idea. You didn't have the scale to go with that. 10 years ago, where people did have good points is that the software stacks were super immature. The cloud guys are now in there because they're getting a lot of requests for I want to be disconnected and not only to mitigate risk from being unplugged from the country that owns that tech company, but also there are real latency concerns and there's data sovereignty. So the first one to jump in was Oracle. And they essentially, it was called cloud a customer and they did a copy and paste of the infrastructure. that was in their public cloud and stuck it on-prem at an enterprise or a government, put a chain link fence around it, and ran services, you could run the entire stack. of applications and software from Oracle on there, including stuff like Fusion, and obviously all the database stuff. So I thought that was a great idea. And quite frankly, everybody jumped in and did that, tried to do that, but not Exactly in this simplistic way right there were virtual embassies that that that were put up and. Now we've got Microsoft, and it's great to see full stack Azure and a few of the key applications like SharePoint, like server, like mail, and stuff like that. And you can run it fully disconnected. for basically as long as you want. Big step up for Microsoft and in my travels to the Middle East, Daniel, this is what this is about. In fact, in the UAE, G42 is running their Cloud stack on top of Azure's sovereign Cloud stack. I think we're going to see this more and more, particularly as the trust in the United States,

Daniel Newman: 

I totally aligned on this one. Not a lot of content to add, but I still think we're not hearing a lot about the scale and opportunity of Sovereign Cloud and these companies bringing these out. This is the opportunity. because I think there's gonna be a lot more regulatory around AI and certainly in certain regions around the country or in the world and where those regulations pop up, these companies are gonna have to be ready to react and the cloud is gonna be the fastest way to deploy sovereign inside of borders where you're operating your business or government or organization.

Patrick Moorhead: Hey, let's jump into a topic that just got all the headlines early on AMD and meta $100 billion, six gigawatt AI infrastructure deal.

Daniel Newman: 

$100 billion of opportunity for AMD. Great way we talked about with Amazon. This was a great way for AMD to break in in a more meaningful way to this inference opportunity. Meta is buying compute like it's going out of style. It's doing TPU. It's doing Meta. AMD just announced an NVIDIA deal. It did announce that it's pulling back on its own most advanced chip. I don't think that means they'll never build chip, but I think they're struggling to get their chip project going. And this is why they're leaning in harder because like my winning argument last week that I'll bring up again, knows he needs more compute. More compute, more compute, more compute, because that gives him the flexibility to deliver $60 plus billion a year of AI-powered revenue in its ads platform, not to mention all its other ambitions around AI. For Lisa and AMD, this was a great way to break in meaningfully. The 10% warrants, I don't want to hear about anybody's BS that that was giving away the company. She's giving away 10% of the company when it adds like $700 billion in additional market cap. I do that deal. I don't know about you, Pat. But this is how you get in. This is how you prove yourself. This is how you break in against a competitor that holds the market share that Nvidia holds. And this is a prove it moment. But I like it. And it's just more indication that these mega hyperscalers are going all in on building the infrastructure for the AI future.

Patrick Moorhead: 

Yeah. had the pleasure of catching up with AMD CEO Lisa Su and kind of talking about the deal. You did a great job kind of going down and peeling the onion here. I did want to lean into the deal part. Listen, 95% market share, you've got to do something different. If you've got a similar technology at a similar price, they're going to go with NVIDIA all day long. And the other thing is R&D. AMD, if they would stay at 10 or 5%, would essentially have 90% less R&D to potentially put. Rounding numbers, of course, don't at me. to invest in the next generations. So this is something they have to do. And my gosh, what a motivated customer, right? You've got 10% of the company in equity. You are going to make some accommodations to move that forward to drive value for you. And finally, I know everybody questioned AMD. You know, people got on the meme bandwagon, and I understand that. But you have to understand the architectures we're dealing with here. NVIDIA has had two generation of scale up. You can put a bigger model in, in one place in a more efficient manner to do training and to do better reasoning for better inference for reasoning models. And they had the CUDA advantage. I mean, you put those two together, and nobody should be surprised. One of the changes, though, is that you have MI 450 with its new OCP compliant rack, is a scale-up architecture, okay? And it is on the same plane as what, I'm not saying it's better, I'm not saying it's worse, I'm just saying it's there. ROCCM, the advantage, the CUDA advantage versus ROCCM, I do believe has, has gone down, CUDA still does have an advantage. And quite frankly, in the age of vibe coding, I'm wondering where that goes long term. So if you're still questioning AMD, I think you're a fool. Um, and, you know, you should kind of rethink, rethink your, your posture on it. I think the biggest threat, I still think the biggest threat to NVIDIA is not AMD, but it's the, uh, the XPU companies. Um, and we can talk about that later, but looking forward to, uh, opining more as we get into, uh, uh, GTC. And. in more chip, chip, chippery news. You've got Intel and Samanova. So there were a couple deals that, or a couple announcements that came out. So Samanova on its own brought out a new SN50 chip, and then Intel talked about what it is doing with Samanova. And let me be clear up front. The gory details are not available on this thing. But from what I know, I do like that Intel Capital is investing into Samanova's $350 million Series E. Get a taste of it. I do believe that Intel was trying to acquire Samanova and whether it was the Grok mega valuation or other forces, I think that deal got killed. I also like the pragmatism of being able to leverage in-rack inference before Intel gets their GPU roadmaps in there. Intel has always been amazing at go-to-market and I think the ability and the relationships that Intel has with the hyperscalers and also the tier two guys could move the needle. I don't think this is an enterprise play yet because an enterprise isn't going to, you know, accept a, you know, an offering unless they believe it's long term. So we'll see what happens. I'm looking forward to more details on this.

Daniel Newman: 

Yeah, I mean, Intel has to have a play here. You know, Intel's got a couple of good tailwinds. This partnership certainly can help. Your note about the details I think is important. The devil lives in those details. So it would be good to know more. But Intel having a bigger toolbox and different ways to make money off of AI chips. And then, of course, Salmanova has some interesting specs around its inference capabilities. And by the way, Intel is having a bit of an inference moment because CPUs are having a moment and CPUs are back, baby. Yeah, it only took a while. You know, we're in the what's old is new again moment. You know, memory got cool again this year. Now CPUs are becoming cool again. And that's obviously got nothing to really do with SambaNova, but it has a lot to do with if Intel can figure out its capacity issues, it can scale. It feels like there's going to be a heck of a lot of demand in this agentic era for, you know, serial processing because agents don't work parallel the way GPUs do. Agents need to work in an order, and that's something logically CPUs are particularly good at. So we're going to get a new CPU war. The AI CPUs, the generative AI CPUs, they won't be like AIPCs. These will actually sell really well. Joke people, relax. It's a joke, Jan. But anyways, Intel 

needs to round out its AI story. And so this is helpful. Good stuff, Daniel.

Patrick Moorhead: 

All right, Daniel. Now it's time to go on the flip where I can train you like a, like a dog.

Patrick Moorhead: 

All right.

Patrick Moorhead: 

Question for the day is SAS being permanently structurally repriced, or is the market just misreading enterprise AI adoption? Let's jump in. All right, time for me to start. SAS is permanently being restructured and repriced. Daniel, this is an easy one. This is like looking, this is old technology versus new technology. But I do, I want to do this with a little bit of a nuance. I believe that Those businesses that aren't, I would call, agentified, agenticified, is that a word? Are absolutely going to be wiped out now. But inside these enterprise SaaS players, they're playing a dual game. And then the question is, is will their agentic technologies and capabilities grow faster and get bigger as what they're doing essentially today? And I think that it's going to be very hard for the new technology to grow at a rate and to a size and to a profit level that the, I'll call it the traditional technology will hit. I do recognize that it takes, when you're in an enterprise, it is going to take some time to bring up these these kind of horizontal agents that can do vertical applications for you. But we have time and what enterprises are going to do is where Things are strategic for them. They're going to attempt to roll their own agents first, and then they're going to pay the AI money to those SaaS providers on stuff that they think is less strategic. I don't have a scenario, Daniel, where they get to the same pricing paradigm, and whether you want to call it the rule of 40 or something like that, I'm just not seeing that long-term. Not because they're not delivering good technology, but because they're going to definitively lose the extra AI spend because other people inside the enterprise feel that they can do it better. And with that, I give the floor to you, Daniel.

Daniel Newman: 

Yeah, it's real simple. The pullback's too much, too far, and while we can have a nuanced debate over a reprice, though reprice is not permanent. It will come back. These companies are going to pivot to consumption-based offerings. They're all continuing. Snowflake reported this week, Synopsys reported this week, Salesforce reported this week. Through these entire earnings, not one of these companies has seen a structural pullback in their revenue or growth rates. because of the SaaS apocalypse. This is purely the market believing that we're going to have our grandson Vibecode, our ERP system, which is going to replace all of the systems and rules and records and rails that we run our businesses on. We're not going to Vibecode a transaction system that manages a supply chain of a major manufacturer. You can be absolutely sure that ain't happening. Plus we clearly don't know how to secure things and neither does Anthropic. It had 24,000 Chinese bots canvassing its platform to steal data and to build better distilled models over there in China. We just aren't ready yet. It doesn't matter whether it's AI is gonna replace software, AI is gonna replace security, AI is gonna replace manufacturing. Next we're gonna say AI is going to replace construction and AI is gonna start paving roads and AI is gonna make the sunrise and AI is gonna help me take a dump When I when I have a constipation, it's just I mean, maybe it'll help me figure out the right medicine I need to take, but it ain't going to help me take that dump. The dump on the stock prices, however, is it's temporary. These things are going to come back. These multiples will grow. These companies are moving to consumption. Nobody's moving away meaningfully from the platform. Yes, sure. There are a few shit codes that will get destroyed, but that's nothing to do with the actual construct of the constructual of the SaaS industry, that just has to do with some bad companies that were overcharging for services that just hadn't yet been created. Now that AI is here, it'll make all these services better. These companies will get a new pricing, but it won't be the current repricing. So I guess we can semi-agree there, but it's going to come back meaningfully from all the way down to where it is, but it isn't going to go all the way back. We can agree on that.

Patrick Moorhead: 

Hey, when do you think the decisions were made that impacted earnings this week by these companies, by the enterprises? When did they make the commitment that found its way into earnings today? Six months ago? A year ago?

Daniel Newman: 

When did the customers make the commitments? Yeah. It depends what you're looking at. I mean, net new subscription, NRR, a lot of that stuff comes in the quarter, upgraded subscriptions, product revenue expansion across the portfolio. I think 25% now of Salesforce is outcome based revenue for their agent force business. So they're starting to do consumption based. So I think that, of course, they have these lock-ins and these long-term contracts, and people are definitely… But I also think enterprises move pretty slow. I mean, look, I think now that we're kind of done debating, the bottom line is the repricing is permanent. I nuance that by saying it is permanent, but it's not going to permanently quite as bad as it is today. I think the baby's been thrown out with the bathwater. I think the consolidation of the industry is real. I think that shit codes and products or sorry, features that became product companies, like if you're a billing software, it just does nothing but billing SAS. I think you're screwed. If you're an expense management software that does nothing but expenses, you probably can be vibe coded. If you are a very, very small business CRM, like a company that just has very minimal needs of customer relationship management, I think you can vibe code that. I definitely do. But when you're talking about the complexity of large enterprises with multi-border governance and tax compliance and we're reporting compliance and, you know, masses of PII and proprietary data you're protecting, we're a long way from vibe coding that out. And I think what you're going to end up is with agentic wrappers, AI wrappers from things like Anthropic that will be, and you're hearing that, the partnerships with the SIs, the partnerships with the, like, I think what we're actually hearing is what's going to happen And I do think it'll cause the rating because those companies become less important because you use an anthropic like wrapper around all of them, but they're not going to go away.

Patrick Moorhead: 

So I guess my point was about timing. Like, let's just assume that the decisions that made its way to earnings. happened a year ago. That was three generations of models ago. And here we are today in the capabilities. Also, I think the thing that might be missing is that it's not that you're going to stop doing business with a SaaS provider. What you're going to do is you're not going to upgrade. And you're going to use it essentially as a giant dumb database.

Daniel Newman: 

Deprecated database, as our friend Satya said.

Patrick Moorhead: 

Well, I don't think that that gets into any PII and you're not changing security.

Daniel Newman: 

And anyways, I think we're actually agreeing largely that they will be less important and they will have a lower multiple. I took the hedge, man. I said, basically, it won't be the current re-rate is not the permanent re-rate, but they will never return. What I will argue is they will never return to the premiums that they once had. Sounds good. It was a good nuance because it was basically the way I could be 100% sure that I would beat you in this argument without having to actually disagree with something I agree with.

Patrick Moorhead: 

It is. And quite frankly, I get tired and annoyed with markets discussions. I mean, if anybody can figure out the freaking markets, stop doing what you're doing and just invest all day long.

Daniel Newman: 

This is where Pat's sob story is. Basically, he has a permanent history of entering markets at peaks and selling in the valleys. My portfolio blows. You got a great real estate portfolio to your credit. So, you know, you've done well in some things and you know, and other things not so well.

Patrick Moorhead: 

Hey, let's jump into something very straightforward, man. A lot of market news. Let's jump into bulls and bears. All right, Daniel, it was NVIDIA week and therefore we will start off with NVIDIA earnings.

Daniel Newman: 

Good earnings, what's next?

Patrick Moorhead: 

I'm just kidding. Great earnings and a giant sell-off.

Daniel Newman: 

Good God, good God. I don't know about you, how many reporters were you trying to answer this question to? But why are we selling on that report, that print? Everything, margin good, data center number good, overall number good, six billion of robotics entering into the next trillion dollars of market cap opportunity. Good. Forecast, great. Like what did they guide? Six billion over the consensus? I mean, and it's selling. The market just fundamentally does not believe that there's not another shoe to drop here. That's all it is. Like there's a shoe to drop. Nobody believes right now that this thing can be up and to the right. I think it's also a lot of like kind of what I would call the ecosystem risk around NVIDIA. Like NVIDIA's margins in products and sales are probably pretty safe. unless the shoe drops, unless OpenAI collapses, unless Oracle can't build its data centers, unless we end up with a 17% unemployment like Citrini said. Like if all those things happen, there is a problem at some point, there's nobody to buy this stuff. And I mean, I know the market's super forward-looking, but I think there's a little bit of fear that's come out. I mean, I know you and I have texted in private and we're not gonna discuss our private, but we read some of this crap and even us in the middle of this stuff can wake up and all feel a little overwhelmed. by everything that's happening right now. And it makes it really hard to be confident that this party can continue in perpetuity. So an amazing print. If we're just talking about the print, freaking amazing. But if you're talking about why things are selling, I think it's way more nuanced.

Patrick Moorhead: I still don't understand why it was down. I have yet to get, Daniel, I respect the heck out of you and your market's capability. really did NVIDIA go down? Was it the competitive front? Was it puts and calls that were set to 200? Like what on earth happened though?

Daniel Newman: 

I genuinely think that, you know, four and a half trillion dollar market cap company, there's a lot of value in that. I think that, like I said, the extended risk of the customers that buy NVIDIA concern people. I think people do still have concern about kind of the circular deals that are kind of creating all this momentum in the industry. There's a lack of sort of true clarity on how much, what's the right word? how much token, how much revenue tokens are generating in actual profitability, meaning as, because to keep up generation to generation, to keep buying all this CapEx to create efficient tokens so that people, because like, there's a lot of like, there is some subsidies on with all these models. Like, Pat, you and I are probably consuming way more than $99 a month worth of tokens with all the crap we're doing, right? So how do they become profitable? The things that are using all this technology really generating meaningful profit, and there's a few companies that are, but there's a heck of a lot that aren't, and that are gonna have to spend a fortune to get there. So I think that's a lot of the concern, in my opinion.

Patrick Moorhead: 

Yeah, I net it down to terminal value, right? Because of that risk, it was the terminal value that's being questioned. And, you know, finance 101, right? weighted average, cost of capital, cash flows, infinitum. I get it. Hey, let's jump down to Salesforce. So we hit the essence of the debate a little bit in our flip topic. But here's the net. I mean, Salesforce had a very good quarter, right? They beat on revenue, they beat on EPS, they freaking beat on EPS by 25%. Okay. CRPO, they beat, they missed on guidance, slight miss on the full year, but 22,000 deals with agent force in Q4, 50% quarter on quarter growth, and the stock was down. right, for all the reasons that we debated in the flip. It's just not good enough. You know, I watched Mark Benioff get on Kramer and make his case. Man, I have to applaud a Fortune 500 tech CEO who bangs it out and does the debate. And he did deliver a pretty compelling He talked about the other two SaaS apocalypses that he had been there. But I think when it comes to market, what they're going to have to see is continued growth to negate what I talked about, which was when you made the decision and the state of AI that was out there.

Daniel Newman: 

Yeah, I mean, look, it was, again, this goes back to our debate. The print was pretty good. The guide is pretty good. The partnerships with the AI platforms are telling that that's got some stability. The move over to kind of more consumption is really important. And to your point, like Mark has done a good job of being out in the, you know, gladiator in the arena fighting. Yeah. But it is a battle that's gonna be fought for some time.

Patrick Moorhead: 

Yeah, for sure. Hey, let's jump into another software company who probably doesn't want to be called a software company right now. And that is Synopsys. Dan, how'd they do?

Daniel Newman: 

Yeah, I mean, they did good. High end of the guidance. The earnings was above the guidance. They reiterated their year. They expanded a buyback. They clearly think their stock might be oversold. This is a company, I think, again, Are we going to vibe code a platform to design the most advanced semiconductor chips and to simulate robots and, and, and, and massive physical components and I think. I think not. Is it possible? Maybe. But I think this company, that's going to be really important that they're out in front of that, that they're talking about the combination of their data moats, their IP, the relationships and accessibility, the trust they've built, the security around their capabilities are going to be really important to make sure that they lock in. And of course, it was good to see the ANSYS number show up and really now a newly combined company. I think has a pretty important mode. And there's a reason that NVIDIA and them are so deeply partnered. And it was a pretty good quarter.

Patrick Moorhead: 

Yeah, the company did what they say they're going to do. They did have beats. And if there's a software company and services company that I think is the least likely to get taken out, it's EDA as a category and Synopsys in there. And if you look at where the puck is, it's combining EDA and system design software together. And that's why I keep saying that it's the next power broker category, along with WFA, sorry, WFE category out there. All right, speed round here. Let's go into Dell earnings. God, where do we start? Like I saw the numbers, Daniel, and I literally thought, Like I'm reading the wrong company. I mean, these guys absolutely freaking cooked monster monster. I mean, beat and beat handily destroyed pretty much every everything. out there, stock went up 10 points. I mean, product revenue, I mean, revenue was up 30% year on year. EPS was up 20, beat by 26%. And it was all about data center and all about ISG. And surprisingly, client computing rose 14%, which may or may not be based on people trying to pre-buy before the memory apocalypse. out there, but Dell is moving with the market. Neo clouds are getting bigger and the reality of sovereign AI clouds is turning more real by the quarter. I think Dell has pricing power in areas where they can get commodities. Dell is very well known and has a very capable and industry-leading commodities practice. And I don't mean selling commodities on the Chicago Board of Trade, but basically supply chain management. They do long-term agreements, which really, really help you up front. And it gets more equalized kind of every successive quarter versus the competition. But I believe that Dell will end on top of a lot of other companies just because of what they can do in commodities.

Daniel Newman: 

Got it, yeah, absolutely. You know, we got to fly through these last few because we got to get going, but Dell, absolute monster result. The AI number is staggering. You know, congratulations to the team. And by the way, it's going to be fun to watch their core server business start to rip now that CPUs are ripping too. You know, the PC part of the world is still a little under, it's not super exciting. I don't think it will be for some time, by the way. That's right. All right, Daniel Snowflake, what happened?

Patrick Moorhead: 

Software company.

Daniel Newman: 

Strong numbers, strong results selling. I mean, look, this is another case of a company that's super high forward multiple, super high amount of stock-based compensation, isn't creating a ton of cash flow yet. And I think the market's just nervous about the composition. I do think the technology is robust. I think it has a real platform to scale in the agent era. I think it's doing the right things. This is a, you know, this is one that just, we beat the crap out of this topic today. They are a victim of the SaaS apocalypse. And I think they're one that will be okay though.

Patrick Moorhead: 

Yeah, I mean, they showed that AI is not just talk, right? They drove real consumption there. I think it's the 45X forward revenue on the stock that's weighing.

Daniel Newman: 

And they have a huge amount of stock based compensation.

Patrick Moorhead: 

Yeah. Daniel, IBM shares dropped based on Anthropic saying it did some deals with some GSIs on COBOL. I got cited in a way that I normally don't get cited on CNBC. They put up a tweet that I had that basically called it dumb. Uh, and why would I be so strong on that? Well, first of all, uh, IBM, uh, makes more profit dollars, uh, moving their clients from a cobalt, let's say a Java you, and you can't replace the mainframe. Uh, why do they make more money? Uh, they're going to buy more mainframes. You're going to buy more services. Uh, they're going to buy more red hat. They're going to buy more everything. Uh, when that happens and IBM. has been doing stuff with Anthropic and COBOL and moving that forward all ready. So this was a just complete and utter nonsense.

Daniel Newman: 

I agree. I actually retweeted your tweet. So my opinion is your opinion on this one. Don't have much to add. Just the fact that this is one of those where baby getting thrown out with the bathwater.

Patrick Moorhead: 

Yeah. All right. Speed round. Ever, ever pure formally pure storage reported revenue 1.6 billion adjusted EPS at 69 cents and a rebrand.

Daniel Newman: Solid. I mean, look, good quarter. I think storage space, flash, you know, as we see memory constraints, storage becomes a bigger, bigger narrative. I think the company's doing a lot of the right things. I do think it's sold. I don't remember now for sure. I actually think it's sold off the news. Didn't make a ton of sense to me, but, you know, this is a company that's kind of quiet, heads down, performing, and now it's performing under a new brand, new name. So good from Everpure. I think this theme storage has a tailwind.

Patrick Moorhead: It does. And their transformation into a data company makes sense. They're already a data company a lot more than a lot of other so-called storage companies.

Daniel Newman: 

Good ad, by the way.

Patrick Moorhead: 

Data, they really are pivoting. All right. HP actually had a good quarter. They had a beat on revenue. They had a beat on EPS, maintained full year EPS. It's set on the low end due to softening in the market and the stock reacted kind of as you expected, sold on 6%. And this is one of these just classic, man, like self-inflicted or market. It's you know, with a lot of the changes at HP, you know, the fact that they can crank out something like that, I think shows stability that even when you have, I would say, the top person, number two and number three, kind of exiting, you can still hold together. And I understand the timing. of of of earnings i'm not i'm not confused but it's still uh it's still good you know printing is uh is is is a drag right it was off two percent uh year over year hardware was down six percent uh memory cost is going to be the big challenge uh going in and, you know, HP doesn't have an enterprise platform to lean on or a huge software business that can absorb some of the challenges that you're going to see on the PC side.

Daniel Newman: 

And there you have it, folks. That was well said. The memory thing was my I'm obsessed with how these companies are going to manage this memory issue. They seem to have ideas. We're going to see it start showing up in the numbers. You know, HP is a got a great provenance. But the challenge is, of course, they're in some businesses right now that are genuinely in a tougher situation, they don't have that big data center access, but they do have the future of work play. And I think that's where they're going to need to lean in a lot is kind of a lot of what they've invested in with worker productivity and driving more productivity. Because as we know, as Block removes 40% of his workforce, the other 60 is going to have to be exponentially more productive. And we don't really hit on that today, but the sky is falling. We should end on that note. The sky is falling. Yeah.

Patrick Moorhead: 

Listen, I do like the HP strategy that pulls together kind of multiple solutions into experiences. It is just going to be tough because they don't have have a have a business to fall back on that cooks in the data center. So, hey, Daniel, great to see you, buddy. And I want to thank everybody for tuning in an extra long show. Hopefully, you actually was only 50 normally long show. Yeah.

Daniel Newman: 

You and I can even make a speed round and take a long time. I know.

Patrick Moorhead: 

Can't take it. Great seeing everybody. Hit that subscribe button. Tell your friends and your family and your pets about the show. Subscribe. Take care. Have a great week.

‍