Anthropic at $1.2 Trillion, AMD's Blowout Quarter, and the PE-Backed AI Enterprise Play | Ep. 304

The handpicked topics for this week are:

‍

1. Anthropic and OpenAI Launch PE-Backed Enterprise JVs on the Same Day — Both companies announced private equity joint ventures, with OpenAI backed by Bain, Brookfield, and Advent, and Anthropic partnering with Blackstone, Goldman Sachs, Apollo, and General Atlantic. Daniel's read is that this is fundamentally a distribution play, using private equity portfolio companies as a deployment channel for AI at scale. Pat sees it as the clearest admission yet that enterprise AI cannot be self-implemented at scale without specialized consulting support, and flags that mid-tier systems integrators (SIs) could get cut out of the middle. (The Decode)

‍

2. Anthropic Signs Massive Compute Deal with SpaceX Colossus — Anthropic urgently needed compute and SpaceX had 300 megawatts and 220,000 GPUs sitting at Colossus One in Memphis without enough business to fill them. Pat's take is blunt: this move is pragmatic. Anthropic needs it, xAI has it. Daniel adds that Dario himself said they planned for 10x growth and got 80x, and this deal is the fast backfill that reality demanded. The side note both hosts flag: Anthropic is running on H100s, H200s, and B200s, which puts the whole "Anthropic only runs on Trainium and TPUs" narrative to rest. (The Decode)

‍

3. Cerebris Files for a $3.5 Billion IPO at $26.6 Billion Valuation — This marks their second attempt at an IPO after pulling the first filing. The architecture is genuinely unique, a complete wafer with massive on-chip SRAM and interconnects built directly onto the wafer rather than copper or photonics. Pat calls it the first credible Western alternative for AI inference. Daniel's framing cuts through: you do not have to beat NVIDIA to sell right now. You just need to have availability. The more interesting headline, both hosts agree, is that Sam Altman and Greg Brockman are angel investors, which adds fuel to the ongoing OpenAI lawsuit. (The Decode)

‍

4. NVIDIA and Corning Announce $500 Million Optical Partnership — Three new US factories, co-packaged optics for Vera Rubin, and a supply chain strategy that mirrors what NVIDIA did with Coherent. Pat's context: this is vertical integration through investment rather than acquisition. Daniel's observation is that the pace of movement toward co-packaged optics is accelerating faster than anyone expected, and his “rule of and” applies here too. Copper is not going away. Optics are being added on top because the data volumes moving across these racks are outrunning what copper alone can handle. US manufacturing in North Carolina and Texas is a strategic bonus. (The Decode)

‍

5. IBM Think 2026: Day Zero, Sovereign Core, and the Quantum Plus AI Bet — Pat moderated on stage with CEO Arvind Krishna and calls this IBM's best showing in five years. Arvind opened with the AI divide, the gap between companies still running POCs and companies already in production, and framed where IBM sits as day zero, not because nothing has happened, but because enterprise AI deployment at scale is still so early. Daniel's biggest takeaways: watsonX Orchestrate updates, Sovereign Core going GA with policy at runtime, and the Confluent acquisition potentially being IBM's most important asset since Red Hat, given that 40% of Fortune 500 companies run on it and real-time streaming data is foundational to agentic systems. Both hosts land on quantum plus AI as IBM's next inflection moment. (The Decode)

‍

6. ServiceNow Knowledge 2026: Enterprise SaaS 2.0 is Emerging — Daniel got there on day three of the event and noted the conference was densely packed. His observation: enterprises have not gotten the memo from Wall Street that SaaS is supposedly dead. His emerging thesis is that middleware could make a comeback for AI, with companies needing a layer that lets agents work across any infrastructure, any app, and within the rules of their specific business. Pat agrees and adds that the growth question is about mix, not survival. (The Decode)

‍

7. The Flip: Is Anthropic at $1.2 Trillion the Most Important Company in Enterprise Tech? — Daniel took the affirmative citing that Claude Code is deeply entrenched in developer workflows. Anthropic went from $9 billion to $45 billion ARR in months. Every major hyperscaler is both a customer and an investor. The PE JVs are turning verticals into Anthropic engines. Dario said they planned for 10x and got 80x. Pat's counter: the enterprise trust gap is real after what Anthropic pulled on pricing and performance. Microsoft has 2 billion users across 365, Azure, and Copilot. NVIDIA is the infrastructure Anthropic runs on. And workforce replacement, which is how Anthropic extracts its terminal value, is not arriving as fast as the valuation suggests. In reality, both hosts admit their notes looked almost identical. (The Flip)

‍

8. AMD — Lisa Su guided AI data center growth up from 60% to 80%. With OpEx growing 83%, net income up 95%, free cash flow ripping, and CPUs growing at nearly 40% without price increases, Pat reads this as unit market share gains coming soon. Daniel's framing: AMD is now a two-headed juggernaut with CPUs and GPUs for the data center. And Helios has not even started shipping yet. Both hosts take a victory lap for previously calling this one. (Bulls and Bears)

‍

9. Palantir — Triple beat on revenue, EPS, and forward guidance. Rule of 40 at 145%. Government revenue up 84%, 47 deals over $10 million, and the largest guidance raise in the company's history. Daniel's take: Palantir is redefining the category entirely. It’s not a software company in the Salesforce or ServiceNow sense. It’s technology, plus ontology, plus people, deployed at the deepest layers inside governments and enterprises. Pat adds that the four deployed FTE model lets them stand up AIP POCs within a week, which is why they are winning business at this pace. (Bulls and Bears)

‍

10. ARM — AGI processor demand doubled from $1 billion to $2 billion within 45 days. Record revenue, strong pipeline, royalty growth at 21% for the full year. The stock ripped after hours, then sold the next day when management confirmed only enough supply for $1 billion of that $2 billion demand. Pat's read: 50% CPU market share with hyperscalers at the core level is the most underdiscussed signal on the call. Daniel adds that the worry about ARM competing with its own customer base in custom silicon has been quietly swept away by the sheer volume of compute demand. (Bulls and Bears)

‍

11. Supermicro — A board member allegedly used a hairdryer to remove labels from GPU boxes being shipped to China. Approximately 20% of their revenue has reportedly been illegally shipped to China. They beat on EPS and Q4 guide but missed Q3 revenue versus consensus. Stock still ripped 18%. Daniel's take: if you are selling picks and shovels during a gold rush and you are this messed up, he cannot imagine owning it with the overhang that is building. (Bulls and Bears)

‍

12. Lattice Semi and Coherent — Lattice revenue up 42%, back into growth, guiding to 50% year-on-year at midpoint. The AMI acquisition at $1.65 billion doubles their serviceable market from $6 billion to $12 billion and puts them inside every AI server on the planet at the BIOS and platform firmware layer. Pat calls the timing right: core financials crushing it, time to make a move. Coherent printed 21% year-on-year growth, 55% EPS growth, margins expanding, debt coming down, entered the S&P 500, and sits at the center of the co-packaged optics trend that is accelerating. Pat's choke point note: Indium phosphide capacity is the constraint. Six-inch fabs are doubling capacity in 2026, a quarter ahead of plan, and competitors are still ramping their transitions. (Bulls and Bears)

‍

Want the full breakdown from IBM Think and ServiceNow Knowledge, and check out our on-the-ground coverage linked in the show notes.

‍

Be part of our community. Hit that subscribe button and let us know what you want us to cover next week in the comments.

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 absolutely know what we're talking about. Welcome back to The Six Five Pod is episode 304. I am broadcasting from my office. Daniel is broadcasting from a prison cell in Lubbock, Texas. Daniel, how are you doing, my friend?

DANIEL NEWMAN: 

How did you know that I'm in prison?

PATRICK MOORHEAD: 

Should be a meme. Nondescript walls. Should be a meme. Yeah. I mean, the curtains, you did make your bed.

DANIEL NEWMAN:

 I do iron, iron beds, mattress. The mattress is, is, is this thick, you know, this thick and, you know, one pillow, one pillow. Exactly.

PATRICK MOORHEAD:

 So, Hey, you wouldn't want anybody to steal the pillow. You wouldn't want anybody to sell it in the prison. I mean, You know, let's get real. That was a busy week. Um, I was Daniel and I were both at IBM think 2026 in Austin and Daniel is an overachiever and he made it over to service. Now we'll talk about a lot, a lot about IBM think, but it was nice getting up on stage with Arvind Krishna. That was a, a fun, fun thing to do. Yeah.

DANIEL NEWMAN: 

I just, I just want to give a shout out to my awesome daughter who's graduating with her doctorate, occupational therapy here at Texas Tech, doing the big things, big girl things. So, you know, Arthur, I mean you on stage with the CEO of a Fortune 100 company, that doesn't seem like that big of a deal to me when I've got Kids graduating and I think you've got a kid graduating pretty soon too. And by the way, I want to point out both of our kids, not only graduated, but are employed coming into their graduate coming into their graduations, which I don't know. But based on what I'm reading and hearing, that seems like a pretty big accomplishment these days, Pat. So we should.

PATRICK MOORHEAD: 

Yeah, it is. It is. Yeah. And my son, you know, will continue the streak of engineers in the family. I was the family disappointment. Not only was I not an engineer, but I didn't get a secondary degree. So anyways, he's carrying that forward. I mean, to have a doctor in the family, Daniel, like, like, is that a, is that a first for the, for the Newman's?

DANIEL NEWMAN:

 In the immediate family, we've got a bunch of lawyers and doctors in the extended family. But yeah, we're a bunch of dummies over on our side. I mean, I've got the I've got a master's. My sister has a master's plus 60, but didn't actually finish her doctorate. So, you know, every generation, we're getting a little smarter, I guess you could say.

PATRICK MOORHEAD: 

That's amazing. Yeah. I, I, I almost graduated high school. So, but now, Hey, we got a great show for you. Um, so much to talk about. We've got opening on Anthropic, uh, doing, uh, uh, getting in the middle of PE. Uh, I'm sure that's, that would, that, that'll end well. Anthropic signing big deals with, uh, SpaceX. Rebirth's filing for an IPO. We've got, uh, Whoa, editor's gonna have to take that out. We got Nvidia and Corning doing an alliance. We've got IBM Think, Boston. We're gonna not tell you what the flip's gonna be until we get there, but a lot of bulls and bears. We've got AMD, we've got Palantir, we've got Arm, we've got Supermicro, we've got Lattice Semi. It's gonna be an exciting thing. So let's dive in to the D code. All right, Anthropic and OpenAI launching PE-backed enterprise services, JVs, on the same day. What the heck is going on here, Daniel?

DANIEL NEWMAN: Well, I mean, this was an interesting one to me, Pat. You know, we basically on the, what was it, on Monday this week, You know, they created like these JV companies, many investors, some of the big ones, Bain, Brookfield. I don't know. I didn't memorize all of these. Advent, you know, committing a few billion of equity and a significant return to these companies. And by the way, then Anthropic did one and theirs was with Blackstone and Goldman Sachs and a handful of others, including Apollo, General Atlantic, you know, and If I'm reading this right, this has a lot to do with kind of that Palantir model of how they are using ontology and AI and then sort of adding this value-added services. And what they're basically looking at is the distribution, the portfolio of the companies across all these JVs, sorry, across all these private equity back and take the AI platforms and create this massive distribution. of these solutions. It also to me is kind of interesting, because it says something to me about the ability for AI to be successful inside of the enterprise. And the expectation is, you know, and I think you've seen this in generations, you know, we talk a lot about, and by the way, it's a little tangential, but like, we went from an era where enterprises built their own software, And then SAS came out, because really, enterprises don't want to be in the software business. They wanted to have software to help them grow their core businesses. And then we got to a point of such efficiency with SAS software, and now AI is being introduced. And now, with AI, realizing you can't just off-the-shelf implement AI to the scale, just like in the early days you couldn't just off-the-shelf implement CRM or ERP software to scale. And so, I think what they're kind of looking at is what the business model in the future is going to look like, which is going to be some hybridized, not entropic taking over every type of software and then doing everything for every business across every category, but it's going to take Companies like Anthropic with companies like these PE-backed firms basically taking the software, putting it together, packaging it, and making it high value inside of enterprises so that it can deliver the kind of productivity and efficiency gains that were sort of being promised in the disintermediation conversations that continue to go on. Um, you know, I still feel like these things are a little loosey goosey. Don't fully pretend to understand exactly what's going on here. But the biggest thing that I do think is happening is there's a clear admission by these anthropic AI that their software can't just be, you know, it's the whole, you can't just vibe code every application and put it into your organization without the support of, you know, highly capable specialized consulting teams. Maybe it's an SI, maybe it's one of these PE back firms. To be able to collaborate to make software high value and highly available. So anyways, you know, I think that's kind of what's going on here. I'd love to get your take two on it because as I read it. You know, it feels to me like this is the admission from these companies that there's a lot of work to be done to really start to be able to extract value that meets sort of the promise that Dario and Sam are telling the market is going to be available to them.

PATRICK MOORHEAD: 

Yeah, so a couple thoughts. So first and foremost, you go to PE when you want to, when you can't get money from venture capitalists, you aren't a public company, and you can't get a loan, right? So it's kind of like the the end of the line for this. And I think that that that speaks volume. The second part is this, I see this primarily as a distribution deal, right? Look at all the the companies that are their PE backed. The ability to bring AI into those PE based companies in an accelerated fashion seems like a good thing to do. So, you know, there was some conversation about the losers being Accenture or Deloitte or folks like this. I don't know if Accenture and Deloitte do most of their projects with the smaller PE-backed companies. And listen, PE-backed companies aren't small. They're just not Fortune 500 companies that most of the larger SIs work with. So, it's a $200 billion market. And I do think kind of mid-tier SIs have the opportunity to get cut.

DANIEL NEWMAN: 

Is your read, Pat, that this is like taking all these businesses and just putting an anthropic wrapper around them and then promising them? 17.5% returns on their money for doing this?

PATRICK MOORHEAD: 

Well, if you look at the PE companies, I think each of them will probably create their own platform and then they will offer that platform to, they'll create some sort of standardization. You know, PE companies, they come in, they try to get you on a uniform accounting system, ERP system, um, and, and uniform as much as they can. And the thesis being it's going to cut out cost, right? that's, you know, PE is all about driving costs down, driving the value up. And sometimes, you know, they even connect their PE companies together, they merge them as well. And I see this as operationally trying to jumpstart the folks who might not be jumpstarted. And, you know, quite frankly, you know, you look at why, you know, PE typically will buy companies that are distressed That's that's one or somewhere where they have a fee. The company cannot get financing any other way. So, yeah, we will. We will see. All right. Next topic, Anthropic and SpaceX signing massive. compute deal. Anthropic has just, you know, you just go to their uptime page and you can see they are not able, they do not have enough compute to do what they need to do. So it looks like a 300 megawatt, 220,000 GPU Colossus One down there in Memphis. So, You know, I think this is very straightforward. I mean, Anthropic is getting two things going on. Anthropic is getting access to compute that's available now. And it looks like SpaceX, which basically is XAI, doesn't have enough business to be able to, you know, occupy all those all those GPUs. So yeah, I think it's just a very pragmatic type of deal. And you've got Anthropic that needs it, and you've got XAI who has it. So I don't know if there's much news beyond that. I know people tried to put orbital compute in there, but I just think that to me just looks like a strategic tack on the very end. Right. Maybe it was a OK, lower your prices, but you have to support orbital orbital type of operations. Dan, is there anything more to this?

DANIEL NEWMAN: 

Yeah, I mean, the only thing I'll say is this feels like a big middle finger to to to Sam Altman. in the middle of this trial and everything that's going on. XAI, one, because it's kind of almost self-defeating to some extent that they had to admit that all this compute they've built, they don't need, that they have all this extra capacity for others. Which is saying, is Grok not doing that well? Which is a bit of an indictment heading into their IPO, but at the same time, now he just filled it up with a shitload of revenue. that should help further justify that expected massive IPO evaluation. But also just knowing that Dario, he came out this week talking about how they planned for 10x, they got 80x growth. They just were unprepared for the amount of compute required and to be able to get like a quick backfill like that, you know, people always want to make, you know, so I said a little bit like with this lawsuit is just wild. I mean, even when we get to the next topic on cerebris, we're going to realize there's another tie-in to the lawsuit there. I mean, this thing just, this thing has tentacles to it. But then again, you know, the second that Anthropic went over there, the bubble bears wanted to be like, well, this is bad for nebulous and this is bad for core weave. It's like, You know, again, this is a relatively small amount of additional quickly available compute. You're talking about a handful of hundreds of megawatts when we're talking about building hundreds of gigawatts. You know, so it's just interesting how people always want to do that. But overall, I think it was a quick and good fit for Dario. I think it was a interesting stopgap for some maybe overextended investment by Musk. And it's definitely Musk would rather be paired up with Anthropic than do anything to help open AI at this point. So off we go.

PATRICK MOORHEAD: 

Yeah, I hope Anthropic gets better. What I was struck at too, Daniel, was, you know, they're going to be ready almost immediately. And what gets really scratching my head here is, well, wait a second. I thought Anthropic only ran on TPUs, sorry, Tranium, and then TPUs. And here we have this, you know- Welcome to the party. Yeah, exactly. So, you know, this is a clear beneficiary because Kloss says H100s, H200s, and I think B200s or B300s. So anyways, let's go. I hope Anthropic stops sucking in a couple of weeks. We'll see. You're a Codex guy now, right? You're a Codex guy now. You know, as soon as I literally have a free moment, I mean, it's crazy how many developers that I know, even my son in a week pivoted over to Codex and he's just talking about how amazing it is. We'll see. Hey, here's a more straightforward story. Cerebris filed for a $3.5 billion IPO at $26.6 billion of valuation. This was the second go at it. They filed once and then they pulled it back.

DANIEL NEWMAN: 

I mean, isn't this just a major inference chip with a bunch of SRAM on it? A different, you know, so so the IPO is interesting. Again, they've got, what, about $20 billion of commits. They actually got money from OpenAI. And interestingly enough, probably the most interesting thing about this story is the whole Sam Altman and Greg Brockman are actually angels in this company, again, tied back to the lawsuit. And people are questioning their lack of disclosure about all this when they did the deal. So that's probably the most interesting headline about this overall deal. You know, I think it's timely. I mean, we've had this incredible melt up on infrastructure and, you know, if if cerebris can deploy and show that they are. you know, a solid alternative. As I keep saying, you don't even have to be the best. You don't even have to be top five. You just have to have availability right now and you'll be able to move it. Um, I think they're doing some interesting things, uh, from a technology standpoint, but like some of the stuff I read about, you know, cerebrus going after and competing with NVIDIA. I just, all these headlines to me are just so nonsensical. It's like, you know, every chip is not going toe to toe with NVIDIA. It's just, this is a different approach to adding scale imprints. Um, heavy, heavy memory on, on chip, uh, versus, you know, going out to HBM. I, but the deal itself pretty straightforward.

PATRICK MOORHEAD: 

Yeah, so the architecture is amazing. I mean, it's basically a complete wafer. And Dan, you were right, a lot of SRAM, I feel like it's a cluster on a wafer, right? And the interconnects are literally on the wafer as opposed to being copper or photonics. Um, you know, the, the interesting part is that it is a little limited based on the size of the model that you can, you can throw on there. It used to be a training solution, but now it's an inference solution. And that's okay because that's where. the growth is. One of the investors you didn't mention, Daniel, was AMD, put an investment into Cerebris back in February. And that was at a $23 billion valuation. A couple of things, CEO says he's not going to sell. you know, that could be a credible signal is retaining 10.3 million shares. And I think that that, you know, we don't know what he has sold so far. But yeah, Andrew Feldman, sorry, I didn't mention his name. So it is interesting, right? This is the first credible kind of Western alternative for AI inference. Rock was in there, but they got acquired or acquired or licensed, whatever. So I'm really, really interested to see. I do think we need some heterogeneity out there even more. And this architecture is truly unique. Okay, guys, my notes start to look like your notes. Yeah, yeah, I'm doing I'm doing a world of models now, which uses three different models and then and then kind of picks the picks the best one. All right, next topic. NVIDIA and Corning announced an optical partnership, $500 million investment, three new US factories and co-packaged optics for Vera Rubin. So this is the same play with vertical integration that NVIDIA has taken. Essentially, you instead of buying the company outright, you make investments and then you get supply chain. You get some supply chain guarantees, very similar to what we saw with Coherent. So this is all about co-packaged optics, right? Next step. step change and infrastructure doesn't mean copper is going away. It just means you're going to see more co-packaged optics. There's three or three different ways you can do, uh, you can do CPO that, uh, have different, uh, different puts and takes different risks, uh, different, uh, different benefit. Um, I think that, you know, at, at GTC, uh, when NVIDIA first came out with this, NVIDIA did a pretty good job, uh, talking about articulating the benefits, but. Copper, you know, made a strong push back. You know, it's just like us saying that spinning discs would go away. And, you know, most storage is done on spinning discs. Copper, you know, you can use special cables, you can use special algorithms across it to keep it going. US manufacturing is a nice thing. New facilities in North Carolina and Texas, getting back to a more reliant, lower risk supply chain. You might have to pay more, but you got to factor in not getting it if it's coming in from Taiwan. or China. So supplies to strategy and Corning just, you know, shareholders woke up to a completely different type of stock.

DANIEL NEWMAN: Yeah, I mean, for me, all this was was a really, really strong signal of how fast we are trying to move towards co-packaged optics. To your point, and I think to many points, this is such, again, not the optics are about to replace all things copper, but there is a need and desire to move to optics, you know, based on the amount of data that we want to be able to transmit across these, up these racks and across these racks, and then across these data centers. But I, you know, I think just the recent coherent announcement, the momentum announcement, this announcement, like there is clearly more pace and velocity and desire to get to this outcome. Because, you know, we kind of keep hearing like it's three or four years old. It's like I just feel like everything's accelerating. I feel like the time-space continuum that we live our life in is like, you know, is moving at like a 1.5X. So, you know, when we think something's going to be three years out, you know, remember we said it'd be a trillion dollars if semiconductors wouldn't happen or if infrastructure would never happen, probably till the end of the decade. Now we're going to get a trillion dollars of infrastructure this year. And yes, some of it is based on pricing and inelasticity of the economics. So I guess my point is just we're moving very, we're moving very, very fast. towards a higher propensity or a higher portion of the transmit being done optically. It's just moving very quickly. But again, it's not going to displace copper people, so don't sell all your credo. It's just, it's going to be in addition to, we just need to move more data against more compute.

PATRICK MOORHEAD: 

You know, it frustrates me that people don't understand called the law of and the law of and yeah, like Patrick's law of everybody's talking to everybody.

DANIEL NEWMAN:

 It's not your law.

PATRICK MOORHEAD: 

Exactly. Like, every every hyperscaler is going to talk to every chip maker. Every chip maker is going to talk to every foundry. It's just like mainframes are supposed to die. And, you know, Z was up on IBM 51% last quarter, you know, so and what happens is, you know, copper is cheaper. It's more reliable than photonics. So the industry's, you know, figuring out and once we get to scale, and believe it or not, Optics used to be kind of like quantum is today. You know, it's like five years away every year. But it's easier. Five years away. Exactly. All right, let's wrap up the Decode. Daniel, what did you learn at IBM Think?

DANIEL NEWMAN: 

Well, first of all, I learned that Pat Moorhead is a really good moderator. I mean, I just heard terrific things about Pat and Arvind Krishna up on the big stage. I was just a fan in the stands, clapping my hands, cheering, singing, go, go, go. But, you know, I thought it was a comprehensive, you know, the keynote he gave was what I thought was really good. He said something that I want to just reiterate. You know, it is so early right now. Day zero is kind of what he called it. And his point about AI is it's not day zero because nothing's happened. It's day zero because we're so early in terms of like taking these sort of capabilities and bringing them into enterprise scale. You know, I thought they'd made some good updates to their next gen Watson X orchestrate. You know, they were able to communicate. I think what they're doing with sovereign core and bringing that GA, I think that's a really important You know, again, the rules and rails of business and infrastructure is going to be having a policy, you know, they call it a policy at runtime for sovereign AI. So I thought that was a really, really good one. And they brought some good partners in there. I think, you know, just some of the headliners, AMD, Dell, Elastic, Intel, Mistral, Mongo, Palo Alto, all part of their sovereign core. You know, I think also Krishna was very focused on kind of the quantum eras here. He continues to, you know, that is their AI by the way. What they're doing with AI is basically building an enterprise-scale deliverable of AI. Smaller models, lower open source, less infrastructure, sovereign runtime, all those things that we just mentioned. But what they're not trying to do is build chips at scale. They do some for themselves. What they're not trying to do is build large-language models and frontier models in the lab. They're basically saying, we will be the partner to help you. We talked about that FDE model, and they are kind of an in-a-box FDE model for delivering AI into the enterprise. So, you know, that was my big one, but then like on quantum, I think that is their opportunity to be their AI inflection moment, you know. They're building some of the most sophisticated quantum computers. They have the runtime language in Qiskit that's basically being used for the applications and development. And it's not about quantum or AI. This goes back to Daniel Newman's rule of and. And it's quantum plus AI. And that's where we're going to start to see really interesting developments. I had a chance to talk to Cleveland Clinic there on R65. There'll be an episode out. I'll look into some of the things they're doing. And by the way, Pat, you and I being workout guys, you know, seeing the modeling of molecules and proteins is actually very, very interesting. A lot of stuff we think about with peptides and with, you know, different, you know, things, you know, you may or may not inject in yourself. Like, all the molecular development of these compounds and these drugs that are used to create superhumans or to heal things or to improve conditions or to deliver longevity, you know, Quantum plus AI is going to be really interesting. So I think that's a that's a path for IBM's next wave of growth. And I was good that he leaned into that.

PATRICK MOORHEAD: 

Yeah, I I think this was the best showing that IBM has had over the last five years, and I think the timing is good. You know, with all this complexity and jargon and tech and stuff like that, I think IBM came in with a very good senior level executive rollout. Um, like Arvin's first 15 minutes of his keynote were, were, were really good and even better than our discussion on stage the day before. That's a joke folks. Yeah. He, he came out with the AI divide, uh, which Daniel, I know, I know that you, you had mentioned, it's kind of the difference between, you know, these are my words, not theirs, the winners and the losers, right. The folks that are still doing POCs and the ones that are just, uh, rocking and then He went through an AI operating model that had different elements of it that just made sense. And then the payoff was, what does IBM do? They do AI, they do hybrid, and they do quantum. Danny did a good job on the quantum stuff. And I think one of the differentiators when I step back here is, first of all, you have to look at IBM inside of the regulated industry space, right? It's where, you know, it is their North Star. So whether it's governments, healthcare, energy, defense, all these different regulated industries out there that that's what they're focused on. One differentiator is cross cloud, right? If you look at what they're what they are delivering at the end of the day, They are software services that arbitrate between wherever the application is, whether it's on-prem, whether it's Colo, whether it's public cloud, whether it's SaaS, pulling that all together. I think Sovereign Core was a super interesting offering. I don't know if I like the name yet. Maybe it will will grow on me, because it's not necessarily sovereign only, but it could be good for sovereign. And sovereign is just not a, oh, I'm a country and I want my own sovereign AI. capabilities. It's also those regulated industries inside of a said country that isn't afraid of the United States unplugging or snooping on the data that's out there. And it's full stack. I think the final comment I want to make is, well, first of all, let's add confluent. I think this was You know this could confluent could be potentially the most important assets since Red Hat. Right. Forty percent of Fortune 500 and in real time streaming data. And you know I think the way to look at this is agents. you know, if you're, you know, gosh, the basic of basic things, right, you run a query, you've got 10 agents that are that are out there looking for stuff. And then you take that to the next, next layer, which is a system of action, where they're actually doing something automatically for you. That is all data that's happening in real time, new data is is coming in. So We will see what happens there. Cyberincore is a full stack, right? All the way from models to chatbots, to developer programs, to the data layers. You can pick your own models. Governance is integrated in there. In fact, I think they had 20 templates. of things to be mindful of. So yeah, I'm impressed. So hey, Daniel, just to wrap up, I know you were at ServiceNow. Just some quick thoughts from you on that.

DANIEL NEWMAN: 

Yeah, at ServiceNow, now some interesting things around agentic model testing. They are partnering with NVIDIA to do some work on that. optimize. I don't know why that word didn't come to me. You know, I also think, you know, it continues to, I guess, stepping out. Probably the most interesting observation I had there besides some of the announcements was just how crazy busy it was. You know, I don't think I've seen a bigger disconnect between Wall Street and Main Street than when I go to service now. And I got there, Pat, I didn't get there until the third day. And one thing you and I can tell people about usually when there's an event that's three or four days long, first day, maybe second day busy, and then you start to see a pretty significant drop off by third and fourth day. I literally got there on the third day, midday. And Pat, you were at the Venetian. I was at the Venetian. You know that hallway when you go past the Starbucks. You literally could not walk.

PATRICK MOORHEAD: 

Yeah.

DANIEL NEWMAN: 

There were so many people in there and it was so crowded. And all I can say is I said to myself, I said, you know, whatever's going on. The enterprise has not gotten the memo yet from Wall Street that enterprise software is apparently supposed to die. The investment of people and their time to be there. And again, I still continue to think there's going to need to be a central platform where companies build, we just talked about sovereignty, but kind of like the rules and rails of your business's runtime and workflows. And I do think the workflow is an interesting potential winning point. Meaning when a company says, do I build it with the infrastructure layer? Do I build it the platform layer? Do I build it the SaaS layer? I think you want to build it. I think, Pat, you know how everything that's old is new? I kind of think middleware could end up, and that's why I like IBM too, but I kind of think middleware could make a comeback for AI because like, Ultimately, you want to be able to build across any infrastructure, you want agents that work across any app, and you want all these things to make sure that they work within the rules of your business. So I think the demise of enterprise software might be as absurd in a few years as the supposed demise of the mainframe. It doesn't mean that the rewriting is not real. But the actual use in the proposed, you know, I think that's totally being missed.

PATRICK MOORHEAD: 

Yeah, I need to check out what future equity is saying about this, but I'm with you on the I'm with you on the I mean, it's not going away. It's just a growth. that I look at, right? You're gonna have some people that are gonna not use the higher margin AI features and data features that come with it. And you'll have others who will lean in on it. And I think it's the mix of what it looks like going into the future. Okay, good stuff. On the decode now we want to go into the flip a lot of good great topics we could have chosen from here but a pretty provocative one we love talking about anthropic and I've got a chance to check out my Forbes piece that I put on there on determining whether enterprises can trust. Anthropic moving forward. So here's the topic. Anthropic at $1.2 trillion is the most important company in enterprise tech, more important than NVIDIA, Microsoft and OpenAI. Let's jump in and see who gets the baton. Yeah. All right, Daniel. Big Dan, big head Daniel with the beard. 

DANIEL NEWMAN:

Fat Dan. Fat Dan. Everything that…  Fat Dan.

PATRICK MOORHEAD: 

That's the four. So come on, amaze me.

DANIEL NEWMAN: 

 I mean, look, Pat, every once in a while, the center of gravity in enterprise tech moves. A few years ago, it moved to Nvidia. Maybe last year you could say it moved to Google, but this year it's clearly moved to Anthropic. You know, no company goes from nine to 30 to, over 45, I think $45 billion now of ARR inside of just a couple of months if you're not the most important company in the world. And by the way, what technology is every enterprise that is by coding and building applications building on? Sure, Codex is cool this week, but Cloud Code is deeply entrenched now in every workflow on the planet. And by the way, All these companies now are basically making Anthropic their big customers and big investment point. Amazon, pouring money into them and also, you know, leveraging their compute at massive scale because they want to have the biggest and most important tech company as a customer and they want to invest in them. Google, doing the same. $40 billion in. What's that? A $200 billion committed spend back. They have a multi-hyperscaler anchor in Anthropic. This is something that OpenAI is just getting around to now. Oh, and by the way, Anthropic is now running on SpaceX Colossus. So they are everywhere and anywhere that you need to be. You know, I also think that the capital is voting here. You know, you have a one point two trillion dollar valuation. It was a trillion a month ago. It was nine hundred billion the month before. And by the way, remember, they raised it like three hundred billion less than a quarter ago. So they triple quadruple revenue. Dario himself said it. what other business on the planet right now is growing at 80 times? Yeah. So Dario said, yeah, we were planning for 10 times and yet we got 80. And now on top of that, you know, basically, Anthropic is out there partnering with the JVs, doing these JVs, doing it with the PE companies, because every one of these companies now want to turn themselves into Anthropic engines of various verticals, whether it's manufacturing, whether it's healthcare, whether it's financial services, they want their entire business to run on the entropic platform. There's never been anything like it. There's never been a technology that disintermediates more businesses in real time, and there's never been a voting with people's wallets that's been more considerable than this. You see all that money being sucked out of market caps of software companies, it's being sucked out of market caps of of data companies, it's all going straight to Anthropic and give it more time. And as features get released week after week, we will see that this 1.2 will quickly turn to 1.5, will turn to 2. And before you blink, Anthropic will not only be the most important enterprise company in the world, it will also be the most valuable company in the world.

PATRICK MOORHEAD: 

Yeah, so let's look at what importance and relevance actually means. And I know valuation is a great marketing number of what something could be. And it's the terminal value of all future cash flows. But let's just put that aside. Trust at the enterprise, trust gap at the enterprise layer is real. And quite frankly, what they pulled with performance, price, and quality half-truths, you know, can enterprises truly trust Anthropic? You know, maybe as a pass-through, but certainly not directly. And so let's look at Microsoft. I mean, Microsoft has the enterprise distribution. If you add up Microsoft 365, Azure, Copilot, you're looking at over 2 billion users out there. That's something that I would consider absolute power. If you look at NVIDIA, I think the market share itself is a great indication that even though Anthropica is doing a lot with AWS and a lot with Google in the future on TPUs, I think this colossus deal with SpaceX, XAI clearly shows that two things that Anthropic needs NVIDIA, but also that Anthropic runs on. Runs on NVIDIA. You know, what is it going to take, like a week or something? They said that they would be getting this online. I don't think we know exactly how many, um, you know, the percentage that Anthropic runs on TPU, uh, tranium or, or GPU, but it's certainly going to increase in the future if nothing else, just for capacity, uh, just for capacity and, and availability. Um, so I think the, um, You know, the final thought here is that, you know, it's really going to be about workplace replacement. And I think you and I have even, you know, you've even said that, hey, these jobs are just not going away. Kind of Dario, you know, what he talks about is it's kind of nonsense. Not that there won't be any implications to it, but I just, don't think that it's going to be as big. And workforce replacement is how Anthropic gets its value and how it shows value. So I rest my case for this simulated debate.

DANIEL NEWMAN: 

I don't know. I don't even think they're top five right now. They're probably top one most discussed company. Yeah. But I mean, Dario moves market cap through fear porn. I like Anthropic, but I mean, he's using fear porn. I mean, he's he's he's executing perfectly by every time he says jobs go away, more market cap Anthropic. Every time he says that industries will be disrupted, more market cap to he's just taking market cap without actually having taken revenue because it's actually additive. The gross product's growing. The companies that are that are supposedly taking market cap are actually getting bigger now, but are worth less.

PATRICK MOORHEAD: 

Yeah, we haven't seen the best from OpenAI yet, and OpenAI has the capacity. I'm guessing that Google will drop a new model at IO that's coming up very, very shortly. And we'll see kind of a resurgence from that. I think in the end, I just don't see models as a differentiator long-term and as a moat just yet, I think. to capture all that revenue, all these model companies are going to have to go upstream and displace the aggregators. But if you look at the power that the people with the infrastructure have, the Googles, the Microsofts, folks like that, I think it's going to be a tough one. So hey, let's go into the final, final segment of the show, Daniel's favorite segment, Bulls and Bears. Let's dive in. All right, Daniel. You get the choice one here. AMD, first quarter, stock up 20%. What the heck is going on here?

DANIEL NEWMAN: 

Yeah, I mean, sorry, I got to take all the good stuff here. But, you know, what did Lisa say AI data center, she expected to start growing at 60%. And now that number has been revised up to 80%. And by the way, what a tailwind. I mean, you you already had this sort of It was it was somewhat ambiguous how big the GP business could be for them because she's been very coy in terms of making claims and calls at this. But as we at Futurum have sort of forecasted, you know, a 10 plus percent market share opportunity, this is going to quickly cross into a business that could be worth one hundred billion dollars by the end of the decade in revenue, not worth in revenue. But now what you didn't have, and I don't think any of us really put on our bingo card heading into 2026, would be that all the work that the company's done to create market dominance in its data center CPU space, in the Epic business space, would become so invaluable. So now this is a two-headed juggernaut that basically has CPUs for the data center, GPUs for the data center. And I mean, they're growing from a much larger base now, almost at a 40% clip, and that includes they're kind of, yeah, client market and embedded markets, which have been far from exciting. And you know what, it doesn't matter anymore, because the OPEX is growing 83%. Net income is up 95%. Free cash flow is ripping. Um, you know, the company, uh, and margins are staying strong. I mean, they don't have Nvidia margins because the client business and embedded business, but as the data center continues to become more and more a disproportionate part of the revenue stream, those numbers will start to be more and more beautiful as well. And again, this is going to go back to another Pat and Dan's law of and. So when we say nice things about AMD, it does not mean we're saying NVIDIA is not good. It doesn't mean we don't think the ARM AGI is going to work. And heck, Pat, even you and I will probably, we've even acknowledged that Qualcomm AI250s will sell. I mean, right now, if it can be built, it will be sold. That's a new law, you know. But by the way, I just want to quick shout. Did you see Qualcomm went from like 130 to like 200 bucks in like a week?

PATRICK MOORHEAD: 

No, I know. I know. It's amazing.

DANIEL NEWMAN: 

But anyways, I mean, and here's the thing, Pat, with AMD is Helios hasn't really even started shipping yet. So all the rack scale stuff, we're still really talking about shipments of GPUs and CPUs individually into Oracle, where they got a big deal into OpenAI, where they have that big six gigawatt deal. You know they're selling lots of CPU, data center compute, and doing business with Meta, with Microsoft, and others. So, you know, the people that counted AMD out were wrong once again. Pat, the people that called that AMD could succeed even with NVIDIA's dominance, you and I, it's kind of like a victory lap moment. We've called it, it's arrived and it will continue to be, it's going to only continue to grow from here.

PATRICK MOORHEAD: 

Yeah, the opportunity to do Power Lunch the day after earnings came out and they put me on the spot, basically said, hey, is the market appropriately reacting? I think at the time it was up 16%. the, uh, the day after. And, you know, I'm not an equities analyst and I'm not certified, but it's like, I mean, to me, it was simple, right. Based on a business fundamentals. And the short answer is, is yes. Uh, and then they asked me about Lisa Sue. Um, and basically if you look at everything Lisa has said, historically, um, very little hasn't become true. Or the other way to look at it is most things, uh, that she has said are true. And it's, it's very, um, She's very conservative, and I know some particular retail investors get frustrated at it, but she just takes a very different tack. And by the way, CPUs are cool and sexy and they doubled their TAM. on and they crushed it. And by the way, they crushed it on CPUs without raising prices. Okay. Which was different from what happened with Intel. So to me, the only way that I can look at that is they're going to gain a unit market share as well. The one thing that we haven't covered yet is Lisa had talked about new MI 450 deals. Expansion of current ones and then new ones out there. So going through my head, I'm wondering, is this gonna be Google? Is this gonna be AWS? Who is gonna step up here? So yeah, I mean, great, uh, great earnings and, and very conservative company. Can't wait to see, uh, uh, what they kind of can't get what they do. And by the way, you're, you're, you know, our law of hands, you know, everybody wants to frame this as, okay, Andy's good in video. This must be really bad for them. Uh, there is a lot of compute to go around and be is sold out. And, um, you know, people are going to, you know, people are going to want to, I mean, in the Nvidia multiple is, is kind of ridiculous at this point. I know we're not talking about low. Yeah, it's low, right? It's ridiculously low. And they asked me, you know, on Power Lunch, you know, what the heck's going on here? We've never, we've never dealt with a company that quite frankly, if you just do an average PE for this sector, it would be nine over $9 trillion. Okay. So, um, which is bigger than, you know, the GDP of India and France, uh, put together, right. Doesn't know how to deal with that right now.

DANIEL NEWMAN: 

There's only so much dollars to go around. And so a dollar in somewhere has to, you know, liquidity is not endless. Um, but yeah, it's interesting, Pat, you mentioned that cause I, I did three different TV spots this week and I got asked on all of them. Is this a like what to what detriment is AMD success to Nvidia? Everyone asks that and they continue to ask that. And it's like nobody's realizing how big the TAM is. I mean, yes, if you are purely valuating a company on its terminal market share. meaning that it has a continuous market share of 96% of the data center GPUs in perpetuity. But when the difference was they had 96% when it was a few hundred billion dollar market. It's gonna be a few trillion dollar market and they're gonna probably still have 75 or 80% of that. And so it's just, it's sort of a silly question, but it does get headlines. And so I guess, You know, I actually said that I think I actually got myself in trouble with one because I actually I said I keep getting asked the same silly question. I like that. I like. Yeah. Well, we'll see if I get invited back. Maybe I won't. Yeah.

PATRICK MOORHEAD: 

Hey, let's jump into Palantir. The SaaSpocalypse narrative hit Palantir before this, but man, did they come off and just fricking crush it. Basically, they had a super, super triple beat, right? They beat on revenue, they beat on EPS, they beat on forward guidance and rule of 40. I mean, they're at 145% of that. And I think, I think everybody needs to look at that. It is funny. I can't believe the industry was debating downstream AI demand. Not that there was no risk, but here we have a quote-unquote software company who is just absolutely crushing it with AI. They have a secret a weapon here with four deployed engineers, right? FTEs. So to drive AIP in there, they essentially go in there and create, you know, create the POC for them in like a week. And then this is why they're knocking back business as well. Two types of sales motions and market motions in there. We've got the government right up 84%, 47 deals at over $10 million. But I think what surprised everybody was the amount of commercial business and the amount of growth that they're doing. Final comment, I didn't hit this earlier. Their, their raise was the largest guidance raise in company history out there. That was just, that was crazy, right? They raised, they raised it by, you know, 10% over expectations, which is, is pretty awesome.

DANIEL NEWMAN: 

Yeah. You know, Palantir is the most interesting AI, uh, consulting platform on the planet. It's treated like a hyper-growth enterprise technology company, but they have a massive consulting business. And what they've basically been able to do is wrap a ontology AI platform with all the things that an Accenture does, put it all under one roof, and do it at immense scale with a tiny fraction of the headcount. And they're growing at just this remarkable pace. but they're not necessarily attached to the same kind of pure license seat ARR kind of that all the other companies are. So they're sort of redefining the category, and I think they're going to sort of become the model by which AI companies are evaluated in the post-SAS ARR era. And so they continue to be, I call them an end of one. They continue to be completely unique. But don't confuse them as a software company in the vein of a Salesforce or a ServiceNow or a Workday. They do something completely different. They do a lot of bespoke work. They do a lot of work at the most complicated and deepest layers inside of governments and inside of enterprises that no other software can necessarily do. But that's not all done because just the pure technological capabilities, it's the technology, the ontology and the people that enables them to do that.

PATRICK MOORHEAD: 

Yeah, good, good adders here, Daniel. Hey, let's jump into another chip company and call them a chip company. Now. That is that is ARM. They doubled their AGI processor for that, I think. within like a month, but their stock did not do good things. What's going on?

DANIEL NEWMAN: 

Well, it had a massive run-up into earnings, as did so many other names. I mean, they saw the demand for that new AGI chip, and this near-term demand doubled from $1 to $2 billion. you know, signaling that there's some very large hyperscalers, maybe even some that have their own chip that are going to look to buy this as well. Because by the way, you know, they kind of know it's going to work on all the things they're going to do so they can get additional compute capacity. I actually think the biggest win, you know, because they had record revenue. They have, you know, they had really strong, you know, pipeline. They did get a great bump after hour. Then the next day it kind of sold. And, you know, the royalty growth was maybe a little slower. Um, but I mean, it's still Um, you know, and I think that, uh, the overall space for just CPU demand, I think it couldn't have come at a better time than they almost nailed it at the time they came to market. And I think there was some worry when they came out, like, was it going to be like co-op partition with their current customer base? And I think in the era now where we have as much demand for CPUs as we do, I almost think that that got immediately and gently swept away. Like it's not really an issue at this point because frankly, we need more compute. and them being able to offer a sort of merchant arm based at scale probably will augment nicely to all their custom arm based partners.

PATRICK MOORHEAD: 

Yeah. You forgot to say a rocket ship after we need more compute, Daniel. I'm very disappointed in you. Your audience needs that there.

DANIEL NEWMAN: 

Yeah, I had a rocket ship in my post in case anyone missed it.

PATRICK MOORHEAD: 

Yeah. Listen, they beat on revenue. They beat on EPS. They raised their guide. But the thing that made the biggest difference was that people got super excited about doubling AGI from 1 billion to 2 billion, I think within 45 days, which is crazy. Now on the call, they asked if the supply was there and they said they've got supply for one billion dollars of that revenue. And right, you can see it on the ticker, right, as that was discussed that the stock, the stock, stock went down. Some other things that I thought was interesting is royalty at 21% for the full year is probably the most underdiscussed quality signal out there. And it's interesting, I, I do firmly believe that the royalty and licensing business won't go down in data center based on this, but I think we will know when AGI hits in probably six months after that. Another point I want to make is ARM as an architecture is just screaming, right? You combine Meta's Graviton5 deal and Qualcomm hyperscaler engagement, they are taking per core unit share. So I think it was discussed on the call that a 50% market share with the hyperscalers, that is absolutely nuts. And again, that's cores that's not that's not chip but that's that that is is super super, super, super important. So I'm really interested in the future to know what, so Rene Haas is now in charge of SoftBank International, which, you know, involves, you know, a lot of the acquisitions and a lot of the other businesses they're building. I'm super excited to see what the company is doing there. All right, let's shift to super micro that has been embroiled in controversy. Supposedly a board member was aware and actually taken a hairdryer to a label on a GPU box that ended up being shipped to China. And they put a label on it that was a detuned version or a different address of where it was being shipped. But apparently 20% of their revenue has been being illegally shipped to China. So their stock was down big time. So coming into this thing, the only question was, could they give confidence to investors on what things could be? So they did beat on EPS and Q4 guide that had been detuned down from consensus. They missed on Q3 revenue versus a consensus. So I, um, I just, you know, triple digit growth above consensus guide indictment, everybody's waiting on, but the stock still ripped like 18%. Right. But I just this this will leave a mark. There's no way it's not. There have been seven incidents prior that, that that, you know, in a way you really have to of put that into your supply chain strategy, whether you want to, you know, take, you know, take the risk on Supermicro.

DANIEL NEWMAN: 

You miss like in the middle of this melt up. So, you know, no, I'm saying like in the overall, if you're selling freaking picks and shovels during the gold rush, like this company is so messed up. I just I couldn't imagine owning it with all this overhang on it. But I guess, you know, there's always people that like jumping out of planes without parachutes. Yeah. I had two other quickies. Can I hit them? Sure. Yeah. I felt that there was a couple of other semi companies that have been absolutely ripping that aren't necessarily big names, but one was Lattice Semi and the other was Coherent. They both printed this week. Lattice saw its revenue got 42 percent. It is absolutely back into growth. You and I spent some time with their new CEO, Ford Tammer. You know, they've hit 50% year-on-year midpoint as their guide. They're beating and raising. And they had this interesting AMI acquisition, 1.65 billion, a big bet for the company. But it really brings them into a whole bunch more discussions. They're playing in the BIOS, BMC platform firmware space. It's going to give them a, you know, it doubles their service addressable market from six to 12 billion. It seemed like a really good move for the company. And I mean, basically every AI server on the planet needs what AMI has. So the FPGAs, which is needed in most cases as well, but they've just gotten themselves into another really interesting and strong position. This seems to be the signature move for Ford since he's arrived as CEO for the company. And obviously, Jim Anderson, who when I mentioned coherent, we'll talk about that in a second, brought this company to a certain point and Ford came in right at the kind of absolute at the absolute hell for inventory point for the chip cycle. And now this company is climbing its way back up. So I'll pause on this one and we can quickly hit coherent.

PATRICK MOORHEAD: 

Yeah, so on Lattice, I mean, it's all data center. And I think the market fully understood what the AMI, like AMI BIOS, like, they're essentially the 800 pound gorilla in all, in all bio. So you can think strategically, the control point that that gives Lattice right now. On fundamentals, one thing I love about what Lattice does is they'll share a 16 quarter view on revenue, gross margin, operating profit, and EPS, and everything is on the upswing. There was a tough downswing. post post pandemic and chewing off inventory. And none of these industrial markets are are swinging around the room doing great. But the good news is there is basically a trough on the industrial market. I think we're finally here on that stuff. So it's pretty much up and up in the right. Um, on this, you know, and final comment when Ford joined, right, he had a lot of M and a experience. So I thought immediately they were either going to try to sell themselves or, or buy a, buy a company. And, you know, here we are, right. It's probably not good read to go out and buy a company in the trough, but now that their fundamentals are crushing it, core financials look great. It was time. It was time to make a, it was time to time to time to make a move.

DANIEL NEWMAN: 

Yeah, absolutely. And then, you know, with the Owen Corning deal with our own Corning Corning, still want to call it I think of the Pink Panther dude. But the NVIDIA deal, like, you know, how hot is coherent? And, you know, Jim's another one. Jim's a lot like Elisa, right? You know, he's never overstated. This company sits in an incredibly good position right now with optics. They had a strong, you know, 21 percent year on year growth, 27 percent pro forma because of some of the deals that the company has made and some of the things have sold, 55% growth in EPS, margin growth, debts being paid down, entered the S&P 500, and is absolutely the center of a trend that's about to explode. They don't talk their order book, which I think they could get a bunch more credit if they probably started touting how much backlog they have, but I guess Jim's probably more comfortable just letting quarter to quarter to quarter, the valuation and the revenues and the margins continue to expand. But it feels like, you know, this company is on pure offense right now. The investment that came in from Nvidia, you know, and again, you know, their portfolio, you know, is broad. It's not just. You know, it's the modules, the optics, it's so much of the materials. I mean, they do so many parts of the value. Thermals, I mean, they're in so many different places that are going to be valuable. But the optics are the story, for sure.

PATRICK MOORHEAD: 

Yeah. And I think the reason that there was a sell-off was because the analysts had a disconnect Um, the analysts were disconnected between dates, right? Capacity is the constraint here, not demand. And it's all about Indian phosphide. Indian phosphide is the choke point. And a lot of talk about these, uh, six inch, six inch fabs are doubling capacity in 2026. And by the way, that's a quarter ahead of plan and then doubling. that capacity in 2027. So, and six inches, 4x the capacity, half the die cost, better yields, and the competitors are still ramping their transitions. And this is what people are missing. Final point, their manufacturing is done A lot of it in the United States and also with our allies that I think should give them a jump up at a minimum as a tiebreaker, but the ability also to charge more money for it. Daniel, great adders. I'm glad we talked Lattice and Coherent. We want to thank you for being part of our community. Have a great week as this is airing on Monday. Tell your friends, tell your family, tell your dogs to tune in and check out our short clips as well on X and other platforms. Hopefully we will get to Instagram and TikTok at a future date. Take care. We appreciate you.

‍