The Six Five Pod | EP 322: AI Writes Its Own Rules, Copilot Bills by the Task, and Memory Names Its Price

The handpicked topics for this week are:

1. White House Gathers AI Leaders Behind Voluntary Safeguards: Executives from NVIDIA, AMD, OpenAI, Anthropic, SpaceX, Microsoft, Google, and Amazon met at the White House to agree on a safety framework after a string of agent containment failures. Daniel Newman described two camps. The labs asking to be regulated are, in his view, partly chasing regulatory capture and positioning for multi-trillion-dollar IPOs, while Jensen Huang's camp argues the risk can be managed technologically. The group landed on immediate self-regulation, peer evaluation of models, and greater board oversight, with Mark Zuckerberg reading out the outcome. Newman called it a productive meeting with the right people in the room, since pausing or stopping AI isn't a realistic option. (The Decode)
2. NVIDIA Puts Agents in a Managed Sandbox: In the same week, NVIDIA introduced its open agent safety platform, which combines a sandboxed agent runtime with a reference design and can use BlueField-4 for enforcement outside the host. Patrick Moorhead highlighted Hugging Face CEO Clem Delangue's claim that OpenAI could not have hacked Hugging Face if it had been running the platform. Moorhead credited NVIDIA's leadership while noting the framework is open, runs on all x86, including Intel and AMD, and can integrate DPUs from AMD and the hyperscalers. He expects AWS, which has not yet signed on, to arrive with an approach that spans GPUs and Trainium. (The Decode)
3. Microsoft Resets Copilot Around Work Completed: CEO Satya Nadella positioned the new Copilot as the operating system of business, with a Home tab that combines chat and cowork, a Code tab that builds apps and workflows from plain language, and Autopilot for persistent background agents. Patrick Moorhead, who was briefed by the product team, believes this version has a high chance of sticking, flagging a pricing model that charges for work done instead of per seat. He named governance, productivity data, and Microsoft Fabric as Microsoft's moat. Daniel Newman cited an ETR pulse of 50 Fortune company CIOs in which about 70% want AI's pace slowed because they can't keep up. He argued Microsoft should be the adult in the room that sets the enterprise pace, the same play ServiceNow is running with Flow by wrapping vibe-coded tools in governance and SLAs. (The Decode)
4. Synopsys Brings OpenAI and Amazon Into AI-Native Chip Design: Synopsys launched GPT Synopsys, a frontier model co-designed with OpenAI and layered on Synopsys' full tool chain. Daniel Newman read the deal as OpenAI conceding that EDA tooling, IP, and experience are essential to cutting chip design cycles to six months or less. Amazon became a lead customer for Synopsys' application-optimized silicon IP and expanded its EDA and agentic engineering relationship, and Synopsys launched long-horizon Agent Engineer agents with Autopilot orchestration, citing 40 to 50 engagements already underway. Patrick Moorhead shared Synopsys chief product officer Ravi Subramanian's framing that AI reasons and explores while Synopsys generates and validates, along with the customer line "In AI, we believe, but in physics, we trust." Moorhead also warned that zero data retention doesn't mean companies aren't sharing IP, and pointed to confidential computing as the next shoe to drop. (The Decode)
5. Anthropic, OpenAI, and Google Match Prices on New Models: Claude Sonnet 5.5, GPT-6.1 Sol, and Gemini 4 Argon all list at $2 input and $10 output, with Argon still in security testing. Patrick Moorhead argued the price that matters is the cost of a known good output. He cited Signal65 findings that OpenAI's $200 Pro plan now delivers less output for the same price, and that GPT-6.1 Sol at max effort raised quality without lowering the bill per correct task. He compared the labs' efficiency push to Chinese open model developers, citing DeepSeek's 85% KV cache reduction. Daniel Newman called Signal65's Pinnacle testing the most important resource for enterprises trying to understand the token economics of running agents in production. (The Decode)
6. AMD Makes an $8.2 Billion World Labs Bet: AMD agreed to acquire Fei-Fei Li's World Labs for $8.2 billion, about 1.6 times the $5 billion valuation Bloomberg reported. Daniel Newman called it a talent grab and a bet on physical AI and world models, made possible by AMD's trillion-dollar market value, though the deal came with almost no disclosed financials. After a Q&A with AMD AI chief Vamsi Boppana, Patrick Moorhead described it as a talent and model buy with little near-term revenue, scoped across content, media, robotics, embedded, and data center. He said the team will need to keep building models to retain talent, and could give AMD sharper insight into its hardware roadmap. (The Decode)
7. The Flip: Will the White House AI Accord Solve Real Problems?: Patrick Moorhead argued yes. Getting competitors behind common responsibilities makes safety hard to dismiss, independent evaluation and board-level review offer concrete mechanisms, behavior is already shifting with OpenAI pausing frontier training after its containment failures, and the FTC's existing investigation provides a backstop. Daniel Newman countered that the accord's "believe" and "should" language reads like a mission statement, the pledge is voluntary with no enforcement, the floated oversight committee could come from the signatories themselves, and OpenAI and Anthropic have already committed to third-party evaluators. He pointed to reports of Meta's Muse agent exposing users' private data the same week, and said companies playing offense on safety ship evidence while companies playing defense ship photo ops. Note: The Flip assigns Patrick Moorhead and Daniel Newman opposing sides of a simulated debate, positions argued are not necessarily their own views. (The Flip)
8. Synopsys ($SNPS) Raises Guidance and Adds a $1 Billion Buyback: Synopsys shares gained 10% to 12% after Investor Day as the company raised revenue guidance from about $10.81 billion to nearly $11.2 billion, lifted margin guidance, and added a $1 billion buyback. Daniel Newman points to the systems side of the business, spanning physical AI, robotics, and simulation. Patrick Moorhead suggests Synopsys' terminal value is underscoped in a market of "two and a half players," referring to Synopsys, Cadence, and to an extent, Siemens. He expects AI to bring more customers into chip design, gated mainly by leading-node capacity at TSMC, with Samsung and Intel adding capacity. (Bulls and Bears)
9. HPE ($HPE) Lifts Its Networking Outlook at Investor Day: Four weeks after its September 2 earnings call, HPE raised its 2027 networking guide from a 14-to-17 range to the high teens and low 20s, which Patrick Moorhead tied to AI monetization and Helios inside AMD's rack. Networking margins hit a record 40% in Q3, and a $1.2 billion Helios rack order arrived, business Moorhead called high margin since HPE passes on lower-margin neocloud deals. Daniel Newman noted integration savings nearing $1 billion in annual run rate and a stock that has roughly tripled in a year. Moorhead argued HPE should be modeled more like Cisco than Dell. (Bulls and Bears)
10. Micron ($MU) Posts 87% Margins: Micron delivered $54 billion in revenue and $32.87 in EPS at about 87% margins, and guided to $61 billion, $38 a share, and 86% to 87% margins. Daniel Newman noted the growth comes almost entirely from pricing, with supply constrained in 2027 and likely 2028, and said roughly $70 billion in cash could turn Micron into a major ecosystem investor. Patrick Moorhead added that customers put down $12 billion in deposits to fund clean rooms arriving in late 2028, and that Goldman, Rosenblatt, D.A. Davidson, and Mizuho all raised targets, led by Rosenblatt's move from $1,500 to $1,900. Newman added that some accelerator makers are returning wafer allocations because they can't secure memory, capacity he expects NVIDIA to absorb. (Bulls and Bears)
11. Accenture ($ACN) Jumps 16%: Accenture rallied 16% in a day, though Patrick Moorhead noted half of EPS growth came from a lower share count after $2.3 billion in buybacks for the quarter and $7.5 billion for the year. Fixed-price work, including outcome-based contracts, now makes up 65% of bookings, which ties margin to delivery instead of the rate card. Moorhead doesn't think Accenture is out of the woods yet with roughly a million employees. Daniel Newman argued the "AI kills services" call has been wrong so far, because enterprises want to deploy agents but lack the resources to do it safely and securely, and they turn to partners like Accenture to get it done. (Bulls and Bears)

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

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

Patrick Moorhead:

Welcome everybody to The Six Five Pod. We made it 322 episodes, Daniel. And I feel like I've worked an entire day today. Oh, I have, because I'm over in Europe. But it's good to be back. It's great to see you, Bestie. I've really missed you. I need my Bestie time.

‍

Daniel Newman:
Oh, I know. Well, it's hard even when you're in the same time zone not to miss me. But now it's even worse when you get to be up thinking about it all day knowing that I'm just sleeping in. I got up at five and did a leg day. I got home late last night, but there's no better way to punish yourself than rolling out of bed. 91 sleep score, though, I managed to get six and a half hours of sleep and four hours of restorative in my six and a half.

‍

Patrick Moorhead:

 I mean, that is completely amazing. And I'm impressed. Well, my sleep scores are not in the 90s. I'm in here for Monaco for Lenovo's 360 Accelerate Europe and Meta program. uh, met a couple of your as well. And I'm here to nice job on stage. In fa How many sessions did you do? I really did three. I did three. It was funny. You know, you show up and they gave you a certain time frame. It's a stand up presentation. You know, I was going in and, you know. My previous year, I did 30 minutes. And then they came in and said, you've got nine minutes. And I had to negotiate for extra minutes. But no, it went well. I got a lot of good feedback on it. And it was all about tokenomics and basically known good enterprise work. And I shilled the shit out of Signal 65 a lot.

‍

Daniel Newman: 

That's a lot better than you quoting semi analysis. So that makes

‍

Patrick Moorhead: 

It is, it is, it is.

‍

Daniel Newman: 

Well, I mean, if I got a mole in the room to send me the slide and said, Pat just cited semi-analysis. I know it was Dworkash, but it, it said semi-analysis slash Dworkash.

‍

Patrick Moorhead: 

I'm like, well, you know, maybe this, This was just a way, a cry for help that maybe, you know, you would give access to me to your MCP server and I could start using your data. In fact, somebody may have suggested that to me.

‍

Daniel Newman: 

I think it's great, you know, to see our and my data on the screens. We get Synopsys Investor Day when they quoted. our numbers in their custom accelerator growth.

‍

Patrick Moorhead: 

That was great to see. Yeah, you were at Synopsys Investor Day and you did Fox and you did Schwab Network. And I'm sure I think you were either on Yahoo as well.

‍

Daniel Newman: 

Did that, yeah. And I actually Sadly had to turn CNBC down because I already booked something else in the morning and they've got all these rules. You can't do the two networks. Oh, and you know, I try to be good like whoever books first, you know, you never screw the one that books you first because. Again, it's a, it's a small world. You know, you've seen these producers go from, from place to place. Bridges can't be burned. But yeah, it was, I mean, it was busy. I mean, dude, the White House, I'd be not sure we'll hit this all like, there's no rest in this business. Like people are always like, how are you doing? Are you getting any rest out there? I'm like, has there been like a quiet week? Has there been like a, where you didn't, you know, and by the way, we have this administration that loves to drop shit on the weekend, like five o'clock. I guess his future is open. It's like, yeah, we're not we're going to go ahead and not bomb Iran or whatever massive sea chain. It inevitably starts on the weekend. And then whatever most like the most bad news that you're going to deliver is Friday at like 7 p.m. It's like, here's the thing that's about to happen. It's horrible. So you can have all weekend. And then by Sunday, there's like a reversal right ahead of market opens.

‍

Patrick Moorhead: 

Yeah, it was a busy week. You had the White House AI meeting talking about AI guardrails, Micron earnings this week, HPE Network Investor Day, OpenAI Dev Day as well. We also saw what I consider a Microsoft co-pilot reset. That was interesting. Oh, and AMD, you know, did an $8 billion acquisition of World Labs. We're going to be talking about that. And NVIDIA introduced an engineering solution to all this agents escaping from their cages. So why don't we, Daniel, dive into the D code? All right, Daniel, White House plus AI leaders and famous people arrived at the White House to have discussions about AI safeguards. I know we're going to debate this on the flip, so let's talk about what actually happened.

‍

Daniel Newman: 

Yeah, I mean, look, in the wake of escaping agents with a PDoOM score of at least 10, And for those that have never heard of the P Doom score, that's the probability of doom or demise. And I think the Cox in PR campaign was a 10 percent, a 10 percent chance that the world would end. President Trump said, okay, let's go ahead and bring the most important leaders, and I won't name all of them, but let's just say the top execs, NVIDIA, AMD, OpenAI, Anthropic, SpaceX, Microsoft, Google, Amazon, and so on and so forth. And let's get this group together, let's bring them into a room, and let's have an open discussion about the public positions that these companies have, how they're handling safety and security of their platforms. And let's come up with a framework. And let's have AI write the framework so that there's a misspelling under the signature. Wasn't it like the United States?

‍

Patrick Moorhead: Yeah, I'm pretty sure that was in there to show it wasn't AI.

‍

Daniel Newman: 

By the way, I'm joking. I don't know. Who knows, right? But that was funny. You're right. That's actually funny that maybe that was the intent. They actually say, hey, this was written by a human by the way. And you know, this was, I think the big question is you've got this kind of multi-camp going on where it's like, hey, regulate us, or as Dario's Saturday Night Live character said, I urge you to urge me to stop. You know, but Dario and Sam saying regulate us, And I think a lot of it is the desire for regulatory capture, protecting their market leadership, making the barriers to entry higher, more rules about what models are released and which models are considered safe to use, an assault on the open models market and somewhat a positioning exercise to justify multi-trillion dollar IPOs. But there is questions of real safety inside of it. Like, yeah, we need regulation. And so the other side of this is like, you don't know, you know, that Jensen is PDoom absolute zero. He's saying there is absolutely zero. I always like to say I think it's a non-zero, but it's certainly not 10. And that's like any technology, right? I mean, every technology has come with some risk. You know, we've opened up, put our grid online, you know, where people can connect to our power and to our water, you know, through tech. I mean, those are non-zero events that something really bad could happen. This obviously just accelerates that potential risk. But the zero or absolute zero crew is like, look, we can build agents, you give them no capabilities, You slowly assign to them what you allow them to do. And so that's the other side of the camp. It's like this can be technologically managed.

‍

Patrick Moorhead: Yeah.

‍

Daniel Newman: 

And so I think, you know, the TLDR of where they landed and Zuck was the kind of the reader of that was that they landed with these safeguards that really meant companies would do more of their own self-regulation starting immediately. Now, this is the big question of like, hey, this small subset of companies that have massive control over the financials, the economics, the business and the power, are they really incented to be safe, to prioritize safety? You know, because a lot of it was, hey, maybe there's an evaluative layer where they can evaluate each other's models. There's more accountability. There's, of course, as Jensen likes to talk about, there's the, you know, liability attached to any sort of damage. Think about a plane that falls out of the sky, a car that, you know, an ADAS system that crashes. Think about a drug, a pharmaceutical that builds a drug that makes people sick. There is liability tied to this, and that's always been part of a guardrail of every industry is not building and putting products out that could harm. But in the end, it was also like boards having greater oversight and visibility. Again, I'll hold my opinion because I wanna make sure that our flip is fun, but let's just say, I think some of the job is accomplished. I certainly think this was a good meeting with having the right people in the room and at least coming away with, hey, this is something we can action immediately because the option of pausing or stopping isn't a real option.

‍

Patrick Moorhead: 

Yeah. And by the way, the same week on that Monday, NVIDIA introduced what they call the open safety, open agent safety platform that combines open shell runtime with a sensory reference design. And you can even use Bluefield 4 for enforcement outside the host. Which, it's so funny, you had Clem from Hugging Face get on and basically say, if OpenAI had been using this, they would not have been able to hack us. So I thought that was probably the biggest statement point out there. I did a pretty long write-up on it, leaned into it on social. I think this is super important. Looking at the future, obviously, we have to be confident that this technology is going to have a low PDoOM number. And by literally managing your agents, uh, with a, a managed sandbox that also has some hardware enablement in the DPU, uh, you can, you can really make it happen. And I just want to have kind of hats off to, to Nvidia, uh, for the leadership that they're providing here. You know, I know people are all, you know, a lot of people, particularly competitors are quick to jump in and say, oh, uh, you know, this is another, um, a move to block out a competition. Uh, Somebody has to lead and do elements that are open. The framework is open. And obviously, if you're going to use Bluefield 4 for enforcement outside the host, that's NVIDIA. But AMD has their DPU. as do the hyperscalers, and they can integrate that in and do the work. There were some people who hadn't signed up, like AWS, and I'm expecting that they will come in and have an arrangement that does the same thing, but works across GPUs and obviously Tranium. The software that NVIDIA is bringing out works on all x86, including Intel and AMD. So I'm just interested to see what AMD has a comment on. So good stuff. We'll debate the reality of it, if it made a difference or not. But hey, let's get to the next topic. And this is Microsoft, and I'll call it the new co-pilot. Satya positioned us as the new operating system of business. And I think this one has a very high degree of sticking. And what I mean by that is sticking in a way that people really enjoy using it out of the gate. What it did is it adds a home tab, which is essentially chat plus co-work. You have a tab called code, right? That's apps and workflows from plain language. It's kind of, you know, a citizen app, uh, app developer, and then autopilot for those persistent agents and, uh, persistent, uh, persistent workflows and, and stuff like that. So, um, I got brief talk to the talk to the product folks. And when I looked at it, I did sit back and say, this is something that I would actually, I would actually like, like to use. Um, No offense. The big news, though, is is C plus meter, right? They're not just charging for tokens. Sorry. They're not just charging a per headcount basis, but essentially they're charging. They're charging, charging for for work done. Some of the details is not available exactly yet when it comes to the always on autopilot and code, but I think this is a really good step forward. Big picture, governance and the data is a moat. for Microsoft. They've got access to all of your productivity information. If you have your SaaS apps with them, you've got access to that. Microsoft Fabric essentially is a a fabric that connects all the other non-Microsoft data to your AI. It's one of the fastest growing and most used new products inside of Microsoft's entire cadre. So I'm optimistic here. I'm not bay-breathing this, but just looking at it and looking at the way that people work and they want to get it done, they've given an on-ramp for the newbies. which is the home tab, which suggests things. It gets you into doing things. And then once you start doing things, it asks you more interesting things it can do. If you want to create an app for your group or something like that, it gives you that ability. And then finally, all those persistent agents that want to be headless and just operate in the background, it has a construct for that too.

‍

Daniel Newman: 

Yeah. A couple of things, I'm going to zoom out a little bit here. I was on Varney and he asked me about Microsoft and I was actually saying like, you know, one of the AI adults in the room, right? This is a company that builds products for enterprises and understands enterprise as well. And that's part of the reason why I think people maybe ruled them out a little early in some ways of winning. All the credit was going to OpenAI and Claude. We kind of got the enterprise software thing wrong. And some of it's in the data. You know, we, our team at ETR did a quick pulse following the pacing the frontier conversation. We talked to 50 CIOs of fortune companies. And one of the things was is they all actually wanted to see the pace of AI slowed down. Isn't that interesting? Like not all, but like 70%. So the vast majority of CIOs are up against the problem is that they actually can't keep up with AI. This is really opportunistic in my opinion for companies like Microsoft, right? When a CIO is saying like, look, we don't actually trust at all why they want to slow down. Like they all said, like we don't trust the reasoning that they want to slow down, but we all want to slow down. Like we literally can't keep up with the options, with the technology, with what's happening. Companies like Microsoft should be the adult in the room. They should be the rational of like, let's make AI work for the enterprise at scale through our tooling, right? And let's kind of set the enterprise pace. And so a lot of it to me is like, look, you know, make CoPilot, utility a lot higher, which has been the complaint. It just doesn't work that well. And I think the enterprises have not made full commitments. And again, the other part of this is like enterprises do still want the business rules and rails. You saw ServiceNow came out with Flow this week, which is like basically a vibe coded ITSM. And they're basically, so what's happening now is software companies are effectively vibe coding tools, but what they're doing is they're vibe coding tools, and then layering the rules of business governance, and then SLAs around it. Because the problem is you can vibe code something, but can you actually manage it, make sure it's updated, make sure the data is safe and protected, and most enterprises don't wanna do that. So they're like, we'll just do that for you. And then we'll put, so this is something that Microsoft can do, keep everything in containers, data retention, save all the things that companies really are going to care about. Microsoft can put the wrapper around it while still offering the frontier value. And so this is where Microsoft should win. I mean, whether they win or not, we will see. But this is what they are inherently good at, and they already have distribution.

‍

Patrick Moorhead: 

Yeah, and speaking at the macro level, I mean, if you think about what Microsoft offers, and I was on a panel today talking about who would be the, who would be the winners. And, and my first bet was the CSPs, because not only do many of them have their own models, they also are introducing routing in there to route to the best model per, let's say, a multi-agent agent. And they have the infrastructure all there. And, you know, I'm loving the competition. I mean, heck, even OpenAI added a routing system to open models. through one of their inference providers. So I think we're finally getting to increase in maturity in enterprise AI. We started off in this very messy, messy zone. The funny part is, if you look at three years ago with what we were talking about, which was responsible AI, the importance of data, and a focus on the outcome, we're kind of back there again. And it just seems to be more important. And IP theft was a rumor or a scare tactic three years ago. The reality is, it's there. I had a slide in my keynote yesterday that showed within one year, Daniel, Anthropic went from standing up a service for pharmaceutical companies to creating a company to make their own drugs and having a wet lab that also controls robots inside of the wet lab to make scary stuff like drugs. So crazy times, and it's no better time to be part of this. So Daniel, no, what could go wrong? I don't know. There's only like a one to 10% chance of something going really wrong. So I feel pretty good about that. Joking, joking, folks. Hey, Daniel, Synopsys had its Investor Day. They also dropped some news before Investor Day. One was with AWS, and the other was with OpenAI. But they also unleashed a thing called Agent Engineer and Autopilot.

‍

Daniel Newman: 

Yeah, they had a lot to announce. And Big day. I actually ran into their CEO on the street in New York just randomly. Like before I actually, I was like walking down the street. I went to a bodega to grab a turkey sandwich. Cause you know, when I'm on trips, I eat these luxurious dinners in my hotel room by myself. But I literally just passed him on the street. It's like amazing. Cause you know, anyone that's been in New York, there's like 20 million people there. So running randomly into people, this is funny. But anyway, Yeah, look, this was another industry that Dario and Sam had said would be eaten by AI. I don't know if you remember that. There was a period of time where the EDAs got sold off really, really hard because this was just part of the AI eating software, and the EDA is just software. And then what was going to happen is, you know, the OpenAI would just build their own chip design tool. I think there was rumors, right, that the OpenAI chip was vibe coded. without any tooling. I think now we know that's probably not the case. And I think the big headline was the new product called GPT Synopsys. Now, there were some other big headlines business-wise, but the big one is basically OpenAI acquiesced and kind of walked back now that they've already you know, pillaged the market and taken all that market cap and moved it to their value and walked back and said, oh, wow, it turns out that all the tooling, all the knowledge, all the IP, all the experience that these EDA companies have is actually really important. to taking chip design and moving it from a multi-year process to tape out to six months or less, which is what everything's moving towards. And so basically they launched a co-designed frontier model that's going to be deployed using open AI for its frontier intelligence and reasoning and layer it on top of the entire tooling of Synopsys to enable the streamlining of chips and and and you know this is a kind of a two-part it's like one is the media techs broadcoms marvellous all get a lift because this will help them do what they do better and faster on behalf of their their ASIC partners and it also creates an interesting conversation about Does this eventually open the door to Google's and Amazon's taking more and more of that co-design and partnership that they have and doing more themselves with that? And again, that's the Jevons paradox versus disintermediation versus augmentation conversation. But what we know for sure is this is a process that is going to hopefully meaningfully shorten cycles on chip design. And it's really promising. But again, I think one of the big important highlights is it's not going to be disrupted purely by AI. The probabilistic nature of open AI itself doesn't work without all the true data and tooling. that sits inside of these companies. So that was the big announcement. You know, I think the Amazon announcement was another huge one. You know, a lead customer for their application optimized Silicon IP, extended their EDA relationship, simulation, agentic engineering. So they will be using the agent engineers. And of course, Synopsys has agreed to more deeply integrate with AWS on their compute storage and AI platform Bedrock. So and then like the third thing just worth mentioning is the agent engineer and autopilot. So they basically launched a series of long horizon agents that enable all the different parts of chip design. There's many parts, but simulation, validation, manufacturing, analog, verification, implementation, and then of course, adding autopilot that can enable orchestration, skills, memory, telemetry, and governance. This is to enable engineers to be more efficient. And they're already able to talk about some partners. They announced some improvements in RTL from, I think it was, sorry, Fujitsu. And then they also talked about, you know, having, I think it's like 40 or 50 different engagements that they already have underway using these platforms. So, yeah, so it was a great day and they revised their guidance upward because of this across the board. So very engaged group of investors in the room. It was a packed house. And, you know, I'll leave it there.

‍

Patrick Moorhead: 

You had a nice chat. I wasn't able to attend the event, but a nice chat with their chief product officer who runs all products and product management strategy, Robbie Subramanian. And he I love the way that he put it. AI fundamentally reasons and explores and synopsis generates and validates. And I really liked that. And it hits to a lot of points that you were talking about, Daniel. And he also relayed the lines customers use. In AI, we believe, but in physics, we trust. So very well thought out here. I thought this was kind of a Claude Forrest moment. for them in a way and their stock didn't rip 25% but it did go up.

‍

Daniel Newman: 

It was up like 10 though, 10, 12.

‍

Patrick Moorhead: 

Yeah, it was really good to see and that was expected. What I like about this as well too is that it busts Um, you know, I, I met with Richard Ho, uh, who runs, uh, Silicon and OpenAI. I met him, met with him at the GSA. I even had dinner with him and it was kind of, it's kind of, you know, joking, like this was the guy that, uh, made the, um, uh, made the disclosure about basically, I'm joking here, uh, vibe coding, um, jalapeno chip, which again, I'm radically simplifying kind of kind of being funny here. But the reality is, is they were working side by side with synopsis the whole time. And and basically tap Richard tank tanks their stock $10 billion. So it's good to see that, you know, you know, the bounce back of this. Data is everything, right? And I'm doing more research on the data regulations. How are you not going to send traces to OpenAI that it can derive knowledge from? I think there's more disclosures to come, Daniel, because zero data retention doesn't mean that you're not sharing your IP.

‍

Daniel Newman: 

It means they see it, they just don't keep it. And that's a promise they're making that we have to trust.

‍

Patrick Moorhead: 

Yeah. Yeah. And I think we're going to see more because I think the next shoe to drop, and I should have talked about this when we were discussing NVIDIA's new agent security platform, is it's confidential computing that will make the big difference on here. I think we're going to see a lot more news to come. And essentially what that means is running the models on-prem. and having the intelligence updated. But, but we will, we will get there. So yeah, it costs two seconds.

‍

Daniel Newman: 

I need to make a quick washroom job. Sure. Thanks, buddy. I'm just gonna pause. I'm pausing. Okay.

‍

Patrick Moorhead: 

All right, Daniel, let's move to the next topic here. In the spirit of new model, new capability every week, and the seven-day-a-week, 24-hours-a-day grind, we have new models from Anthropic, OpenAI, and Google.

‍

Daniel Newman: 

Essentially… That pause is impressive. What's that? The pacing is good.

‍

Patrick Moorhead: 

Well, what's interesting about these three new models is three labs, one price tag, and Sonnet, Sol, and Argonne all list for $2 input and $10 externally. As we have been trying to educate the world, it's not about the token cost. It is about the cost of a known good output. First of all, competition's good. It's good to see Google rearing their head in there. There's no way to test Argonne because it's still being tested. The animals are definitely locked up in the cage. They're doing security testing and likely doing the testing that Anthropic and OpenAI should have probably done with their models. Signal 65 also came out and weighed in on OpenAI having the usage allowance on the $200 pro plan and to see what the output was. And essentially, Same price, less output. And that's not a great place to be yet. And all of this makes sense, right? If you look out, you see these optimizations. What they're doing now is investing a lot of the times and what I believe the Chinese open model folks do, which is they're optimizing for efficiency. Right. You saw DeepSea cut KB cash by eighty five percent. This is really this is really similar to to that. So how GPT six point one sold doing it max. effort? Well, net net, the data quality went up, but the bill per correct task did not come down. So, good stuff. If you want to see any of this information, check out Pinnacle on our Signal 65 website.

‍

Daniel Newman: 

Yeah, no, I don't have a ton to add here, Pat. This is going to be just a weekly thing, it feels like. We just have to talk about what's come out. Like I said, nobody's pacing, so just cut the bullshit, everybody. We're going as fast and furious as we can. I do think, you know, what our team at Pinnacle is doing is probably the most important for the enterprises that are trying to understand the token economics of actually deploying the agents and running them against the different use cases and workloads. And so we'll continue to try to put that out as quickly as possible so that everybody can see what it takes to get a successful task done that delivers business value.

‍

Patrick Moorhead: 

Yeah, good stuff, Daniel. Hey, let's move to the next topic here. And that is? lost my place here. AMD invested $8.2 billion into Fei-Fei Li's startup called World Labs. And she came out with essentially a world model here.

‍

Daniel Newman: 

Yeah, $8.2 billion. She's finally getting paid, right? Isn't she the AI sort of OG that hand labeled the first frontier or first kind of model, you know, gave it all away for free, made no money. So this is, she's getting paid now. Her and her partners, I mean, look, I guess you could call it the two most powerful women in AI now teamed up. Yeah. To work together. And, you know, on the AMD side, like, look, this is what a trillion dollar balance sheet enables you to do is make eight billion dollar bets. And that's what this is here. Like, this is really a bet on kind of what's next. World models, quantum, these are the kind of areas, physical AI, of like where the next trillions of dollars of TAM sit. And that's, you know, not all AI is in the data center. Labeling the physical world, for instance, is going to be a challenge and a task that's going to need to be addressed in the long term and will be driving the future of the frontier. And so to me, this feels like a big buy into physical AI, but it also just feels like a big talent grab. In my opinion, it's a massive investment, but to grab what would be seen as probably the foremost talent in this particular space. And, you know, I think, you know, the economics are a little bit less well known, like how much actual economic value is there to this deal. But again, you know, Lisa can play with the stock. I mean, it's like I said, the company's a trillion dollar value now. This is like Elon doing the cursor deal. It's just like, I can do this. I'm just going to ramrod this right into this massive valuation of the company. I'm going to buy interesting assets and parts and pieces and talent and people, and I'm going to put it together because we need to be building for what's now and we need to be building for what's next. But You know, it's a very interesting move. Like I said, it's just, there's almost no numbers around it. Like there's no numbers around what they own, what the business is, what the revenues look like. This is a speculative play into a future ham that I'm sure AMD feels it needs to be competing in.

‍

Patrick Moorhead: 

Yeah, so AMD was an investor in this company. Feifei was on stage at AMD's AI event as well. So the two know each other well. By the way, NVIDIA was an investor as well. So a couple of takeaways. I think that the most that I got, I got a Q&A with Vamsi Bhopana, who runs AI at AMD. And here's kind of my net on this. This is a talent and model inside buy. It's not a revenue play. You know, you called it zero revenue. Maybe they got some NRE from from some of their partners, but but likely zero revenue. So it's funny, it it also got priced at 1.6 times the $5 billion valuation that Bloomberg reported. And the other thing, and I know people jumped on this, and I would understand if you read Fei-Fei's blog, it's like, this is a physical AI play. But in my Q&A with Vamsi, he was very clear. This scope is AMD-wide. it's horizontal model capabilities for the company across content, media, robotics, and embedded in data center prediction and reasoning. So I think that's, you know, I appreciated his clarification here. The key here is, and I also brought this up in my call with him, is this team has to build models to keep that talent there and to keep that talent and to attract talent as well. So it's interesting. I don't know if AMD is going to release competitive models and where they're going to release it, but I'm excited to say what I can say without hesitation is this team, if they can keep the core team in there, will absolutely give better insights into what the hardware roadmap should look like as well. All right, folks, that is the decode. Why don't we jump into the flip now? So we talked a little bit in the decode about the White House AI Accord. But the question is, is it will it solve real problems? Or, you know, or is this was this just a great photo opportunity for the United States of America? Let's jump in. All right, I am for this. This is going to make practical progress before we get into new legislation. So first of all, getting competitors behind all the common responsibilities can make it really hard to dismiss safety as someone else's problem. I think that's a good thing, right? Independent evaluation, board level review offer concrete mechanisms and not just merely a statement that AI should be safe. And it's certainly better than coming in and saying, regulate me. or recuse me from having any product liability if my agents come and end up killing you or something like that. So the self-governance is where this is heading, right? And it's what the labs actually wanted, aside from getting that liability get out of jail card free. And I do like, you know, when you combine this with stuff that NVIDIA brings out that actually has an engineering solution over the problem, I feel pretty good. about that. So net net, I just think that this wasn't just a photo op, and there were those real, real teeth in here. And quite frankly, I, you know, I think the behavior is already changing. I mean, I know we were kind of joking in the decode about, you know, these new models coming out. The two models from the Frontier Labs in question, Anthropic and OpenAI, it was really about cost reductions. These were smaller models to give more capabilities, though they weren't actually pacing anything on the frontier. We already saw OpenAI pause frontier training after its containment failures. And by the way, a final comment here. There is a backstop here, right? The FTC was investigating before the accord and confirmed the next day. So the existing authority still applies and assigned public standard, I think, gives enforcers and buyers something specific to hold to hold companies to.

‍

Daniel Newman: 

Yeah, Pat, I'd love I'd love to agree with you, but then we'd both be wrong. You know, Mark Zuckerberg, the face of this new pledge and finding out they're very cool new muse agent. is hacking people's iMessaging, taking something that's supposed to be extremely secure and giving away people's addresses that are using Muse to transact in different platforms. It's very safe. Get people access to my home and all my most private interactions with the people closest to me. Very, very good stuff. But in serious, like you know where I stand. I said it this week, if you build the product, you're responsible to make it safe and secure. And that's why the Accord just doesn't doesn't do it with me. Do it for me. You know, start with the language. The document says the signatories believe each company should implement controls and audits. Believe. Should. That's not a commitment. That's a mission statement. Second, there are no teeth. Speaker called it voluntary. The president called it morally binding. Congress didn't announce any way to enforce it. And the court itself admits none of this is law. And what's the oversight that was floated? A committee of maybe 10 people that could come from the same group that signed it? You know, that's basically players refereeing their own game. Third, What's actually new? OpenAI and Anthropic had already committed to third-party evaluators before anyone sat down for lunch. So tell me what changes. Monday morning, which model launch gets held? Which agent loses permission? Which board votes no? No one can answer that from this. So the final part is like, look at the incentives. Same day this was signed, the president said he'd never stifle the growth of this technology. So the message in March is pretty clear. The pledge is the guardrail. And you know, look, we're in the hard ROI era. No CIO signs a seven figure deal because a vendor posted a pledge. They want evidence. Safety should clear the same bar. Companies playing offense on safety ship evidence. Companies playing defense ship photo ops. I'll give Zuck credit. He called it a start, but a start isn't a solution. Show me published control changes, reviews that are actually independent, and a consequence when someone falls short. Until then, this is a great day for communications, but unfortunately not much of a day for safety.

‍

Patrick Moorhead: 

That was pretty good. You got a notification. You might want to get that.

‍

Daniel Newman: 

I know you love when I type during our calls and you like it even more. Listen, if you don't let me type, I should mute, but if you don't let me type, then how am I going to have AI tell me what to say?

‍

Patrick Moorhead: 

I just pre-run it. I have, literally I have workflows that will automatically send it to my inbox.

‍

Daniel Newman: 

But sometimes you become inquisitive during or like you'll say what I wanted to say. So it's like, I need something new to say. No, I'm kidding. You know, it's funny. I did three different one hour long TV shows this week where I was on the entire hour and I never brought any notes with me. And like, people are always like, don't you want to bring your laptop or your keyboard? Yeah. And I always jokingly say like, if I don't know it, like, I shouldn't talk about it, but anyways, it was just funny because like people have become, so actually everybody out there, the point here is. I'd use very little AI for this stuff because like, because I actually- You've talked about it all week, that's part of it. Well, part of it, and I'm gonna use a word, I'm gonna say it very carefully so you don't think I'm saying the wrong word. I'm retarded, meaning when I'm on TV or I'm doing a podcast and I'm reading, it's obvious I'm doing it and I hate it. So it's like, I try to actually know what I'm talking about before I get on because when I'm reading, I'm like this.

‍

Patrick Moorhead: 

Yeah. My biggest day I had, aside from the flip, is really just culminating what I've said publicly on the topic.

‍

Daniel Newman: 

Yeah. Like building your own continuity, right?

‍

Patrick Moorhead: 

That you're consistent. A hundred percent. I'm looking through this. It's like, uh, here's what happened. Pat's take, here's what they said. Here's Pat's take, here's Matt's take. And then I take credit for what Matt said, you know? And then it's like what Daniel said, and I'll take credit for the smart stuff you said. No, actually I don't do that. I don't do that, Bestie. No, I rather, I rather show, uh, my competitors data.

‍

Daniel Newman: 

That is a good idea.

‍

Patrick Moorhead: 

I do appreciate you doing that. I mean, besties let other besties use their data out here. Yeah. Maybe we can work out an agreement. Possibly. I mean, I hear your licensing is so onerous, but yeah, I mean, it's the only value I have left.

‍

Daniel Newman: 

You know, is whatever unique proprietary data I have and You know, I can't give that away because that's it. That's all. That's all we've got. Otherwise, you can just create an agent that speaks like me and shitposts. You could literally be like create an agent that shitposts all day like Dan. What was that? Writing my tweets on the way to the toilet?

‍

Patrick Moorhead: I don't know. I mean, are you telling me your value as the human in the AI loop decision making process like there's no value there?

‍

Daniel Newman: 

No, it's just saying there's increasingly less if all your data is out there for free. Everything you know that everyone else doesn't know is just out there to be had. And then really it's just how you synthesize and the relationships you have, which isn't zero value, but it increasingly becomes competitive when someone's like, You know, same reason like if you can just look something up and get all the info with no wall, would you actually take the meeting? You know, it's like I tell vendors this all the time. It's like if I can literally just ask AI to tell me what you're going to put me in a room to tell me for eight hours and you're not going to have me sit down and have interesting meetings and discussions and conversations that aren't accessible to me, like why the hell would I come? Same reason why people think anyway.

‍

Patrick Moorhead: 

All right. Good, good flip. I feel like you you nipped me on this one. That was pretty good. You pulled in Zuck. But hey, let's move into bulls and bears. Let's see what happened in the marketplace. All right, Daniel, we already talked about what Synopsys said at Investor Day, but it looks like we are going to talk about it again.

‍

Daniel Newman: 

Yeah. You know, I wanted to just make sure we covered a quick quickly because we didn't hit the some of the capital markets part. We talked a lot about the tech. I'll skip that altogether. And besides the fact that, you know, the stock saw a nice double digit gain, they did raise their guidance meaningfully from about 10.81 to almost 11.2 billion. And they also raised their margin guidance.

‍

Patrick Moorhead: 

Yeah.

‍

Daniel Newman: 

So, you know, and on top of that, they put a billion dollar buyback into their plan. And so I don't think I don't think this needs needed a lot of detail. But since we talked mostly about the products they announced and launched and we didn't talk about the numbers. it was the underpinning of what does it mean? And it means more expected revenue, higher margins, and they are investing on the back end to bring down their share count, which is all three things that the investor should like. And on top of that, you know, I did think I didn't mention this, but the system side of the business, the answer side of the business is looking increasingly interesting. You know, we talked about the Fifi Lee deal. We talked about world models. Well, building AI and like If you listen to Micron and other companies talk about why memory is not the same cycle, it comes down to physical AI, robotics, simulation, and design. Synapse is a really interesting company because it's in all those spaces and now it's in the space of actually enabling companies to simulate design of next generation technologies. So very cool, good outlook, positive, and that's all I think we really need to say about that.

‍

Patrick Moorhead: 

Yeah, it's interesting. I know typical sell side goes out 18 months, maybe three years. But if you look at the terminal value of a company like this, I think is increasingly high and underscoped. There is competition in this market, but the reality is there's two and a half players, right? There's Synopsys, there's Cadence, and there's a little bit of Siemens. And then you have some of these AI startup folks, and then you have people augmenting AI on their own. And if you think about it, the number of customers is rising. So I'll call it the market TAM is going up. the ability to monetize AI and agents. That's one thing that they made clear, you know, they had talked about how they're going to use their agents, but they never applied numbers to them. And, I mean, at some point, the ability to create a chip will be increasingly easier for people. And therefore, I think smaller companies will be able to do it. The larger companies will be able to do more chips. I think probably the only gating factor there is there's only so much TSMC leading node. The good news is Samsung and Intel are adding, you know, they've gotten over some of their trouble spots. and are actively adding capabilities to that. So let's go into the next topic. And that was HPE's Investor Day for networking. It's interesting how many people had said, hey, this is going to be a really bad acquisition. Antonio said, hey, this would be very quickly making profits and be being accretive. And sure enough, in the last two earnings calls, that's exactly what was demonstrated. So, you know, the guide moved a full step in, in only four weeks, right? September 2nd earnings call, Antonia was guiding 2027 networking, 14 to 17. and the new range and the high teens and the low 20s, right? So that's a huge boost for networking. And I think it has a lot to do with its ability to monetize AI and quite frankly, Helios inside of AMD's rack. So mix is also that it was was also a story, right? Q3 margins were a record 40% on networking. And that was a 10, 1011 points over its over its 10 Q. And the reason that they said this was happening was because networking is growing as a percentage of their of their business. The other thing that came in, and I'll end on this one, an order, a $1.2 billion for a Helios rack came through with Vulture. I hadn't heard much from them, and a billion isn't 50 billion, but it's good high margin business for the company. And by the way, the reason I'm I'm saying it's high margin, just knowing that HPE doesn't take the lower margin NeoCloud deals.

‍

Daniel Newman: 

Yeah, they've been very focused on business selectivity, we can say. It was a good result, you know, seeing the revenue growth guidance jump, seeing the integration savings targets jump. I believe they're getting closer to a billion dollars of annual run rate cost savings through the integration. Those are the success stories that often don't get told is, hey, by putting these two things together, we found a lot of efficiency. That's often part of how valuations create ratings. Um, you know, the fact is, is networking is another massive frontier. And, you know, if we agree in the story of AI moving from pure API connectivity to more open source, more on prem, more hybrid architectures, HPE is another one of those well positioned adults in the room that I spoke to that, hey, they're the kind of company that will be deploying the infrastructure edge to cloud, which was always the the strategy of HPE. So, you know, this was a much you know, needed deal for HP. And I think it actually revitalized the company. And I think that stocks up like triple in a year. So, I mean, the market is clearly starting to appreciate this transformation, the improvement, paying down debts, accelerating the cash flows, increasing margins. And again, they've done it on their own terms, not necessarily chasing the scale of Dell. They're doing it their way. And by focusing on networking and security, they had a differentiation in how they position themselves, which I think is working out very well.

‍

Patrick Moorhead: 

Yeah, I think HPE should be modeled more like Cisco than Adele from what it does and the value add it's trying to extract out there. And albeit HPE isn't as broad in its networking as Cisco is, it has the compute to go along with it and also the full stack AI software as well. So let's get into our next topic. And that is Micron Earnings. Daniel, you seem to be all over this. This is a company you love to talk about their stock.

‍

Daniel Newman: 

I mean, this is the world series of earnings. The Super Bowl has passed.

‍

Patrick Moorhead: 

I thought it was NVIDIA. No, that's the Super Bowl. Oh, sorry. Sorry. Sorry. OK. Yeah.

‍

Daniel Newman: 

And it is probably one of the most telling about where we are in the cycle and where we are in the trade is a company that, you know, I'm proudly said when it was one tenth of where it sits today. I said this is going to be a massive runner. And it's proving to be the case. I mean, look, this numbers, Pat, they were just absolutely mind boggling. This company to deliver $54 billion of revenue and $32.87 a share of earnings. And their margins, Pat, I mean, margins that make Jensen cry. 87% margin, you know, and all this growth, you know, and here's the thing, like guiding to $61 billion, $38 a share, 86 plus, 86 to 87% margins, incredible supplier side economics with their agreements they have in place, the long-term agreements, the pricing floors, and some of the comments, basically like, We are going to be constrained in 27, likely going to be constrained in 28. So, you know, some of the skeptics are saying, well, growth is slowing. I know, even if all these numbers growth is slowing and they're not necessarily wrong, but that's because no new supply has come online yet. So everything we're seeing right now is makeshift and margin. The entire growth of this business is basically they're making about the same amount of stuff they were making a year ago and they're able to sell it for exponentially more. And then, of course, it's the selection of how they want to distribute across NAND, which they're selling less of, and how they want to position their DRAM against use for HBM. We're seeing an increasing HBM allocation because that's the important memory for these AI systems. And they're able to basically put this mix together and margin together to create this incredible growth. And this doesn't even account for all the capacity adding they're going to need to do. As more capacity comes online and the demand stays strong, and I anticipate it will, these numbers will get even bigger. And the last thing I'll just say is this company, this other, their cash and equivalents ballooned to like $70 billion this quarter. Within a few quarters, that's going to be in the hundreds of billions of dollars. This company is going to be able to go Nvidia on the market. They're going to be able to basically start picking winners. They're going to becoming a massive VC to the market. They're going to be able to start underwriting scale and growth. They are the next major ecosystem player. And so I just think it's important to call that out.

‍

Patrick Moorhead: 

Yeah. The margins were awesome. 87 from 80, 85, essentially.

‍

Daniel Newman: 

Unless you're buying from them.

‍

Patrick Moorhead: 

No, exactly. And I met with a couple of those this week. So, in fact, Micron was on stage here. Their chief sales officer was here. Didn't get a chance to talk to him. He got on the stage and then boogied. But he had some interesting stuff to say. So one of the things that come across are not only do they have pricing power, and by the way, all of the revenue increase came from price increases. Customers also gave $12 billion in deposits. And those deposits fund the clean rooms that those deposits fund don't arrive until late. 2028. And that also shows what a good, what an absolute great, great businesses in one, one conversation that came up was HBM is the is now the lowest margin unit. And I'll be honest with you, I don't fully understand how that's even remotely possible at this point, other than yields on it, because that is where you can leverage the higher prices. And I don't think the competitive dynamics are actually Um, less fierce on HBM. But if you have an answer to that, put it in the, uh, put it in that, put it in the comments. So, um, a lot of, I've been having a lot of discussions too, about, you know, when will we get through this? I had a very. incredible and smart person tell me we're going to be out of this in 2029 and I can't even fathom that. So first of all, if we hit our 2029 growth, If one of the memory makers stumbles on building factories, if Elon remotely hits anywhere near his million robots, which by the way, I do think China will hit, they will continue to crank out those. And what people forget about the most on DRAM is when the PC units are down 20% a year, that demand doesn't go away. Somebody that was trying to replace their five-year-old PC and delays it isn't going to drag that thing out to eight years. It will catch up and they will want to buy a new PC. So that has not destroyed demand. That is just pushing to the right here. Final thing, some re-ratings, Goldman, Rosenblatt, D.A. Davidson, Mizuho, T.D. Cowan, excuse me, not T.D. Cowan, the previous four all had increased re-ratings, up some serious dollars. Probably the biggest one was Rosenblatt, $1,500 to $1,900. dollars. Hey, where's a Futurum equities, Daniel, on this?

‍

Daniel Newman: 

I believe it's 1600. I don't know if we've updated I'd have to ask Rolf.

‍

Patrick Moorhead: 

No, it's beautiful. Just sorry to put you on the spot there.

‍

Daniel Newman: 

I should have all this memorized. I'll look it up at some point.

‍

Patrick Moorhead: 

Well, I know it's not. You're only loosely affiliated with that company.

‍

Daniel Newman: 

So I have to keep my guard. I have to keep the rails up. But yeah, it's a winner. And by the way, one of the interesting factoids I got from an insider and I can't share who or any details, but was effectively some of the accelerator makers are starting to have to give wafer allocations back because they can't secure the memory.

‍

Patrick Moorhead: 

That's 100% and that came up in my GSA panel last week.

‍

Daniel Newman: 

And interesting on that though is who do you think is there to gobble up those wafers? It has to be NVIDIA. Yeah. Because nobody has secured more supply chain than those guys and all I'm saying is Any of the others that didn't have a planned memory allocations or didn't have a memory purchased in advance, it feels like that's just a gift horse to Jensen to be able to go buy that up because you know he bought up the most memory.

‍

Patrick Moorhead: 

All right, let's let's move to Accenture, right? Accenture, right? Accenture is going to zero. AI was going to we don't have a word like saspocalypse or the service providers, but yeah, they were supposed to be killed. It jumped 16% in a day and half the EPS growth came from a lower share share share count, not the finance bots. So, um, What can we say here? Billings, right? 65% of billings are fixed price now, right? And what they said on the call was fixed price work, which includes outcome-based work. Now it sees 65% of their bookings. And this is good because it gives investors confidence that there's stickiness here. And with outcome-based work rather than hours, Margin depends on the delivery, not on the rate card. So, half of the EPS gross was the share count that needs to be noted here. Lower share count for the quarter because of bought back. It bought $2.3 billion of stock in the quarter. and $7.5 billion for the year. So a good sign. I don't think, you know, they're completely out of the woods yet. I don't think that they have had their Claude Force moment. I don't think that they've had their GPT moment yet, but we will see. I think, you know, you've got a million employees or so. I just, I can't imagine that there's not some structural change that's coming up.

‍

Daniel Newman: 

Yeah, look, I mean, I think this is one of those, another, all I can say is everything we've been told along the way just hasn't been true. Software would end, security would end, data services would end, labor would end. And AI, again, ultimately the question is, does it grow the GDP and does it grow it meaningfully, meaning that we're actually creating more work and more outcome and more product. at the very least in the early era of trying to make all this stuff work, people need more help than less. The kind of the idea that the agents just run rampant and build things for you and then you deploy them. I think we've all been through that cycle of, but it was a little bit of like a, what do you call it? Like the peak of disillusion that you would build things. I mean, I remember you and I sitting at Mobile World Congress is building stuff like we're gonna get rid of everything and everybody, and we're gonna use this new data platform that we built in like, And we built kind of cool looking prototypes, but like nothing was production ready. Nothing could actually be deployed at scale, but it was hopeful. And I think a lot of enterprises are going through that same thing. I mean, I've had a lot of conversations with Um, with, with enterprises and, and, and those selling to enterprises. And basically the thing is like, they want to do the thing. Like they want to optimize token, token routing, model, model selection, deployment, building agents and workflows to, to audit. But in the end, they're like, we don't have the resources to deploy this stuff and be sure that it's safe, secure. Um, and so who do they turn to? They turn to an Accenture. I mean, that's the, so I think that. You know, again, ambitiously, we want tech to solve all the problems, but there is still a human in the loop. I still think very strongly about the forward deployed model, doing it with less humans, more efficiently, alongside AI is the ultimate outcome. But there's very few organizations that have really nailed that so far. Just us and Palantir. Anyhow, that's it.

‍

Patrick Moorhead: 

I think we're done. So great conversations here. Daniel, where are you headed next week? You and I are going to be in New York for IBM Analyst Day. We're going to have Arvind and his leadership team there looking forward to that. And you've got a couple other things that you're going to be attending as well.

‍

Daniel Newman: 

I'll be at Marvell Investor Day next week. Um, and I think there's a couple other events I might share on social media later. I'm going to make you all wait. Okay. Yeah, I was, you know, you look at that White House seating chart and think like, I've spent good amounts of time with like half the people in that room.

‍

Patrick Moorhead: 

Yeah, it was it was great to see Lisa there and Michael Dell and Hawk Tan was there. Cristiano was there.

‍

Daniel Newman: 

Lots of videos on Six Five of us having conversations with all the people in that room. It's kind of funny. I know. I know. I said, you know, like I'm only one degree away from being important.

‍

Patrick Moorhead: 

I always tell people, hey, I'm not ring zero. I'm probably ring one or ring two on the on this chart, because I literally people ask me, why weren't you invited to the White House? And I'm like, you got to be kidding me. Come on. Come on, folks. Put a B in front of my number. If somebody comes and offers me, you know, a billion dollars for my company, you know, maybe maybe that that'll I think you should be there because you're you're amazing. Appreciate that. So anyway, everybody, thanks for tuning in with us. Feeling very energetic. It's the end of the day for me. I'm probably going to, oh, I need to do a revenue meeting, but then I'm going to get into the gym stuff and hit the gym. But back to the US tomorrow. Thanks for tuning in. We care about you. Take care.

‍