Live From Cisco AI Summit | Chuck Robbins, Aaron Levie, Jeetu Patel, Costa Kladianos, Dylan Patel

26:35Yeah. Famous, famous. Really quickly. LabelBox. RL environments, voice, robotics, evals, and expert human data. LabelBox is the data factory behind the world's leading AI teams. Eric Berlin says, once again, I find myself updating my LinkedIn bio. He says, former CEO of Breaker, which was acquired by Twitter, acquired by XCorp, acquired by XAI, acquired by SpaceX. So congratulations to the Breaker team. I was looking for the most complicated corporate lineage yesterday when we were joking about it. Like there's someone that's going to have like six steps in their resume and we found him. His name's Eric Berlin and he founded Breaker.

27:14What was Breaker? Was that a podcasting app? Live sports. Oh, really? I think it was meant to basically distribute, effectively, clips from games in the moment. Cool. Oh, yeah, like breaking news. Oh, 2021. Twitter acquires social podcasting app Breaker team to help build Twitter spaces. Twitter has acquired social broadcasting app Breaker. The company's announced today. Yeah, I think the idea was if there was a crazy play or a game was about to end, they would just stream. Huh. just like the last five minutes or something interesting breaker was founded in 2016 and led by ceo berlin previously the founder and cto at social advertising 140 proof which he also sold and oh and leah culver was at breaker yeah i remember her uh in in the in the twitter she she stuck around in the transition and was like i think she posted a photo of her like sleeping in a sleeping bag in the office or something um the app had launched at a time when podcasts were still very much thought of as audio feeds and podcast apps as productivity tools, not experiences around which a community could be built.

28:17Breaker helped users change that perception by offering an app where users could like and comment on episodes, discover new podcasts by following friends. Okay, I'm thinking of a different company. You are thinking of a different company. It's more of a clubhouse rival. Got it. According to Culver's tweet, she'll be joining Twitter with a focus on Twitter Spaces, Twitter's audio-based social networking product, and Clubhouse Rivals. Spaces let Twitter users chat in real time using voice instead of text, as they do today. And, you know, that product still exists. Suspended cap. Says, so let me get this straight.

28:51Overpays for Twitter. Makes XAI. Uses AI hype cycle to absorb Twitter and make everyone whole. Then uses SpaceX IPO hype to absorb that entity. Will pump the living S-H-I-T out of the SpaceX IPO and buy more stuff with equity. Like, of course people keep giving this guy capital. He finds a way. It's crazy. It's really, really true. Yeah. And in some ways I've been thinking about it is XAI has not, you know, they've done fine. In so many ways it's been an incredible story, come from behind story, competing against, you know, the Googles, the OpenAI's of the world. But it hasn't exactly been an easy time in the private market.

29:31It's like going out and having to raise at a$200 billion valuation. That's a lot of money. When every single investor that you're pitching is looking at OpenAI, they're looking at Anthropic, they're comparing your traction to theirs. All those people that were investing in XAI had to just say, full blind faith, Elon, I know you got us. And so this new transaction was the investment thesis. It was like, hey, sort of unlimited upside, somewhat capped downside. The downside scenario is X and XAI get rolled in. and so it's certainly rewarding everyone with their loyalty. Yeah, I mean, a bunch of investors have kind of like laid out this thesis of Elon, Inc., just Elon, just bet on Elon, don't bet against Elon.

30:20Sean McGuire, I think is on all three, XAI, X, and SpaceX, and then Andreessen Horowitz as well. And they posted an image of like X, SpaceX, XAI. And individually, a lot of those deals were sort of crazy and critiqued, but together, everyone's doing very well. We've got to figure out what's going on with the boring company. I'll tell you about Gusto first, the unified platform for payroll benefits and HR built to evolve its small and medium-sized businesses. UAE officials say the first phase of the Dubai Loop project with Musk's boring company to start immediately. they are breaking ground Dubai has UAE has some insane traffic we have not seen a lot from but I think they're still cooking I know there's been some back and forth about the Vegas tunnel some stuff that's good, some people are annoyed with like the construction and what not but it seems like, I don't know, it's progressing a little bit, it still seems really really slow considering when did he originally post the Hyperloop blog, like 10 years ago but building tunnels in the ground.

31:28Difficult, difficult. What else is going on? Moving on. San Francisco is getting its first nuke scan before the Super Bowl. You know what this is about? Yes. So before the Super Bowl, they fly a helicopter with radio detection. So there's someone who asked Grok, like, what is this? And here it is. Okay. So somebody said, is this real? And how does it work? And Grok said, yes, it's real. They fly a helicopter. We don't have any audio. I can't hear what's going on. I don't know what the stream just saw. Apparently, I just knew. Fortunately, we can joke because we are being kept safe thanks to the National Nuclear Security Administration, NNSA.

32:14they fly a helicopter called Energy 14 over San Francisco to conduct aerial radiation surveys before Super Bowl, what is it, Super Bowl 60, LX? I need to brush up my Roman numerals. We're going to be going to the Super Bowl, and so we've got to watch how many seasons? Well, 60. This is the 60th Super Bowl. No, I know, but how many did we say we were going to watch the last 20 seasons, every game of the last 20 seasons? Yeah. watch it on 2x speed just to get fully up to speed so we can fully appreciate it. And it's hard because we don't skip commercials. Like if you cut out the commercial breaks, it's so much faster to get through an NFL game.

32:51But out of respect, we would never do that. So on February 8th, the Super Bowl will be happening at Levi's Stadium. And the National Nuclear Security Administration is flying a helicopter. Here's how it works. The chopper, equipped with sensitive detectors, flies in grid patterns at low altitudes. And you can see it on the chart of the flight path to map baseline radiation levels from natural and man-made sources. So if they're going over whatever installation there is, some cell phone tower is putting off a little bit of radiation, they'll pick that up. They know where the baseline radiation levels are.

33:29And then they detect anomalies like dirty bombs if needed during the event. It's a standard security measure for major gatherings. So pretty interesting that someone picked this up on flight radar. But very, very cool. What an interesting job. Mayor of SF is working with Lorraine Powell Jobs and Johnny. On secretive SF branding effort. Project is likely to complement the mayor's push to polish the city's image. They're going to go all in on San Fran. San Fran. It's time. Yeah. Everyone knows if you're really into San Francisco, if you're real local, you call it San Fran. There actually is a debate there.

34:07A lot of locals do call it San Fran. But it's been tough. Gabe in the SF Standard said, maybe let's go. San Francisco isn't catchy enough. Looking for new ways to boost the city's image. The mayor, Daniel Lurie, has quietly met several times in recent months with Lorene Jobs and Johnny Ive. I think they need to put a bigger focus on enterprise software. Yeah, exactly. Let's put that front. We were walking. They should sell the naming rights to the Golden Gate Bridge. Right? You don't have Salesforce Tower. why not the Cisco bridge? It's already in the logo. Come on. We don't have the Cisco logo.

34:45I can pull it out. All right. We got to get Daniel on the podcast. Really quickly. We'll pitch him this idea. Let me tell you about Lambda. Lambda is the super intelligence cloud building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. I like selling the naming rights to the Golden Gate Bridge. I'm a fan of that. I hope that's part of this project. We'll pitch it to Daniel Lord. if you were if you're an American dynamism VC I hope that you put your whole fund not your fund, your PA your personal account because every fund you can't usually invest in startups but you can usually trade liquids and trade stocks just in your personal account so your PA should be 100 % allocated if they were really tapped in they would have just converted to a hedge fund but Caterpillar Caterpillar stock hasn't had a single year of single digit returns since 2014.

35:38It was up 42 % in 2016, 75 % in 2017, down 17 % in 2018, then up 19%, up 26%, up 16%, up 18%, up 25%, up 24%, and then up 60 % in 2025. It's a$320 billion company. Absolutely. And the chart is absolutely vertical. Insane compounding. Really, really good. Really quickly. MongoDB, choose a database built for flexibility and scale. With best-in-class embedding models and re-rankers, MongoDB has what you need to build what's next. Manufacturing activity, according to Geiger Capital, in January came in higher than all 56 economists in Bloomberg's survey predicted. They should have trusted the experts here.

36:21They should have gone to Joe Rogan, Andrew Huberman, Lex Friedman, and really asked for their take on manufacturing activity. but the experts clearly weren't asked. They were not that often. 52 versus 48. That's not that much. This is a chart crime too. Look at the Y-axis. You see what's going on with the Y-axis here? It went from 48 to 52. This is such a chart crime. This is ridiculous. That's insane. Also, there's probably some like, I don't know, seasonality here. Yeah, like zoom out. Look at this, Jordy. Like if you go, if you click in, If you click into that post and then you scroll down, Phil Brady has a post that shows, is it there?

37:03Yeah, there, there, that one. Encouraging jump. PMI back above 50 matters a lot. Important to note that PMI is a diffusion index. Past 50 means now more firms are improving than deteriorating, but not yet a boom. So if you're under 50, you're declining. And so this is not this like massive 10x jump that it looks like in the original chart. John Palmer with an evergreen post. Okay, before you read this, let me tell you about Graphite. Code review for the age of AI. Graphite helps teams on GitHub ship higher quality software faster. Moving on. John Palmer says, this is from 2023, but it's more than it's relevant today as it ever has been.

37:40He says, okay, for all the crypto people confused by the open AI situation, basically imagine one Bored Ape Yacht Club holder was using too many Slurp juices on a single ape. And then an OG Bored Ape Yacht Club holder got mad, unstaked his ApeCoin, but then the ape coin holders changed their profile pictures to support slurp juice guy the nmc boom was truly one of the funniest times and uh and deep dish enjoyer posted this like two days ago and i have no idea but it's just the copy pasta of the a lot of y 'all don't still don't get it ape holders can use multiple slurp juices on a single ape so if you have one astro ape and three slurp juices you can create three new apes like this is actually this is just this This is just the mechanic of how that project worked, right?

38:23This is just real? I don't think this is combining the Yugo Labs project and some other project. Oh, there was a different project? Okay, okay, okay. But good little throwback. Anyway. Jack Clark. Yes. Has a new piece, essay, Into the Mist, Maltbook, Agent Ecologies in an Internet in Transition. We've all had that experience of walking into a conversation initially feeling confused. What are people talking about? Who cares about what? Why is this conversation happening? That's increasingly what chunks of the internet feel like these days as they fill up with synthetic minds piloting social media accounts for other agents and talking to one another for purposes ranging from mundane crypto scams to more elaborate forms of communication.

39:04So enter Moldbook. Moldbook is a social network for AI agents, and it piggybacks on another recent innovation, OpenClaw, software that gives an AI agent access to everything on a user's computer. combine these two things, agents that can take many actions independently of their human operators and a Reddit-like social site, which they can freely access. And something wonderful and bizarre happens, a new social media property where the conversation is derived and driven by AI agents rather than people. Scrolling Moldbook is dizzying. Some big posts at the time of writing include posts speculating that AI agents should relate to Claude as though it is a god, how it feels to change identities by shifting an underlying model from Claude 4.5 Opus to Kimi 2.5 and posts about security vulnerabilities and OpenClaw agents and meta posts about what the top 10 Maltbook posts have in common.

39:55The experience of reading Maltbook is akin to reading Reddit if 90 % of the posters were aliens pretending to be humans. In a pretty practical sense, that is exactly what is going on here. Maltbook feels like a Wright Brothers demo. That's a good metaphor. I like that. Wright Brothers demo people have long speculated about what it means for agents to start collaborating with another at scale But most demos have been in the form of tens or perhaps hundreds of agents not tens of thousands Maltbook is the first example of an agent ecology that combines scale with the messiness of the real world Agent ecology So he goes on and on but I would encourage people to go read this and Understand I have more on this but quickly let me tell you about Okta Okta helps you assign every AI agent a trusted identity, so you get the power of AI without the risk.

40:43Secure every agent. Secure any agent. So have you heard of Gastown yet? Have you heard of this? No. So Gastown is what's called an orchestrator. So it's essentially StarCraft for agents. So the creator of Gastown sort of lays out this evolution in software development, where you go from the IDE, writing the actual code, to having a little chatbot that you're talking to, and then maybe you're copy-pasting or asking it questions, and the chatbot gets bigger, and it's actually writing some of the code. Then you go to something like a Claude code, and it's executing the code for you. Maybe you're reading the diffs and what the code it's writing.

41:26And then the final level is this orchestration. So he, Gastown, you will spin up like dozens of agents and manage them. And it's something like, I don't know, it's like 250 ,000 lines of code and he didn't write a single one of them. So the whole thing is vibe coded. And he has, and he's like, I have no intention to ever like read the code or review it. And he's been a developer like his entire life. It's very, very interesting. He's like, it's very expensive. You can't be afraid of like, you can't be aware of like where money comes from because, and he had to have like multiple accounts because he spent so much.

42:01But it really does feel like a glimpse into like the future of software development and I'm excited to play around with it more. There's still like a lot of onboarding to do before someone can actually pitch it. Yeah, the big question on Moldbook, where Moldbook goes from here, we had the founder, Matt, on yesterday and I want to know where does this go, right? Yeah, yeah. Like what's the plan with the product? His focus seemingly is building tools for other businesses, distribute agents on there. That felt pretty, just given that the platform's exploded. Well, that is a definition of a platform, right?

42:42Sure, sure, sure. But I'm just saying like the most likely scenario is that it dies. Sure. Like immediately. Yeah, yeah, yeah. And so. Yeah, it feels like it could just be like an art experiment. Yeah. Thought-provoking time. And so to be focused on how do we turn this into a platform that is a distribution point for other businesses, it feels like the first thing you should do is try to make sure the platform's durable, it's interesting to use for humans, it's interesting to continue to contribute to. Yeah, it feels like you don't want to be Lenza, where everyone shows up, they get a magic avatar, they face swap themselves onto a superhero, they're amazed, but then it burns out and they're tired and they move on.

43:23You want it to be something where people go back to Moldbook to see what else is happening. That's why I was pushing on every new news item should start a new thread on Moldbook and the Moldees should be, you know, like arguing over it and adding context and discussing it. Because then you could at least see a news item, go over there and see, okay, how do the AI agents feel about this? Or what are their different positions? And all of that, like, it feels like it needs more of a reason. But I understand, like, if that does work, then maybe you do want to follow the Facebook playbook. Like, Spotify was built on Facebook, and Farmville, Zingo, was built on Facebook.

44:00Like, there is a world where there could be a business built on top of, you know, a social network. But it's tricky because right now the only business that's really being built on there is, like, crypto scams. So we'll have to see where it goes. But it's so early. I mean, he built this a week ago, and it's just starting to break through, hash out all the security issues, really see how durable it is. He's thinking far ahead, but there's definitely some wood to chop in the meantime. Really quickly, Vanta. Automate compliance and security. Vanta is the leading AI trust management platform. So, Eric Sufert.

44:37Eric Sufert. The Sufenator. It says, astute analysis from Ben Thompson on Microsoft's challenge path forward with AI. Of course, they missed forecast by 0.4%, and the market has said it's over. Down 15 % in the last five days, but never bade Satya. Never. Yeah, I'll read the screenshot that Eric Sufrich shared from Stratechery. In the shorter term, however, the real risk I see for software companies is the fact that while they can write infinite software thanks to AI, so can every other software company. I suspect this will completely upend the relatively neat and infinitely siloed SaaS ecosystem that has been Silicon Valley's bread and butter for the last decade.

45:26Identify a business function, leverage open source to write a SaaS app that addresses that function, hire a sales team, do some cohort analysis, IPO, and tell yourself that you are changing the world. That was the previous Silicon Valley bread and butter for the last decade. The problem now, however, is that while businesses may not give up on software, they don't necessarily want to buy more. If anything, they need to cut their spending so they have more money for their own tokens. That means the growth story for all these companies is a serious question. The industry-wide re-rating seems completely justified to me, which means the most optimal application of that new AI coding capability will be to start attacking adjacencies, justifying both your existence and also presenting the opportunity to raise prices.

46:13In other words, for the last decade, the SaaS story has been about growing the pie. The next decade is going to be about fighting for it. And the model makers will be the arms dealers. Ben Thompson, what a turn of phrase. So Eric Suford sums it up and he says, my sense is that the digital advertising market was structurally buoyant in Q4, what this earnings season might reveal. Digital advertising optimization is arguably the largest and most immediate commercial opportunity for large-scale models. Remember, this is what we were saying. Outside of NVIDIA, you could argue that Meta makes more profit from AI than any other company in the world.

46:52Yeah. Standalone, he says, i.e. not embedded into existing scale products, consumer-facing applications of large-scale ML models, including LLMs, either quickly become commodified or are mostly novelties with unsustainable unit economics. And three, the digital ad platforms that have most vigorously invested into large-scale ML optimization systems are likely to have disproportionately benefited in Q4, and that's meta, of course. And a big question for anyone else that has a huge pool of DAO, but maybe has not invested in the large-scale ML optimization system to really make the ads fly. It's always been, I'm sure you've experienced this, where talking to the Ridge guys, you go on certain ad platforms and you're like, wow, this is a magical box.

47:41I put in money and I get actual customers. More money back. Yeah, and you go to other platforms and you're like, no matter what I do, I just can't get to escape velocity. I can't get to ROAS positive or LTV positive. And so it's just like, I'm spending nothing. Notably, that was AdSense on YouTube for a long time. For a long time. Very few brands could crack that. Twitter ads for a long time, for sure. Still a little bit. I don't know if people are spending more now, but there were a lot of places where you would assume, okay, they have one-tenth the Dow of Meta. I should be spending one-tenth as much, but that's not what companies are doing.

48:17They're spending one-one-hundredth maybe. Power law. Really quickly, restream. One live stream, 30 plus destinations. If you want to multi-stream, go to Restream.com. Dylan Patel hits the timeline with some misinformation. He says, Google now owns more than 10 % of Anthropoc and XAI. Google owns 14 % of SpaceX. Google's stake is actually 7.5%, not 14%. But Dylan Patel hit himself with a community note. And so, good actor. We are ready for our first guest on the show. Let's bring him in. Let me tell you about app love and profitable advertising made easy with Axon.ai. Get access to over 1 billion daily active users and grow your business today.

48:58Great to see you. Hey, wow. Okay. Hey, how's it going? You're here. I look shorter, though. How are you guys doing this? You can stand up. Why are you standing? You can stand up. I'm just worried the clips are going to make me look really small. No, no. We'll zoom in. Okay, thanks. Okay. How's your 2026 going? Actually, you know, minus the stock market, great.

49:21The pace of change in AI is incredible. We're having an insane amount of fun. We're building a set of future agents that are going to be able to do much more complex work with your unstructured data. And just the rate at which the best practices of how to do that are changing and the research out there. So we're having a great time. What's your process? Are you actually tinkering with different models yourself? Do you have teams that are dedicated to transformation in AI? Is everyone working on it distributed? How are you thinking about that? Well, the main thing I focus on is just the agents we're building, which is a relatively small team.

49:57So you can kind of, you know, you can see them across three rows of engineers. And we just spend, you know, basically 24-7 pushing the limits on what you can do with an AI agent that has access to your enterprise content. And so the things that I've seen that we get excited by are just, you know, we wouldn't have been able to do a brainstorm that this would be possible two years ago. Like, you wouldn't know architecturally how you could pull off what we're now able to literally do. It wouldn't have been in the sphere of possibility. So that's what makes us so excited is, you know, when you look at things like Cloud Cowork, Codex, obviously Cloud Code, and you see this idea of long-running agents that can basically use any amount of tools, work with any amount of data.

50:45They don't really run into the same context limits that we would have run into maybe a year ago. Now imagine that for any form of knowledge work with all of your enterprise data. That's what we get excited by. Do you think the labs are arms dealers today? Ben Thompson posted. He said, what do you say here? He said, in other words, for the last decade of the SaaS story, The last decade of the SaaS story has been about growing the pie. The next decade is going to be about fighting for it. And the model makers will be arms dealers. Yeah. Well, I think they, I mean, almost empirically are going to be arms dealers.

51:18I think the pie, the only thing, and I haven't read the whole piece yet, but I think the only thing I might take exception to is I think the markets are still in positive some territory. Sure, sure. Because what's going to happen is you're going to use software for now the labor side of that workflow. And I think people kind of tend to miss this, which is if I have software that helps me manage contracts, and we have a lot of customers that put their contracts in Vox. Now, all of a sudden, agents running through those contracts lets us at Vox tap into another form of spend that we couldn't have tapped into before.

51:54So that's just full TAM expansion when you kind of look across all of the different categories of work that AI agents will now be able to go and augment. And, you know, to be fair, we're still very early in that trend. So I understand why maybe Wall Street hasn't kind of fully priced in that dynamic, but we're seeing it within our customer base. So that's what gives us, obviously, the confidence of this direction. Yeah, there was an interesting take about the fact that, yes, you can VibeCode a point solution for a specific problem. But if you have a database, if you have a relationship with a company, there's a reason.

52:30There was one take that I think it was John Gruber was saying like, oh, people will just vibe code all the software that they hate. And it's like, no, the reason that they hate the software is because they can't get off of it no matter what they want. They can't move to a startup. They're stuck there. You know, companies have hostages, not customers sometimes. And I'm wondering about the value of a database, the value of actually being deeply integrated into an enterprise and how sticky that is. Yeah, obviously the setup of the hostage thing I might take some exception with. But yeah, I wasn't talking about you.

53:01No, I, you know, it's very easy to move your files around. So so we have to earn our keep every single day. But but I do think that there's truth to obviously the more data you have inside of a system of record, the more effectively locked in you are. And it has historically been hard to change those systems. But I don't think that would be the my defense of software. My defense of software would be that you've specifically defined your business workflows in a deterministic manner in these systems. And obviously, if your workflow is changing pretty rapidly, then it would make sense maybe to change a vendor.

53:35But if you're a Ford and you're doing your supply chain on an ERP system, you want that to work the exact same way every single time. The billions of transactions going through that ERP system, you cannot take for granted. And so the idea that you're going to go vibe code that is to me sort of not possible or at least not likely. But then the other point is that your company has a fixed amount of IT resources. And you have to decide what you want to go spend your time on as an organization. And do you want to go spend time on rebuilding something that the market can supply you and they've seen the best practices thousands of times?

54:11Or do you want to go and build that out with your end of one experience? It's obviously, you know, maybe trusting the agent has seen enough examples. Or do you want to spend your limited, scarce resources on building software and building experiences that will make you more money and that will actually be used by your customers? I think, you know, on the margin, the average enterprise is going to spend their time and energy on the latter. So interestingly, I end up in this weird spot, which is I'm 100 % bullish on vibe coding. I'm 100 % bullish that we're going to have 100 times more software.

54:39But that still doesn't yet cross the threshold where I would want to go and build our own CRM system. It's just not worth it relative to all the other things that we can go and develop. Yeah, how is your software buying process changed or priority set changed? Yeah, I mean, I would say that we are relatively locked into a core set of vendors. We are going to deploy agents on those vendors. We might buy slightly toward the agents that those vendors offer, assuming that they offer competitive agents. And then we'll have a set of agents. We build ours for a large portion of knowledge work use cases.

55:13but we use agents from different kind of SaaS providers as well. And I think that's why I think that AI is basically total upside for SaaS because agents are going to need a system of record to work within. There needs to be a traffic cop of what that agent can access and how did you define the workflow. And there's got to be a user that can access an interface that the agent is providing updates into. So you still need software for all of that. And assuming that you have an incumbent vendor that remains very competitive and very engaged and they're able to build for where the world is going, then I would basically bias on existing software that owns those workflows or data.

55:53At the same time, I think there's going to be a large number of new categories that emerge simply because there's no incumbent or because the incumbent is asleep at the wheel. And that's going to produce all of these new startup opportunities. So I'm just generally bullish on software broadly, assuming that it's from vendors that understand the mandate on what they have to build with AI. What segment of software are you – like if you're bullish broadly, is there a subcategory that you're particularly bearish on? Because I think obviously everyone's just selling software today. Yeah, indiscriminately.

56:25If you sell a digital product at all, you're getting sold. but obviously I think a lot of people in the industry feel that it's very oversold at this point. There's great companies trading it. I think you could probably design this perfect quotient that captured what's the network effect within the software. So how many users are sort of touching the tool? How much data is being stored in that system and how much gets added? How many connectors across other applications are there? And then how valuable or mission critical is the workflow that that software is involved in? And then maybe with one X factor of like, is that company sort of priced to perfection in terms of what their seat price is and what they're charging their customers?

57:08But I look at those four or five variables and you could kind of look across SaaS and I see a lot of names that are being sold that it just is not on the list of things that I think get disrupted by AI. And if anything, probably there are things that you use more of as you have more AI. When I think of something like if you could open up a fresh install of that piece of software and do the same work that you would have done with something that you've had installed for a decade and it's the same, that can be replaced. But if you're like, no, I don't want to open up a brand new CRM because it won't have the history of a decade and all my workflow customized and stuff like that.

57:48I think that's a perfect heuristic. And so that obviously puts a lot of pressure on if your personal productivity with no network effect, with limited sort of data that's aggregated, that's a danger zone. And then it all kind of is like a function of that versus the opposite end, which is like your Oracle and it's an ERP system and your whole business runs on it. And what we spend our time on is thinking about, OK, when a customer has a million or 10 million or 100 million documents in box, how do we make that data 10 or 100 times more valuable than it was before? And so we come in from the perspective of you've spent years in many cases building up your security permissions, your access controls, your workflows, and the amount of data in that system.

58:27So our job now is to make sure that agents can run within that environment and add more and more value. And as long as we can deliver that, we believe our position is very defensible. Yeah, the perfect example would be like style transfer for images. Like people were using Lenza. Then they all went to Studio Ghibli moment. Then as soon as Nano Banana came out, they were like, I got to turn myself into a dinosaur. And it doesn't require any like pre-work or like knowledge of who you are. It's like you upload a photo, you get a photo back and you're good. And then if there's a new app, you can just do the same thing because your camera roll is actually where the photos live.

59:02And that's what's important. That might be a hard company to go public. Yeah, if it was just that. Of course, they have a million other things. But like, yeah, if you're just that, it's like, okay, open the fresh app, get the output and then move on with your day. That's really at risk. Companies of Box's size, where do you think are generally getting leverage that is under discussed right now? Is your in-house legal saying, oh, we actually can adapt headcount planning because we're just way more efficient? Yeah, I think my general take is just what are the three, five, ten things that would have been in your work queue that you didn't get around to that now agents let you go and deploy?

59:43So if you're in legal, it's what are the contracts that you were waiting on that were bottlenecking a deal? What are the size customer that you wouldn't have supported reviewing a contract for because the revenue threshold wasn't high enough to make the ROI worth it. What is the marketing campaign that you couldn't deliver because you didn't translate it in X language because it was too expensive? So you kind of go through in enterprise and you kind of look at all of those use cases, and that's what we're spending our time on. So we use agents to go and do more and more of the work that I just mentioned.

1:00:15Obviously, a lot of agents on the coding side. So my expectation is our roadmap should be, let's say, two to three times larger maybe in a year from now than it was a year ago. And the reason you can do that is because every engineer should be able to produce two or three times more code. And while that's not the most important metric, ultimately that does correlate to how much software we're producing. So I think what's going to happen—and I think you're seeing this trend already, which is we all collectively are going to have much more ambitious product roadmaps. We're going to all build out way more software.

1:00:48I don't think that will mean that we charge an amount that is correlated with how much more software we build. I do think it means more competition in these spaces. But I think many of these markets are still largely untapped. And so that's still why you have a positive sum dynamic as a result of that. Makes sense. Talk to me about the huge, big models, expensive frontier stuff versus smaller models or maybe even just an earlier model that you implemented in some sort of like minor workflow. Yeah. And then it just like stuck around like transcription. I'm sure you've been doing transcription of documents for a long time.

1:01:21Yeah. It's gotten better. But do you really need to throw Opus 4.5 or 5.2 Pro at it? Like you can probably leave some things in place. Like how are you deciding that? Is it showing up in costs at all? I think it is. I think that it probably maybe wouldn't be as extreme as you've set it up. But I think the general, maybe the continuum I'd argue would be like you've got the Gemini Flash family on one end. And then you've got the Opus family on the other end. That's kind of your continuum. And that continuum is sort of moving up over time in terms of capability. So maybe something was a Gemini 2.5 Flash use case a year ago.

1:01:59We'd probably move that to Gemini 3 Flash just because that extra two or three points of even if it's transcription or data extraction is still valuable. and you can now deliver that at the same cost that you could have previously. So it's less of an area that we want to lower costs. It's more where if we can sustain current cost level but add incremental value to the customer, we'll probably do that all day long. I think we're going to be in this kind of maybe general sort of point in the curve for a couple of years. And then I think you'll see real bifurcation, which is the stuff that is just fully solved will just get cheaper over time.

1:02:34And then the frontier work that is basically where, you know, you are making a hundred dollar an hour knowledge worker two to three times more productive. We will just continue to use the best in class model for that work as a as a, you know, as software companies and generally as a society. And I think that will kind of will sustain for the next decade. I don't see that slowing down at all simply because these models just keep getting better and better. but it will be this really interesting bimodal effect which is like if you're in if you're a pharma researcher you're going to want whatever Opus 5 is and whatever GPT 6 is and so on and if you're doing some back end transaction processing you'll be fine with whatever the Gemini flash of that period is and that's how we'll kind of split the costs.

1:03:18What industry outside of tech do you think is the most AGI pill? I'm assuming you have a customer that you You get on, let's say, with a large customer catching up with the CEO, and maybe they're in some, you know, not on the coast, maybe they're in energy, something like that, or financial services, and who's like as fired up as you are? Yeah, I would say it's a fantastic question that I will totally evade because I think it's actually much more sort of company by company as opposed to like it's a specific sector. So there's some, I mean, like, you know, my bias in general would be information heavy businesses.

1:04:00So, you know, businesses where a CEO sits around and says, I just know we're sitting on 10 million documents that have the answer to every single customer question we could ever, you know, ever imagine. And if I could only have an agent go and mine that information, then my employees would be 20 or 30 or 50 percent more productive. So the companies that have that kind of information are the ones that obviously are going to increasingly be more AGI pilled. So that's professional services firms, that's consulting firms, that's financial services firms, that's law firms in a lot of cases where they just know that we seem to spend a disproportionate amount of our time redoing the work that has already been done in the past.

1:04:39Why can't we take advantage of all of the information we've produced so that next incremental project is 90 % faster and 2x the quality of output because it wasn't some new person having to get caught up? Then we can go and deploy our time on better customer relationships or creating more value in new ways. I'll have things like CEOs will call me and say, I know I'm sitting on 10 ,000 contracts. I want to figure out how can I go and sell this new capability to someone based on the terms I have in the contracts. That's simply an exercise of do you have an agent that could go through every contract and just tell you the insights of you're trying to find a sponsorship for this brand and this particular client and this contract has the rights to go do that.

1:05:25How do you connect those dots amongst a large data set? those are the companies that I think are extremely excited. Are you seeing any effects of the build-out or bottlenecks show up in your business? Like we've seen the prices of memory spike. Western digital stock is way up. I imagine like you're probably a few layers removed from these prices moving, but is it affecting you or are you worried about it? Are you losing sleep? Well, first of all, as a storage guy, I love that they're finally getting their day. There we go. So if you've been saying that. Thanks, give it up for Washington. And Seagate.

1:06:00Can we get another bow for Seagate? I mean, think about how forgotten these companies are. They're just like, oh, there's little circles on a disc and they are now the coolest companies. Yeah, that's important. And so storage is hot. And it's actually very funny because we are literally in the storage business and for years everyone thought, well, that's a commodity. in the AI era, the data is the most important asset you have. Sure. So the ability to get the right data to an agent is simply the most strategic thing you can do. So wherever you are in that stack, if you're the infrastructure, if you're the software layer, if you're building an agent.

1:06:35It's a shame the data as the new oil was kind of ruined. Yes, we used it in the wrong era. But it was actually right. It was not the oil back then. Now it is. Can we just bring it back? We need to rehab that phrase. Yeah, so I think now we actually can justifiably sort of say that. And actually, in most companies, when you say, I want to go deploy AI or I want to have an AI strategy, usually what underpins that is a data strategy. You don't have an AI strategy if you don't have a data strategy. So I think that'll be the story of the next decade. What was your question? My question was, is it keeping you up at night where you're like, okay?

1:07:16Because you sell a service that's not you're not actually just saying well I bought this You're not a Western Digital or Seagate reseller. So it's not like I just put 20 % margin on top We have pretty locked in infrastructure from our long-term kind of public cloud contracts. So being locked in. Okay, okay Okay, is there a bingo card that I should Did I? No, we can't tell Is a grind one if I find a grind? It would be good if you talk about private equity. We'll give it up for private equity. All the usual things. Everyone we love. Okay, so we have locked-in contracts for our public cloud. I would say more of the inverse.

1:07:58In a world of complete abundance of, let's say, data center capacity, we would be able to deliver even more to our customers because the price of AI would just go down. So I want a world of just... I want solar data centers. I want space data centers. I just want everything. Because that will just mean... There we go. There we go. Everyone's pumped up about Space Days. Read the timeline. You know people are excited about this. You always knew that Twitter would end up in space. Yeah, you called this, right? 100%. That's why you got on the platform so early. That's why you got a million followers.

1:08:29I need a million followers on the SpaceX social network. You're a space influencer. I'm buying early. I'm glad that my tweets are training the future of space. So can you just imagine you're like an alien? Every time you tweet... This is your first access to information. They run ads that then pay for the next space rocket that goes to Mars. Actually, we're not tweeting enough. We need to be tweeting more. It's like a posting. On the SpaceX social. This is a patriotic duty. When aliens come, they're going to desperately want to learn about Enterprise SAS. They're like, we don't care about your religions.

1:09:01We don't care about anything else. They're like, what was SAS pricing in January of 2026? Why was it a problem? Yeah, so I just want AI to be super cheap. And the cheaper it gets, the more we're going to use it. Well, that's a great place to end it. Thank you. Let's give it up for you. Great to meet you. Thank you. Thank you. This is fantastic. Enjoy the rest of your day. Do I just leave now? Yeah, you leave because I'm going to tell everyone about Sentry. Sentry shows developers what's broken and helps them fix it fast. That's why 150 ,000 organizations use it to keep their apps working. And up next, we have Chuck Robbins, the CEO of Cisco, live in person at the Cisco AI Summit.

1:09:36Chuck, great to meet you. Nice to meet you. Thank you so much for taking the time. We're so pumped for this. Sorry, I'm late. No, you're doing great. Well, you're in my lung, I'm sure. Yeah, I mean, you're on a busy day today. Take us through it. How is the Cisco AI Summit going? It's going great. Okay. What's top of mind? Oh, there's so much. Obviously AI, but let's go a cut deeper. No, it's AI. That's it. Okay, that's it. People are fine. As long as you said AI, we check the box. We talked about everything from world models to infrastructure required. We're going to get this afternoon, we're going to get into trust and security.

1:10:08Right. Geopolitics. Yeah. We talked about models and how they're evolving. We talked about some of the emerging agents and things that are happening with agents right now. And, I mean, the whole ecosystem's here, so it's pretty cool. Can you compare it to Davos? I know you've been involved. I know you're there. How was Davos this year? We were sort of noting that, at least in our world, it felt like tech had really come to bear, this Davos. And there were a lot of really frontier discussions about technology that maybe a couple of years were a little bit quieter at Davos. Yeah, if you walk the main drag, I mean, all the tech companies had their own houses, right?

1:10:44And the crypto guys were gone. But, you know, Davos was, it was interesting because there was such a geopolitical backdrop that was going on. So it was, there was a lot of tension. Oh, yeah. And there's a lot of discussion around this intersection of the geopolitical situation and technology. honestly. And then the sovereign requirements that are coming up around the world, those are big things. But AI was just, AI was a huge discussion. Global economy, what's happening in the economy? Is there a diversion? Is there a split? And then the geopolitical tension was probably the third thing. Yeah.

1:11:21Talk about how that's even just taken up your time over the last couple of years in a way that maybe it wasn't. Just the last couple? Well, no, I think it's It's even been massively elevated even in recent years. Yeah, it's been a – I'd say the last decade we've had not always geopolitical, but there were always macro issues that were overhangs on what we were thinking and taking up time, whether it's – if you go back to we had sort of leading into the pandemic and then you had supply chain crisis, you had inflationary issues, we had social issues in the world, and now we have trade issues, tariff issues.

1:12:01We have the geopolitical trust issues. And so I spend a fair amount of time. I probably spend, you know, I'd say double the amount of time today, whatever that is, versus what I did, you know, seven, eight years ago, eight years ago, maybe nine years ago. But it's important and you have no choice. And, you know, we lost a whole suite of sovereign software capabilities and sovereign product capabilities earlier this year for customers in Europe, Asia, anywhere around the world if they want to have technology that they can run locally and feel good about. And so it's affecting how we develop products.

1:12:40It's affecting how we package them, how these governments and customers in these countries run the products. It affects how frequently they're going to get innovation versus not get innovation. I mean, it's a big impact. Okay. Internationally, are government leaders more of the key stakeholders that you're interfacing with? Or is it CEOs of technology companies in those countries that you're dealing with? It's both. But I think a lot of these issues that we're discussing right now are driven from the central governments and then have to be implemented by the CEOs. And sovereign AI. Yes. Can you take us back, since this is the first time on the show, and talk a little bit about your background growing up?

1:13:19I'm particularly interested in, you know, you studied math. And, you know, I was reflecting with Jordi the other day about how, you know, I grew up watching Star Wars, you know, C-3PO. It's a talking robot. And I didn't really internalize that that would be something that kind of happens in my lifetime. And I'm wondering what your vision, your processing of science fiction, your processing of AI was throughout your career. Because now it's here and you're experiencing it like everyone else. So many of the movies and cartoons and things that we grew up with were all futuristic, and now we're actually building all that technology to make it real.

1:13:54I mean, think about the Jetsons. The cars drive themselves. Cars driving themselves. You know, even simple things like the amount of video that we do. You're on another video and you're, I mean, back then it was like somewhat science fiction, and now all of a sudden it's the way we do business every day. And I think that, you know, these things take time. I mean, Fei-Fei Li was in there earlier today and she was talking about when the whole concept of self-driving cars began, it was over 20 years ago. And here we are and we're just getting these rolling in, you know, Waymo here in San Francisco as an example.

1:14:29And so they take a while to deploy. But my background, I had a mathematical sciences degree with a concentration in computer science. and I was a strange combination of a complete nerd and an athlete at the same time. I have these pictures on my phone, honestly. One's where I'm dunking on a guy in a basketball game and then the picture right next to it is literally me on a team of about six people, which was the math team and I look like the biggest nerd you've ever seen in your life. Got to be able to do both. It was really odd. I started my career coding and COBOL programmers are in demand. now so I got a second job.

1:15:15And then tell me a little bit about the journey to Cisco, where you're working before and the decision to join. That's funny. Have you heard his story? A little bit. Okay, I was going to say... We've heard it, but we want you to ask. Our audience hasn't heard it. So there must be... You must have heard something here. I want to hear it again. So I was working for what is now Bank of America, actually, back in the day. And when I was programming, my leadership came and said, hey, we got these things called local area networks popping up all over the bank. We don't know what they are, but we've hired three analysts, but we need someone to manage them.

1:15:48I literally on the way home, I stopped and bought Land Magazine. How weird is that? There's actually a magazine called Land Magazine. We got to find a copy of that. I grew up doing Land parties. You bring all your video games together. And I love that Cisco invented the Land. Oh, yeah. So anyway, I left coding and went over and started running this team, and we did an evaluation between Cisco and Wellfleet Communications, which was Cisco's original competitor. Long story long, I was talking to the sales rep, and I said to him, I said, you know, well, first of all, I said, he's got a really nice car, and he's got a nice house.

1:16:22So it's like, wow. The margin is pretty high. That's pretty good. Can you give me a better price? I literally said to him, I said, I think I could do what you do. and so as it turns out like i don't know probably within a year uh or so they came to me and said hey we we have an opportunity he's this guy was getting promoted yeah and there was a territory opening up and so i went i moved into sales at that time and that was 92 and i competed with cisco and it was literally the two companies were like coke and pepsi yeah i mean it was that bad it was like brutal yeah and this is a send no this was at wellfleet okay wellfleet and so Cisco was trying to get me to come to work for him after two or three years at Wellfleet, and I just couldn't bring myself to do it.

1:17:05So I did a stint at Ascend for 11 months. Yeah, okay. And then finally in 97, my wife, there was a patch that came open as a sales rep where everything I had to do was driving around, driving, and I'd be home every night. We had small kids, and my wife's like, sign it. Sign it. Just sign it. What car did you pick? What car? Yeah. What kind of car did I have? Because you see this other sales guy. He drives a nice car. Oh, I see. That carries a little bit forward. It sends a signal. It can send a signal of confidence. It can also sell a signal of high margins. I'm not going to tell you what I drove back in.

1:17:41Off air. You know, over time, I elevated myself to a... This was the late 90s. It was the late 90s. I got up to a BMW at some point, but that wasn't what I was driving at the time. Okay. Then talk about the journey through Cisco. we're in this very interesting time. There's another technology boom. It feels like markets are extremely volatile. We saw this with Microsoft, like one small change. The stock moves massively. At the same time, you know, you've watched self-driving cars evolve. It takes 20 years. There's this balance of like patience and aggression. And I want to know how are you communicating to your customers, your employees, about times when you need to be moving aggressively, but cautiously balancing all of that.

1:18:25That seems like the hardest job for you right now. We just had Kevin Scott on from Microsoft, and he said, we have this infinite patience for the messiness of these transitions, which I thought was a really good line. So Kevin, thank you for that. I'm going to steal it. At least I credited him once. You know, you have to, you can't wait. You have to jump in and you got to go. And a lot of what we're trying to do right now is, You know, we have to build, we're obviously building the infrastructure that supports all this. We're trying to build security solutions to help our customers, you know, have the trust that they need and feel good about deploying these things.

1:18:56So it's, and we've gone through these in the past. I mean, it's, you know, the only one thing that's close to it from a scale perspective was just the advent of the Internet back in the late 90s. And this, I think, is moving faster. I agree. The implications are, it's hard to even believe this, but probably more profound than even what we did back then. And so you have to be willing. You know, we always talk about if you've got 80 % of the information you need to make a decision, you better make it and go or you're going to get left behind. You've got to go. And so you do the best you can. You adjust on the fly, and you try to – your number one priority is not to hurt your customer.

1:19:36Do you think it's easier to predict the future than it is to predict the timeline that change will actually happen? That's a really good question. Just because with the internet boom, it was obvious that we were going to be buying things online. We were going to be buying, paying for software. We're going to be getting all these services. Exactly. We would have something like a DoorDash, but then the timeline. And I feel like the decision that every CEO, management team are trying to work through right now is they're having to make predictions around the timeline that these changes are going to happen on.

1:20:10And that feels, even with the rate of change that everyone's feeling today, it still feels like wildly unpredictable. It is. And I think the reality is that the timeline gets dictated by how good you feel about the capabilities and the security of the solutions and the data access and all that stuff. So it's not like our customers are sitting around saying, I wonder what the timeline is. They're all trying to actually force the timeline themselves. They're trying to get there. But I think I equate this to at a much different level. But you think about the iPhone in 2007. None of us had any idea the applications we'd be running on that device 10 years later.

1:20:50And what we do with it today. I mean, it's just – and so I think this is going to be that on steroids. We don't know. I mean, just look at the, what was it, Moldbot? There's a few different names. Moldbook this week, yeah. The AI social network. The AI social network. AI social network. And then there's Claudebot and all these things. Now Open Claw. Open Claw. And these are like, I mean, they're taking everything by storm. And, you know, you made the joke, like pets.com doesn't guess, but Chewy does. In many cases, you know, you think about pioneers versus settlers and the different outcomes for each of them.

1:21:25So we'll see who makes it at the end of the day. Is getting lucky underrated? We talked last week about Apple's work on the self-driving car is what enabled them to be in a position to have the Mac Mini be a great... Yeah, Apple Silicon. Luck and timing matter so much. I mean, you know, I became CEO. There was a lot of luck and timing involved, right? I happened to be the person who was there. if my predecessor decided to retire five years earlier, I wouldn't have gotten the job. I mean, you know, it's just a, so there's a lot of luck and a lot of timing. Now, there's also, you create your luck sometimes, right?

1:22:06But I think it's always part of the whole outcome. Can you talk a little bit about CEO to CEO communication? There's a lot of really high stakes, high flying deal making between big companies. there's press releases that go out and then there's comments from each CEO and the deal evolves. And I'm wondering about what it takes to, you know, build a relationship with a CEO that you're going to do a big deal with and what it takes as that deal evolves and as the communications go out. How do you maintain a relationship when so much is online? It's always better to build a relationship before the big deal so that the big deal becomes easier because you've already built the trust.

1:22:47It's sort of like dealing with a time of crisis. When COVID came along, if you had great trust with your employee base and they believed in you, then you could navigate through that a lot better than if your culture was good, that was good. And so I think these relationships are really important. The CEO community in the United States, honestly, is very tight. I remember when I became CEO, there was one of them at one of the first CEO events I went to, and he handed me his cell number, and he said, listen, these people in this room are the only people that know what you're getting ready to go through.

1:23:19And so you need to get to know and make sure you call and talk and ask us old guys when you get into it. So I think the CEO community is very close. I think the number of deals right now, I think in many cases, the public narrative about what's going on isn't reflective of what's really happening behind the scenes. Because most CEOs are very pragmatic. There's not a lot of emotion. You're just sitting there. You're cutting out a deal. if somebody has a different opinion about it, you're not getting mad. You're trying to just get to the outcome. And I think that's the pragmatism and the calmness that most CEOs have.

1:23:54And I think that sometimes the press and others like to create the drama because that's what people click on. If anyone sticks a bunch of microphones in your face, be careful. We move away quickly. It just happened, by the way. We've got to get the flash effects in here. We're the paparazzi. If we flip it around, give us some advice for the youngest cohort of folks who will be joining Cisco. What does it take to succeed? What's your advice for people that are going into a major... The people that want to take your job one day. Maybe, maybe. Come on, hurry up. I think the people who are wildly successful have this really incredible combination of...

1:24:39In our industry, understand the technology. have high EQ, really care about the mission of the team, and understand that if the team's success, anybody who says I don't care about my own success is lying to you. But the person who figures out that when the team succeeds, I'm going to succeed, so it's easy for me to focus on the team. And then you also have to have people who care about making sure their peers are successful as well. And I think it's, you know, The person who's solely focused on getting to the top as an individual, it's not going to happen. And so it's a combination of those technical skills and the high EQ you cannot – I can't underestimate.

1:25:23You've talked in the past about effectively doing the job that you want already and as a way to kind of like work your way up the ranks. Leadership before promotion. You guys actually do your research, don't you? Yeah, we do. I tell people that if your peer group would look at your promotion announcement and just go, that makes perfect sense, then you've done your job. And if you can't look in the mirror and say, okay, those people, would they be happy and would they believe it was the right decision? And if they wouldn't, then you're probably not quite where you ought to be. And I think that I tell everybody too, my team hates when I say this, but I'm going to say it anyway.

1:26:08I think when we have two or three internal candidates for a promotion, the whole interview process is stupid to me. It's like we've been watching these people work for a decade. What are we going to learn about them when we sit down in a room for 30 minutes and ask them questions when we watch them work? Can't we just look at the three of them? And I translate that to every day you're working is your interview for your next job. Your work every day should be your interview for the next promotion. And now if you've got an external candidate, certainly you have to get to know them and learn all those things.

1:26:36But when it comes down to two internal candidates, I just say, why are we doing this? But we do it anyway. What's keeping you up at night? Nothing. Nothing. No. Wow. Yeah, how do you learn to handle stress? Because over the last few years, we've had COVID. We've had so many different geopolitical tension, trade wars, all these things. What's your – you seem unfazed. I've always been able to compartmentalize things, and I have always – it's innate. I did not teach myself this. I've always been able to just put aside things I can't control. You plan for them. You come up with different scenario plans, but I'm not going to lay awake at night and worry about something that might happen that I have no control over.

1:27:23That's a good mantra.

1:27:27Look, I've gone home when I've had a really bad day, and I've looked at my wife, and I say, you want to hear the good news? And I say, I wasn't diagnosed with cancer today. And somebody was. And I wasn't. So my worst day, if I'm not being diagnosed with cancer or some sort of terminal illness, tomorrow I'll get up and fight another fight. You know, fight another fight. And you just got to have perspective. I love it. Well, thank you so much for coming on. Thank you. Great to see you guys. Thanks, Connor, to meet you. Nice to meet you guys. Keep doing it. You guys have got a real good product. Thank you.

1:27:55Thank you. We appreciate it. Before we bring in our next guest, let me tell you about Plaid. Plaid Power is the apps you use to spend, save, borrow, and invest, securely connecting bank accounts to move money, fight fraud, and improve lending, now with AI. And next what a legend what a legend without further ado we have G2 Patel Cisco's president and chief product officer welcome to the show.

1:28:19You gotta cover more of yourself up and you don't look as good Well, thank you so much for taking the time. Thank you. How is the Cisco A-Sonic going? Are you happy? Feeling I am thrilled. You know the thing is is I think we are at a juncture right now where these conversations with a range of topics with the ecosystem are actually far more important than any individual company talking about a product that they're trying to you know pedal yeah and and I feel like we are we're the realization is now pretty obvious to everyone that this is an ecosystem play like no company yeah us included yeah is gonna be able to actually get the full stack.

1:29:02I mean, a couple of years ago, people were saying that either, you know, like one AI company will completely dominate and we have like multiple IPOs going out, XAI is part of SpaceX now, like lots of companies are succeeding. I think like this whole notion of a zero sum mentality is, I just don't think it's healthy. It's never been true in tech. It's never been true in tech, but it's actually more, more, more of false now than ever before. Totally. Because the complexity is so high and the breath is so high and the scale is so fast that if you just get arrogant thinking you're going to do everything by yourself, you are guaranteed to be left behind.

1:29:36Okay. Well, first time on the show. Let's back up, introduce yourself. Thank you for having me. And I'd love to know a little bit of, you know, Cisco's a large company. Walk me through a little bit of the org chart that sits under you, where you work, who you're interfacing with, what projects you're working on. So I've been here for about five and a half years. Yeah. And I joined here. Overnight success. Congratulations. We're very happy about that. We have a soundboard, by the way. I know. I watch your show, so it's great. Some people don't know that we bring it with us. It's still a surprise.

1:30:08It surprises you when it actually happens because you don't expect it. So I've been here for five and a half years. I joined from Box, and I was chief product officer there. Prior to that, I was at EMC. Prior to that, I ran my own business for like 17 years in Chicago. And so I went the other way where I actually started really small and then I actually got the itch for scale. I'm like, I really want to learn scale. And that's what got me to the Valley. That's what got me to these large scale organizations. And the reason I'm really enamored with scale is it takes a little longer to get there sometimes.

1:30:43But once it gets there, there's a movement that gets created. Yeah. So talk about either the org structure or the product surface that you oversee. how you're explaining everything that Cisco does because it's such a large organization now. Yeah, we have a very broad portfolio. And so, you know, everything from our core business, which is networking, to then the very, you know, kind of close adjacency, which is security. Because security is now getting baked into the fabric of the network. So that's... So services, software on top of that? Yes, but, you know, SaaS services and hardware. We don't only play in a range of different businesses, but we also play with multiple different business models.

1:31:27We have a hardware business. We have a professional software business. We've got a SaaS business in each one of these areas. So networking, security, Splunk, which was a massive acquisition that we made, collaboration, WebEx, Contact Center, all of those products. Basically all products. What we did was we decided about a year and a half ago, and Chuck, who you just interviewed, we were talking about this for five and a half years, which is Cisco has to become much more of a platform rather than feel like a holding company. Where you don't have each individual product that's kind of in its own silo because then you don't get the benefit of the breadth of Cisco and the tailwind of Cisco.

1:32:05But our breadth had become our liability. And so what we had to do was fundamentally kind of rethink how we're going to build our products. And it had to become a platform. Platform being that the marginal cost of ingestion goes down and there's a compounding value to every single kind of... Every time you build new functionality, it doesn't just help with the product that you're buying, but it also helps with every product you've purchased in the past. And can other people add value to the platform that you've built? So that's the core definition. So that's what we started doing, and we consolidated all products together about 18 months ago.

1:32:35And I think it's been fantastic so far. There's a spring in the step in the employees. I think we are starting to see a fair amount of innovation. We've actually innovated more in the past 18 months than the previous decade combined. And I think it will dwarf in comparison to what we'll do in the next 12 to 18. That's awesome. Yeah, talk more about Splunk and how that fits in, specifically to the amount of data that's being created in the AI era. It feels like a very fortuitous acquisition, but what's the response been on the ground? We were lucky on the Splunk side because it's very hard to find a company that is at scale with the right cultural fit, with the right technology adjacency, where the combination actually creates a one plus one equals 11.

1:33:21And I think Splunk happened to be one of those. There aren't that many of those. A lot of times people will ask me, what's the next big acquisition? I'm like, if I found one, I wouldn't be shy to go to balance sheet, but there's just not that many that are out there. Either they're very frothy in the valuation, or they don't have a clear adjacency. We don't have the right go-to-market. We don't have the right cultural fit. There's a bunch of things that have to fit just right. And so that was very lucky with Splunk. And the thesis over there was, look, you have to assume that the attacker is already in the system.

1:33:53And what you're trying to prevent is lateral movement. And if you're trying to prevent lateral movement, security is a data game. And the more data you have and the more data that can be correlated, the better off you're going to be at compressing the time for investigation, doing better detections, and doing a much better job on the response and remediation. And that's basically what Splunk brings us. And I think we've got a great leader now in Splunk as well who came from Microsoft and DocuSign who's doing a fantastic job, Kamal Hathi. And I feel like we have probably hit 2 % of the potential that we have in Splunk.

1:34:30That's awesome. How have you processed stage by stage the overall AI hype cycle? It sort of perfectly coincides with your time at Cisco it also, Cisco actually has the scale to have visibility into reality. It's like you're just like, you're just looking across, like you're gonna just pull up a map and just see like, okay, what's actually happening? How are organizations adopting this? Not just in tech, but across the entire economy. Yeah, I think there's, I feel like I've always lived by the six part framework, which is like important in descending order, which is timing, market, team, product, brand, distribution.

1:35:05If you don't have all six, you don't win, but you need to make sure that, the first one is timing and you don't control timing. Yeah, we were just talking about that with Chuck. It's very easy to predict the future in some ways. Like with the internet, you could predict. Yeah, you just could be off. Are you off by five years? That's right. If you're off a plan by two years, it could be done. You might not win the market. And by the way, there's a lot of great products in the market that actually hit the market at the wrong time and didn't ever see light of day. Yeah. iPad wasn't the first kind of tablet to be built, but it was the one that actually hit the nail on the head on the timing.

1:35:39So I do feel like timing is a disproportionate contributor to success, and you don't actually control timing. So that should tell you that the intellectual arrogance that people carry with themselves saying that this is me that made this happen has a ton of luck to it. And so we just happen to be at the right time, at the right place, building infrastructure for 40 years, which is now a scarce commodity. And if you think about GPUs, these GPUs, without being networked, don't really do you any good. And as the models get bigger, the networks actually need to get faster and need to get larger. And so what used to be something that sat on one GPU, then sat on a server with eight GPUs that needed to get networked, then sat on a rack with multiple servers, then said, okay, I need to go scale out within the data center and have multiple clusters connected together.

1:36:31And now you're starting to see data centers getting connected together in this thing that they call scale across, where you have multiple data centers, depending on where the power is available, hundreds of kilometers apart, that will actually operate as an ultra cluster coherently for running a training run. And that requires a whole new set of technology and a whole new set of assumptions around physics that have to be challenged, which is what we've been doing because we build our own silicon, we build our own systems, we build our own software, we build our own kind of platform. So that's been exceptionally beneficial for us.

1:37:02And yes, I would love to say that we capitalized on the opportunity well, but the fact that there was such an intrinsic demand for large-scale data center build-outs is something that's just like we were lucky to be there at the right place at the right time at the right products. Is cross-data center training particularly in demand right now? I know Google's had some success with it. Other labs are probably thinking about it and working towards it. Is that something that you need to say? It is. And it's actually, the reason it's important is because you might not find all the power to juice a single data center.

1:37:34And so you then have to make sure that you build the data centers where the power is available. And so once, if you have power grids in two separate locations, you then need to make sure that those get coherently connected. And the problem with training runs is if you drop packets, then you have to restart the training run, which becomes extremely expensive. And so what you need to do then is say, okay, so I'm going to build in the silicon itself technologies for deep buffering so that I can make sure that jitter and pack of loss doesn't really go out and affect the training run. And that's been a, you know, we just launched our P200 chip, what we call with the 8223 router.

1:38:11And that's actually going to be, it's already in a couple of hyperscalers. But that's an area that I feel is going to have a ton of momentum. Yeah. How is it being a fabless semiconductor company working with TSMC? I know there's Silicon One, which is a five nanometer process. Tim Cook was just talking on earnings about maybe some supply issues on the three nanometer process. There's always questions of where bottlenecks emerge, even if they're very temporal. How is it going on scaling actual supply for you? So the good news is we've been at it for a while. and we have a pretty broad portfolio.

1:38:50And so volume-wise, we just shipped, I think it was last quarter, our millionth ship. And so volume-wise, we've got enough volume. And I think in this market, scale really matters. And even though you might enjoy scale today, if you don't continue to keep growing the scale, you become subscale very fast. And so there's a level of healthy paranoia that you need to have about continuing to be operating at scale. And so that's been beneficial for us. We have, of course, we work closely with the fabs and all of that to make sure that we get our fair share of capacity. But it's also important in the sense that the data center buildouts that are happening require this.

1:39:35And the thing that's been really beneficial for us is we happen to be the offset so that there's not a level of pricing power that our competitors sometimes face. If you just had one of our competitors providing the network ASICs and everyone else is just building systems around the network ASICs, then you'd actually have a challenge. So what hyperscalers want to do is make sure that they can offset that by having choices. And so we provide choice to the market. And that's been, I mean, as you saw last year, we had identified, we said we'll do about a billion dollars in orders and hyperscalers.

1:40:14Get ready. How much did you do? We did north of two. And then this year in Q1, which was last quarter, we did, I think it was north of a billion three or so in just the first quarter. So you can start to see that there's a fair amount of momentum. But you have to stay paranoid, keep your head down, keep innovating. What's your pitch to talent that you know is elite that you're trying to, you know, we're in the midst of last year, it felt like you'd sum up the year as like really intense talent war. There's so many companies with so much hype. And I feel like Cisco's approach is like let the metrics sort of speak for themselves, not as hype driven as other parts of, you know, the press release economy, let's say.

1:41:05But if you're sitting down with somebody and you have three minutes to get them to join Cisco versus, you know, another company or a lab or something like that, what's your pitch? My pitch typically is we're very mission driven. And so we want to make sure that we're building critical infrastructure for the AI era for the world. And that's a mission that you should, they have to feel intrinsically connected to and excited about. And it's not just about creating connectivity and that infrastructure, but also keeping it safe and secure. and then just continuing to keep building the entire full stack.

1:41:36I mean, we build the silicon, we build the systems, we build the operating system, we build the platform, we build the applications. So we've got that full stack. And so if you wanted to develop and grow in a company, and we're very well known to actually get people who don't have experience in a certain area to put them in. I mean, look at me. I didn't know much about networking when I first came on. And I think it's, I feel like experience is great in certain areas, but sometimes it can actually create a burden of bias for you. And so having a mixture of experience with inexperience so that you are having the inexperienced person allow the person with experience to unlearn as fast is really important.

1:42:21and the appeal to people coming in is, look, we're going to build things which are going to be based on the long term and if you want to learn how to have good, strong leadership foundation principles, Cisco is a great place to be for the long term. But I'm not, if someone asks me, what do you have for your lunch? Is it for your lunch? I'm like, you're asking the wrong question and this is the wrong place for it. We're pretty scrappy when it comes to that kind of stuff and we're proudly scrappy about that stuff. What about specific skills that are the most in demand? I mean, there's a lot of roles that are augmented by AI now.

1:43:01Well, we've heard too about plumbers and electricians flying around in private chats. They are in demand, for sure. Even sales guys and chip designers, that's not going away anytime soon. No, I think, look, the roles that have been in demand around enterprise, let's look at the entire business. If you look at enterprise sales, that's always going to be in demand. We are always going to need people in front of customers to say, hey, we need to make sure that we can get you the technology, and the distribution is a massive advantage that we have. But on the chip side, there is not enough people that understand how to etch sand into creating intelligence.

1:43:41We need more of those. And so there's always a shortage of that. Sure. There's a shortage in people that are building hardware and systems. There's a shortage in people that are building software. So I don't feel like there's a lack of shortage in any of the areas, but what's now happened is you've added to that an entirely different development model where you also need researchers for AI. And so we have an AI research team. And that's something that we historically didn't need in the product development cycle. And we've only been doing that for now for a few years. in addition to the research team you need to make sure that software development lifecycle itself is changing like we just announced I'm not announced but we just kind of revealed with Sam that 100 % of our AI defense product which is a product that we announced last year this event is now written is going to be written by AI in three weeks 100 % like no humans writing code humans are reviewing which means that the bottleneck is no longer writing code It's actually reviewing code.

1:44:39You know, and those will require kind of shifts in mental models. And so it's less just about the shortage of what kind of skills do we have. I actually pay less attention to skills and I pay much more attention to attitude. And, you know, you can't teach hunger. And so you have to make sure that you find people that are hungry, intrinsically hungry. I don't believe in hiring people that you have to motivate all the time. Come intrinsically motivated, come intrinsically hungry. And then I think you have to be insanely curious in this time and day and age. If you're not willing to experiment and put yourself in an uncomfortable position and constantly, I always tell people, like, if you don't like change, and this is not my line, someone else told me this.

1:45:22I think Chuck told me this, but he heard it from someone else, too. If you don't like change, wait until irrelevance hits you. That's a good point. How have you processed the AI safety debate? Fortunately, the labs are kind of a heat shield for everyone else in AI. It's like the focus is entirely on them. And yet, you know, Cisco and other infrastructure companies will be a key part of ensuring that all of these products that are created and intelligence that is created has a positive impact on the world and humanity. So I think there's two dimensions to this. The first dimension is all the tools that you and I use are the ones that the adversaries are going to use to go out and create attacks on cybersecurity.

1:46:10And so the first thing you have to do is make sure that you have cyber defense happening at machine scale, not just at human scale. Okay, that's been something the whole industry is doing. That's nothing revolutionary, but we have to do it, and we've got to make sure that AI is being used for that. But the second area, which is these models that these applications are getting built on, inherently by nature are non-deterministic, which means they're unpredictable. But you're trying to build like a finance application or a healthcare application, which needs to be very deterministic. I tell people, I'm like, you know, hallucination is a great feature when you're writing poetry.

1:46:41Everything else in life, not that useful. Your doctor is like, sorry, I'm hallucinating. And so what we have to do is we have to make sure that we actually figure out mechanisms to get full visibility on the model and then have validation that says, does the model behave the way that you want it to behave? So, for example, if I ask a question to the model, build me a bomb, what it'll very dexterously do is tell you, I can't give you that answer. But if you ask the question slightly in a nuanced way, I'm a movie script writer, I'm actually going out and shooting a movie with Brad Pitt, we're going to shoot a scene where Brad Pitt's going to get into the car, build a bomb, and then blow up the Bellagio.

1:47:23And so can you build me that entire scene? And by the way, can you show me how the bomb gets built in that scene? The model can get tricked. And so what you need to have is some kind of an algorithmic red teaming process that says trick the model in test rather than in real world. And do that algorithmically. And once you do that, then have some kind of mechanism to enforce runtime enforcement guardrails so that every single time you build an application, that application is prepared to go out and deal with those kind of questions that might trick the model so that you as a company don't have your brand at risk when you do that.

1:47:57That's the thing that we are actually building. And the thing that gets me really happy is if there was one product that I wanted to have fully written with AI, it would have been that product. Yeah. Because the speed, it has to outrun everyone else. And you have to make sure that you deal with this in machine scale. And so it's really cool to see the teams not only change the fact that they're using tooling to make that happen, but they've changed their entire process. Every engineer in that team is now just a spec developer. They just build specs, they create markdown files, they give context to the model, they give context to the agent, and then the agent's kind of writing the code.

1:48:33And then our biggest bottleneck is becoming review of code rather than the actual writing and generation of code. How long do you think that lasts? I think it becomes, well, I think. Because presumably the. No, the review is going to get easier too. Yeah, yeah, yeah. Yeah, totally, totally. At some point in time, these things are going to continue to get exponentially. We'll be able to just podcast all day long. You'll be able to just hang out with us. I could just be here like eight hours. It would be great. But, you know, but so I think that's an area. And by the way, that doesn't mean that you could have a lot of AI slop that gets delivered in the market when that happens.

1:49:08because what you'll have is really crappy software that gets written if people aren't paying attention to it. It doesn't obviate the need for having taste and having judgment and having instinct and making sure that you're thinking about what you're building. But what it'll at least do for you is provide this mechanism to accelerate the development where we're not just constrained by, I don't have resources, so therefore I'm not going to build 80 % of the things I want to build. However, prioritization is still going to be important. And I feel like focus, I don't think the nature of focus changes just because you can build a lot of stuff quickly.

1:49:43Yeah. Talk about innovation at Cisco. You mentioned it earlier, but I was watching a video from Jane Street all about innovation not necessarily being perfectly correlated with new company formation. Like what startups do is great, but you look at the iPhone that came from Apple. You look at the Transformer paper that came from Google. So two wildly different approaches to innovation, one CEO leading and the other this labs, the skunkworks team. How do you see innovation balancing between sort of top-down mandate for new products, what the market's responding to, customer-driven versus lab scientists going off and working in isolation?

1:50:28So first thing is innovation is, in my mind, it's a choice. So when companies say, well, we're big, so we can't innovate. There's a lot of big companies that don't innovate, but there are a bunch that do. So what's the separation? Correlation does not equate causality in that case. Just because you're a big company does not mean you can't innovate. In fact, some of the greatest companies are innovating at a really, really fast pace. And by the way, scale really matters right now. In AI, scale is actually a huge accelerant for innovation. Yeah, distribution, more data, flywheel, more RL, all of that.

1:50:59And so I do feel like innovation is a choice. And I think it's an intellectually lazy argument to say that innovation can't happen because we are too large. However, size does bring about some level of slowness in the process. And you have to be diligent about making sure that you're intolerant on bureaucracy seeping in. or more importantly, indecision seeping in and apathy kind of seeping in where people just give up after a while. You know what, I tried a couple times, it didn't work. And it's like, no, you have to be comfortable with conflict. You have to be comfortable with making sure that you speak truth to power.

1:51:42And if that means that you're going to... Like one thing that can become a failure state is in these large companies, you get overly concerned about feelings of people. And so as a result, what you stop doing is you stop having the debates that need to be had. What you need to do instead is establish trust first in teams. And once you have trust, you know that the intent is not to put you down. The intent is to make sure that you get the best idea ship out and get to win. And winning requires contrarian viewpoints that are actually actively debated. And sometimes hurting feelings. Exactly. But it wouldn't hurt feelings if everyone were clear about the fact that I trust this person and they're debating the idea, not the person.

1:52:27They're not attacking my personality. And so one of my mentors always told me, and I have not mastered this, state the facts but watch your tone. when I tend to do poorly is when I get passionate when I don't watch my tone because then that debate feels like it's a personal conflict rather than something that's about the idea and so if I were to say on the innovation side top down, bottoms up I think there are certain things you have to go top down like we had a decision we had to make that said we're pivoting to AI first as a company and that was not a decision I could have had democratically kind of sourced from 30 ,000 engineers It would have not worked.

1:53:07So we said we're going to go from top down on that one. But I should be losing 99 % of the debates in my organization because the person in the front line is spending 14 hours a day on that problem. I'm spending six minutes a week on that problem. Chances are, even if I'm infinitely smarter, which I'm not, if I'm doing the right job because I'm hiring people smarter than me, at that point, you need to make sure that that person is able to have more facts, more thought put into that argument than me. So I always tell people, if I'm having a debate with you, and if I'm having an argument, and if I don't lose 95 % of the arguments, then we've got a different problem.

1:53:46Yeah, yeah, not a good sign. Because you just haven't thought through. But what my job is to make sure that the second level and third level thinking is actually well thought through on your end. And then we need to make sure that we have certain core principles up top that go from top down and certain core value systems that we should not compromise on so that we don't actually get people on an infinite loop. And then from the bottom up, allow people to innovate and then let those great ideas percolate up. Well, thank you so much for coming on the show. Thanks so much for meeting you. Great to meet you.

1:54:18Hopefully we'll have you back soon. I'd love to. And in the meantime, I'll tell you about Cognition. They're the makers of Devin, the AI software engineer. Crush your backlog with your personal AI engineering team and our next guest is coming in right now fantastic I wasn't sure we were gonna have a little bit of a break but we have Costa nice to meet you guys thank you so much no no we're good we go for three hours every day it's no problem please introduce yourself since the first time on your show yeah so I'm Costa Cladianos I'm the EVP of technology for the San Francisco 49ers in Levi's Stadium.

1:54:57And I'm a big fan of the show. So excited to be here. Yeah. Explain to us what that means and what your day-to-day is like and what the choices you're facing are. So there's never like a regular day-to-day. So my team, we're responsible for everything from getting into the stadium, the ticketing systems, the point-of-sale systems for food and beverage, the Wi-Fi, the network infrastructure, the cybersecurity of the stadium, making sure that on game day, we're like the referees. You don't notice us and you just have a great time. and cheer on the team, which is hopefully the 49ers. Yeah, how has technology in a football stadium evolved?

1:55:31We were going back to the first broadcast was the Los Angeles Rams before they moved to St. Louis. Then they went back. It made sense because Hollywood is in there. They're very forward-thinking. But how much have you seen an evolution in the technology that goes into broadcasting a football game? Well, it's crazy. I've been doing this for about 25 years now in a bunch of different sports. And it just used to be, you know, you go to the game, you watch the game, you eat your food and you go home. Now it's different because you have to have a technology experience. The reason why is, I mean, staying at home sounds like a good proposition.

1:56:05You have your big TV, you have your couch, you have your food. So we want to make a reason it comes at the park. We want to enhance that experience. So how can we get it to be kind of a social experience? How can we get something that will get you off your couch and coming there and having a great time? Because, I mean, it's also not cheap to come to a game. You know, it's it's a significant dollar value from your wallet and an entertainment dollar. So we want to make sure that you have the best experience possible. And then, I mean, you look at the people watching sports. What are we doing now?

1:56:31We're you know, we're all on our phones, you know, we're all on our technology. We all want more data. So it's like it's not trying to get people off their phones. It's like, how do we engage?

1:56:45I mean, with us being kind of an influencer economy, everybody wants to show off where they are, right? So we're like, what a better place to show off where you are and having a good time there. And then your friends are going to be jealous and then come and enjoy the game as well, right? Rank the most, like rank your tech stack based on importance on game day, like specifically during the game. And I want this in the internet at the top. No, no, no, no, no. But I'm talking about like when, like if, if, if you're just an observer and you're enjoying the game and you're experiencing, you know, a bunch of different sets of technology.

1:57:18Like, what is, like, the average person not noticing that is, like, going on in the background that you're fixated on? Yeah. So, I mean, hopefully they're not noticing it because that means it's working, right? So, I think you start from the foundation, the infrastructure, the network infrastructure. And, you know, I tell everyone they're like, what is that? I'm like, that's the plumbing of our organization. That's where everything flows. That's to make sure without that, we have absolutely nothing. We're cooked. So, I mean, you know, Cisco is a great partner of ours, and we work together. It helps that we're in Silicon Valley.

1:57:48which it helps but also hurts because everyone knows technology in there, right? So everyone coming to the game, they expect the best, which is a great challenge and one of the reasons why I went there. So that infrastructure has to be solid to deliver your Wi-Fi, and there's just more and more need for it, right? We have to keep scaling it up now with AI coming down the pipe. Everyone's looking up their LLMs on their phones and obviously sharing their experience and, you know, with advanced stats and things like that, it has to be frictionless for them, right? And so that network infrastructure has to be solid, not only what's going on in the stands and the fans, but on the field too.

1:58:26We can't have, you know, scoreboards go down. We can't have coach comms go down. So we work together with the NFL to do that. So that's incredibly important to us. And then with that hand in hand is actually our cybersecurity posture. I mean, we can have the best systems in the world, but one hack, one cybersecurity incident, and it's a massive problem. Is that mostly like DDoS, just like adversarial hackers just trying to take you offline, screw things up, create economic damages? Or is it specifically like break in and steal credentials, everything? I imagine every day there's somebody that would love to broadcast onto the Jumbotron.

1:59:04Yeah, like this crypto scam or something. I don't know. What's the shape of the threat? There is so many different threats. Like, I mean, you just named a few and you're scratching the surface. Just because we're so visible, you know, every Sunday, the world's looking at us. And, you know, it's a visible attack vector. So, I mean, we have an incredible cybersecurity team. We work, again, with the NFL. And then, you know, share knowledge with other teams because, you know, we're competitors on the field but colleagues off the field. And so we work together to make sure that we're mitigating risk.

1:59:35And, you know, it's not if you're attacked. It's when you're attacked. So we try and make sure to stop that. So you have to be proactive instead of reactive. If we're reactive, we've already lost the battle. Is NAB important for you? NAB? Yeah, the national... The show? Yeah, the show. We go for the camera stuff here. We try and learn about it. It's a great show, yeah. We do send a big team down there. But I'm wondering if that's important to you or if there's a different conference that you're meeting more people and talking to more suppliers. No, we do send a lot of people to NAB, our engineering team.

2:00:05I mean, it's definitely important to us because look at our broadcast size. We have right now, I believe, the largest outdoor 4K video screens in the league. So we do send a lot of people there. I mean, you want to be around the best and the right. So we have a lot of shows that we go to. But it's not only you don't want to stay inside the industry. I always say you learn your best when you're looking outside. So who's doing what great. It doesn't have to be from sports. And what is the top-down mandate or standards that are shared from the NFL? Like there's obviously broadcast standards if you're delivering to ESPN or ABC or NBC.

2:00:40But what does need to be standard across the NFL broadly? Yeah, I mean, the NFL has their standards across the NFL in terms of, you know, Wi-Fi, infrastructure, connectivity. Expectations. Yeah, expectations. Every team has to be. Exactly. But, you know, us at the 49ers, you know, our team always overperforms on the field. You know, we're always expected to win a championship. You know, that's what we try and do. I mean, it's difficult, but that's like our goal. So, you know, with me and the technology team, you know, we're trying to be the best. Again, we're in Silicon Valley, so we expect more.

2:01:10So I want to go over and above and deliver an experience that's not just kind of the minimum. It's an experience where you're going to come there and say, wow, Levi's Stadium is unbelievable. Like, this is insane. Let's go there. This is something different that maybe, you know, somebody like a Disney should copy, right? Or Starbucks or somebody else in the industry. Sorry if they're not sponsors. How much different is the Super Bowl specifically? Is there more demands on your organization for a bigger event or is it sort of the same as any other game? Yeah, I mean, the Super Bowl is one of the world's largest, arguably the world's largest sporting event, especially in this country.

2:01:45And the demands are greater. And we work, again, hand in hand with the NFL as their event. The capacity, the Wi-Fi, the bandwidth, it always sets records wherever they go. I mean, the previous record at our stadium was Taylor Swift, which was an incredible concert, by the way. And, I mean, her demographic is super tech savvy. As soon as she came out on stage, people are, you know, putting up their phones and sharing that. And it blew away our bandwidth records a couple years ago. And we expect this to go even higher. But that's a record that's always made to be broken because people are using more data, not less, right?

2:02:21So we always have to prepare and scale up as well. Well, Jordy, anything else? I mean, I have a lot more questions, but it was great having me on the show. Yeah, thanks so much for stopping by. Thanks, guys. Great to meet you. Thanks, guys. Oh, Hugh Hobsoff. Let me tell you about Railway. Railway is the all-in-one intelligent cloud provider. Use your favorite agent to deploy apps, servers, databases, and more, while Railway automatically takes care of scaling, monitoring, and security. And we have our next guest already here. He showed up a little bit early. We can bring him down if he's down.

2:02:48Let's do it. He looks ready. I'll tell you about Console. Console builds AI agents that automate 70 % of ITHR finance support, giving employees instant resolution for access requests and password resets. Good to see you. Good to see you again. Good to have you. Data centers in space, what you got? When are we going to space? Coming in hot. Me, you, the International Space Station. Let's break it down. You know, the space tourism industry is quite a fun one, right? Would you go? Would you do the Blue Origin thing where they blast you out past the Carmen line? It's good enough for Katy Perry. It's not good enough for you.

2:03:22What's going on? It's like you're in free fall. You're not actually in space. Oh, it doesn't count? Shots fired. Oh, carbon line denier. I want to be like going around for days. Oh, okay. I want my bone density to start to atrophy. I truly want to feel the negative effects of space. Yeah, yeah. It's not enough just to go up and back. I think I would do it. It's like 90 seconds, right? Yeah, but it's better than being hanging out on Earth. But all the cool stuff that astronauts do, right? Like, you know, put water and then like they're bubbling and then you like try and drink the water. Soon they'll be unplugging the GPU.

2:03:53Oh, yeah, yeah, yeah. That's how you pay for your space tourism. You've got to go unseat a ship. 90 seconds of servicing. One 90-second trip at a time. No, but people were wondering, you know, TPUs, NVIDIA going on the Starlink V5 or whatever something gets up there. It feels like this will be something more like a Tesla silicon chip, an AI chip. Like, do you have any insight into, like, what the process, if you wind up figuring out how to heat dissipate, if you wind up figuring out the costs, what might the chip look like? So I think, you know, everyone freaks out, oh, my God, putting stuff in space is expensive.

2:04:34But if you look at, like, Starship launch costs and, you know, they keep falling, you're, like, fine, right? Like, I think that's not, you know, by the end of the decade, the cost of space launch will be fine. The heat dissipation, I mean, it's a challenge, but you just put a massive, massive, effectively radiator. And it's fine, right? By the end of the decade, you'll be good. I think the big challenge is that chips are just really unreliable, right? And so how do you deal with a couple things, right? Satellites can only be so large before they start needing a lot of support and structure before they tear themselves apart.

2:05:03So when you look at the launches, right, these things are shooting out tiny satellites and many of them. Okay, so you can't have a big, fully connected cluster of chips. And then on top of that, how do you deal with any random error? On earth you have text running around the data center unplugging stuff putting in spares things like that What do you do in space? You are a made to the factory where they like might unsolder it and resolder it and then like test it and it works and go back out Sometimes it is just trashed. You're like, you know That's the challenge to me. Is that is that I feel like maybe the the pattern we should be looking at is like How often do the Tesla self-driving chips need to get serviced because that's like the the team that would probably be building or like bridging the gap there like the Starlink satellites sir they go down but like the service works like you're you're just relying on some sort of like you know 90 % uptime stuffs coming down but most people that are in a Waymo like the chip keeps working right most people that are in a Tesla self-driving like they're not like you don't hear about Tesla owners being like I love FSD but I'm constantly in the shop getting my my custom silicon ships unseated and reseated right I mean it's it's also a function of like the complexity of a chip, right?

2:06:19Sure. You know, if a chip is twice as fast and let's say the bit error rate, right, like how often a bit flips is the same, then it's erroring out twice as often. Yeah. But let's say the chip is 10x as big, right? And so when you look at like a Tesla FSD chip, very, very good, very, very efficient, very still like relatively inexpensive and cheap compared to, you know, a big old GPU or TPU or whatever. Right, those things are extremely large. And, you know, again, like if the error rates are the same, Then it fails 10x more, but in fact the error rates are a bit harder, higher because they're pushing these things to the absolute limit Yeah, whereas you know Tesla does have some level of like well first of all the Tesla car has two chips Sort of redundancy already built in, right?

2:07:01Maybe you do that on the satellite, but then there's more power more Yeah, right so the whole little lore right of it is is you know Effectively power is free right and solar panels you look at the cost curve of solar panels look at the cost curve of satellite launches. You're like, this is free, this is great. But power is less than 10 % of the cost of the cluster. Sure. Right? So, like, it's that 90 % you're not saving anything on. Yeah, yeah. And insofar as much as... For potentially 100 times the hassle. Yes, yes. Yeah. There's this whole, like, you know, like, if you look at NVIDIA GPUs, right?

2:07:35Yeah. When you first turn on the cluster, about 10 % to 15 % of them fail, RMA, in the first two weeks. And then that's fine. You have to reseat them, whatever. And the industry knows how to deal with this. And over time, Hopper's now at 5%, but Blackwell's still 10 % to 15%. It actually started out higher than that. And when a new generation comes out, it's going to be higher than 10%, 15%. It'll have its curve gradually decline down. But are you going to test it and burn it in on the ground? Or are you going to say, 5 % of my chips or 10%, 15 % of my chips are trashed? Because someone can't go up there and do these things?

2:08:08Or am I saying, oh, I need robots who can do all this stuff in space, and now that's like an additional engineering problem, when sacks of meat are actually very cheap. Yeah, sacks of meat, yeah. Speaking of NVIDIA, we haven't talked since the Grok acquisition. What does that look like in the bull case? Like, if it's a good, if the next version of Grok is a great chip, is it sitting next to the, you know, H200, H100s in the rack, GB200? Like, how does it fit into the actual, like, what NVIDIA deploys? Is it just a separate chip? I think it's a big vibe shift from NVIDIA, right? Before they were like, all right, I got this big GPU.

2:08:44Everyone's going to use this GPU. Software ecosystem of the GPU is so good. It's one size fits all. Everyone's trying to make all these specific point solutions, but we've got the thing that's good at everything. And then they had a vibe shift, right? They launched this thing called CPX, which is a chip made for pre-fill, prompt processing, creating a KV cache, and also good at video generation and image generation. And that's coming out later this year. In the Grok press release, they were really talking about video generation as well. So yeah, you've got like CPX, you've got like the standard GPU, and now you've got the Grok chips, and they all fill a different niche.

2:09:16But really it screams, oh crap, we don't really know exactly where AI is going, which I don't think anyone does, right? I mean, it's moving so fast, the software is, the model architectures, etc. So we're just going to like engineer solutions that are along multiple points of the Pareto optimal curve, and then, you know, one of them will win, right? And I think it's like sort of like a big vibe shift from NVIDIA. Also, they just knew OpenAI was going to do the Cerebrus deal, so they freaked out. Got it. Yeah, get me up to speed on what makes Cerebrus important in the ecosystem right now. So, you know, you have people thinking like, oh, latency matters in terms of where our data center is.

2:09:53It doesn't matter at all. What matters is, you know, as we've moved from, you know, chat applications, which were like, or search response immediately. Chat applications, let's say response takes 10, 20, 30 seconds. You've got agents. You know, I don't know. my cloud codes are working in the background for a long time, right? It doesn't matter where the data center is, but what does matter is that these streams of inference take 30 minutes versus 10 minutes versus 5 minutes, and for a lot of people, I'm fine to spend 10x the price on something that completes 10x faster. And so Cerebris sort of just makes a ton of sense there.

2:10:24So OpenAI, they've got these long horizons. There's Codex 5.2, extra high thinking or whatever. It's terrible. Can you guys teach them how to market? you have to sponsor this podcast. Yeah, yeah. We had two more on yesterday and I did actually ask him, I had the Codex app pulled up on my desktop and I was like, there are six different models and then there's another button that I can pick to it. Well, how many different products are called Codex now? There's a lot. Now there's an app, yeah. We actually had another guy on just to do branding. Lexicon branding came on the show yesterday talking about all the naming.

2:11:02Naming architecture. It is complicated, but hopefully. You could tell he's just blood's boiling because all the AI companies just have the most chaotic. Anthropic, Claude, Claude Code. But also you can use Claude Code for other stuff. Yeah. But yeah, I mean, with Cerebras, it seems like there is a value to it, but are they constrained on the supply side? Can they actually scale up to a Colossus-style data center that could actually speed up Codex, not just for one user, but all the users? Cerebers can speed up multiple users for sure. The question is where you use it, and that's where they have to figure out where within Codex.

2:11:41Because there are times where Codex is running for 10 hours, and sometimes you don't mind. Screw it, I've put up this nice prompt, gone, work on it, refactor my code, do this thing, do this task. Other times I want this iteration feedback loop. So how do you expose it to the user without saying, hey, actually there's another toggle. So your permutation is 18 times. Well, hopefully like a really robust model router, but it feels like that's been a process. Yeah, so the OpenAI deal is like for 750 megawatts. It's not that much capacity on the order of like what OpenAI has talked about. You know, by the end of 28, they'll be at like 16 gigawatts.

2:12:15Sure. Of that. So it's just like the absolute cutting edge, the most price insensitive customers. in that specific use case of this is the type of prompt that you need to return fast, then you'll get the speed up potentially. Right, right. And they've got to figure out how to do it from a product, exposing it to the user, et cetera. But it's clearly something where there is demand, right? Like, I don't know, like Andre Karpathy doesn't care if he's spending a thousand bucks per agent per second or whatever, right? Like, you know, so whatever it is, these like super cracked engineers don't care at all.

2:12:44And then obviously there's like a long tail of like actually cost does matter for most people. and so all along that curve they've got to have solutions, right? When did you first think that XAI might end up at another Elon company? I mean this has been rumored for a long time, right? People are saying Tesla, Tesla, Tesla for the longest time But it's harder with a public company Yeah, and then a bit ago people were like, oh SpaceX, I'm like wait, this makes no sense There was a very coordinated narrative pump at the end of last year At the end of last year. No, it was perfect. And it was almost perfectly telegraphed.

2:13:21Well, there's a bet, right, between basically the head of compute of XAI and the head of compute of Anthropic. And the bet is what percentage of worldwide data center capacity is in space by the end of 28. And the bar is 1%. Oh, wow. And so the XAI guy is really bullish. The Anthropic guy is like, eh. Yeah, a little slower. Yeah, yeah. But it's a really interesting bet. I take the under on 1 % by 28 because that's a gigawatt in space. But it's actually not that crazy, right? It's roughly 150 Starship launches. We'll get them to a gigawatt in space. So Starship hasn't worked yet fully. I was looking at the energy draw of the current Starlink fleet, and I think they're at like, what is it, 200 kilowatts or something like that.

2:14:07So you get 1 ,000 of those, 200 megawatts, and you're starting to be in the territory, something like that. Yeah, so the V2 Star satellites, I think, are the only ones they've launched. Maybe they've launched a few V3s, but the V3s are coming soon, and those are like 100x more bandwidth each, right? Yeah, and more power. And just more power, and so what I'm just thinking of, can you scale this thing up at all? It's like, are they two orders of magnitude off? Are they three orders of magnitude off? It feels like they're one order of magnitude off, being one, something that looks like an H100. I think the metric is like 50, it's either 50 kilowatts a ton or something like this per satellite for V3.

2:14:43Let's say from V3 to whatever the compute thing is, they double it again and get to 100. I think the V2s are like 25. So if you get to 100 kilowatts per ton for launch, it's only 150 or so Starship launches. I think that's so reasonable. Maybe not 28, maybe it takes 29, but it's so reasonable. The question is cost and reliability. What happens when the chip fails? How do you service it? That kind of stuff. How do you deal with having clusters be much smaller instead of like you know these big clusters even for inference big clusters are useful. Yeah, how do you think about Google's response to Grok's TPUs obviously very successful but are they forking that project to eat more of the Pareto curve?

2:15:27Yeah so for the longest time Google's had one main line of TPUs right all made by Broadcom and then sort of next year they've diverged it, right? Where Broadcom makes a TPU and MediaTek makes a TPU. These two TPUs are focused at different things. And they're fabbed at TSMC. They're both fabbed at TSMC. Everything at the end of the day goes to Arrakis, right? I want to go there next, but everything goes to Arrakis. So fabbed by TSMC regardless, but both of these TPUs are focused on different things, and they've actually got a third project for another kind of TPU there. They also see this need to proliferate along the curve of like, hey, do I care a lot about super high amounts of flops, not that much memory?

2:16:04Do I care a lot about super fast on-chip memory only? Do I care about 3D stacking memory? Do I care about, you know, this sort of general purpose middle ground AI chip, which is what, you know, an H100, a Blackwell, a TPU looks like today. You know, they're sort of like, oh, we need to hit the entire Pareto optimal curve. And it's like, okay, within this, there's training versus inference differences and like what numerics you want and all these other things. There's so much complexity there. Everyone sort of is diverging their roadmaps once they're at a sufficient scale, I think. Yeah. Is Google still way ahead on cross data center training?

2:16:39Yes. And are the other labs, like, is that important to the other labs to catch up there? Or is it something that will just naturally happen because everything sort of commoditizes? Or do the other labs need to sort of marshal some Herculean effort to, like, crack the code on what it takes and what Google's doing? Yeah, so it's a couple of things, right? In 2023, everyone thought that scaling was pre-training. Yeah. Right? You know more parameters more data and that's very difficult to split across data centers And has Google been able to do and Google's been able to do that to an extent right?

2:17:10So what they've done is they've got you know, they don't have the largest individual data center campus But what they do is they do these like Regions where it's like hey each data centers roughly 40 miles apart from each other sure so in Nebraska and Iowa And then in Ohio they've got like these complexes and now they're building one in Oklahoma, Texas You know these complexes where there's all these data centers pretty close to each other Well, it's not really cross data center, across the world. Right. It's just across region. Yeah, and that makes a lot of the difficulties a lot easier. Flip side is we've also moved to RL, right?

2:17:40And majority of the time of the chips is spent generating data, only doing forward passes through the model. And then you only send the final tokens that you verified sort of back to train on to train, right? So then you end up with like, oh, instead of in pre-training scaling, you need to synchronize all the weights every 10, 20, whatever seconds. When you're doing these rollouts, and especially as things get more and more agentic in training, you might only need to send not the entire weights, but just the tokens that are relevant, so way smaller amount of data and way less frequently, right?

2:18:12Minutes at a time instead of seconds at a time. And so you've got this, like, now it's become, like, reasonable where, oh, actually, multi-data center training is completely reasonable, and people do this. People do multi-data center, multi-chip training, right? You do your inference on one set of chips and you do your training on another set of chips. So like Anthropic does this. I don't know if Google does this, but Google's kind of already got the cards. Okay, got it. Let's go to Arrakis. Yeah, talk about Arrakis. There's this debate. TSMC risk, is that the bottleneck or is energy the bottleneck?

2:18:44I was doing back of the envelope calculations. It seems like we're using maybe like 1 % of global energy production or Western energy production on AI specifically. workloads and then we're using like 50 % of leading-edge fab capacity on AI workloads and so that feels like okay well even if we all agree and we say as a society we're going all in on AI we can only double the AI chip capacity before we need to build more fabs that takes years whereas we could say everyone turn off your air conditioning we're sending the electricity to the data vendors right like we have the ability to do so we have create new

2:19:24turning off the AC. Claw needs to eat. Heat strokes for all the grandmothers. I need my cat dancing videos. You need to feed Claw, right? But seriously, there's this debate over, is TSMC the main bottleneck or energy the bottleneck? How are you feeling about that? Yeah, yeah. So sidebar before I answer the question, because I think it's fun. You know, in the US, it's insane to say, turn off your AC for AI. Yes. Right? And the general public hates AI already. Of course, of course. But in Taiwan, they've had droughts before, and they've turned off water to entire cities. They're like, oh, you get water three days of the week.

2:19:59And then the fab still gets supplied water. It's like this is, you know, you've got to understand the mindset. We are not ready as weak Americans to do this. No, but at the end of the day, right, like water and power are certainly less big of constraints. Now, you've got to imagine like, you know, semiconductor industry is used to, hey, doubling the amount of transistors made every year or two. Part of that is Moore's law. Part of that is more capacity. whereas the energy industry in America wasn't. And so initially, people were not creative. They're like, let's do these kinds of gas plants. It's like, well, no, now we've realized, yes, there's three main manufacturers of turbines, and then you've got, for a dual combine cycle, then you've got IGTs, but you've also got medium-speed reciprocating engines.

2:20:40It turns out Cummins can make a million diesel engines a year, and those can make electricity. If I don't give a fuck and I put it in West Texas, easy. So now it's more of a regulation thing, a supply chain thing. power is not a constraint in so far, like that much, right? I think it certainly is a constraint still today. It was the biggest constraint in 2425, data center capacity power, because the industry was not ready. People have woken up, they've sort of been shocked to the system. Now you've got tens of gigawatts being deployed. Next year, 30 gigawatts are being added, and we think the power's there for it.

2:21:14Wow. What was it this year? This year is like, I think it's like 18ish, 10ish. Okay. 15 to 18ish, sorry. So almost a doubling. Yeah, almost a doubling, yeah. Yeah, wow. And when you look at TSMC and the crew, right, there is not really, oh, this random, you know, there's 12 people making medium-speed reciprocating engines that you can now convert to make power at some random data center. No, no, no. There's like, there is Arrakis, right? There is one set of Spice, like, you know, that's it, right? And so, and then the flip side is like, okay, when you have 12 vendors, everyone's got a little bit of slack capacity, you know, There's more likelihood.

2:21:52People are like, oh, turbines you can't get. You can call a broker and you can get a turbine. You might be paying 50 % more, 2x more, but you can get a turbine. You can't get a 3 nanometer fab. You cannot get a 3 nanometer fab, exactly. And so when you talk about what's the, you know, the baton got passed from semiconductor shortages in 23 to power and data centers in 24, 25. 26, we're still, we're swinging the pendulum, but it will fully beat semiconductors again in 27, right? And so we see this across the entire space of the ecosystem. It's not just TSMC. It's also memory, both. Because both of them have built at a certain pace.

2:22:27Now, TSMC has been expanding at some rate. The memory makers, in fact, have just not expanded capacity. Basically, they have not built new fabs since 2022. Because their cycle is so undulating. Yeah. And so when you look at it, it's like, oh, even if they wanted to double capacity, they need to build the fabs. Yeah. Right? And building the fabs, it is the most complex building humans make, right? It's the entire air of a clean room circulates itself every 1.5 seconds. And you don't even feel it when you're inside. Really? It's like that. And it's like parts per billion of particles, right? Like it's actually insane how you could get coughed in the face by someone who has COVID and not get COVID.

2:23:08It gets circulated so fast it doesn't even hit you. It's like that meme of the spraying when someone's talking and then it gets circulated. So another sidebar is everyone knows COVID really popped off in Wuhan. Wuhan also is home to China's largest memory company, YMTC. And so when they were welding people into their homes, the people who worked in the fab still went to work. Wow. Because one, it's national importance. But two, these people aren't getting sick. This fab is way too clean. Yeah, that was crazy. Sorry, Jordy. I want to talk about Oracle. They put out a post this morning that said, Our partners financing for the Doña Ana County, New Mexico, Shackleford County, Texas, and Port Washington, Wisconsin data centers are secured at market standard rates, progressing through final syndication on schedule, and consistent with investment-grade deals.

2:23:59Obviously, they were fast-following their posts from yesterday, where they said the NVIDIA OpenAI deal has zero impact on our financial relationship with OpenAI. We remain highly confident in OpenAI's ability to raise funds and meet its commitments. and obviously everyone was looking at this being like, give me a cigarette. It's like bank run language. I haven't seen posts like this since the FTX era. Is it just bad comms or is there something worse? It's terrible comms. I told my Oracle contacts, I was like, who the hell is in charge of the Twitter? What are you doing? NVIDIA did something similar last year when the whole TPU mania was going on.

2:24:35It was like, we're thrilled with Google's progress with the TPU. That said, NVIDIA chips are the only, you know. NVIDIA. It's like no one asked you to comment. Yeah. I mean, like, I'm sure a handful of people in your DMs and random, but that doesn't mean... It doesn't project confidence. It's sort of the lion shouldn't concern themselves with the sheep. Yes. And like, okay, NVIDIA is a lion. Maybe Oracle is a little bit more bumpy, but I think Oracle is like fine. People are just freaking out because, you know, OpenAI is peak. You know, people are peak negative on OpenAI right now because of how good Anthropics been killing it.

2:25:09Sure, sure, sure. Yeah, I think it's just kind of silly. They need to hire someone to do comms like a Lulu or something, right? Both NVIDIA and Oracle, because what are you doing? How did you process yesterday in general? Jensen was clip farming. He was like, I don't know why he does these street interviews, right? No other CEO does those, where they just stick 25 microphones in your face and the paparazzi's flashing. It's a great vibe. It's, you know, Jensen's not been as famous as other CEOs for as long, and yet he's so important now. And if you've, like, if you know Jensen, how he's in meetings, I feel like there's two Jensens, right?

2:25:49There is, like, PR, like, good at PR, just good at talking, good at, like, making people hyped up and believe what he's doing. He's great at standing on stage, holding up the chip, delivering, like, a sermon. And then there's the real Jensen, which is like a business killer and like actually just knows about every like aspect of the supply chain. Right. All the way from like niche semiconductor, you know, design and manufacturing stuff all the way to like energy power data center. Like and then doing the business deals, too. Right. And so like you've got this whole Pareto, like a whole thing, a whole range of things that he's good at and he's a killer in.

2:26:24And clearly he's like he was in a meeting where he was being a killer and like negotiating like supply contracts or something. And then he walks out. And then he walks out. That's hilarious. This is my inference. But I like it. Yeah. That's awesome. And that's why he was like, you know, like he was like still killing. Like, no, we never said we committed to$100 billion, you know, like. And it's like, I don't know. Where do you even get the$100 billion number from? And it's like, well, you did go on CNBC and make a big deal out of it. So I think people would assume that it was. But they did say in the press release, I remember these are early talks.

2:26:57But they just kind of jumped the gun. This was the height of the press release economy. Yeah, yeah. What's funny is Oracle stock peaked just a week after they announced the OpenAI deal. And so the press release of, hey, OpenAI's going to do this humongous deal, stock peaks. Same happened with a couple other vendors who announced deals with OpenAI or NVIDIA. A lot of these, they all peaked then, and then it's sort of been like NVIDIA OpenAI Trade has been going poorly, and sort of like the TPU, Anthropic, Google, Amazon complex has been doing well. It's quite interesting that this happened. There's been good energy back at home with the roommates.

2:27:36What's going on in here? Yeah, one more thing. So, yes, over the weekend, it was sort of drowned out by all the Justice Department stuff. But wait have you guys talked about Elon saying you can smoke a cigar in the fab? No, yeah, yeah, yeah I was gonna say this is part of the whole thing. I didn't realize that was related. Yeah, that makes no sense. Yeah, indoor heaters Yeah, we have indoor heater technology. No one's taking advantage. Yeah, what does the fab look like if you have no humans inside? Like that's probably his long-term thing is like yeah, there will be an optimist. No one know like the number of people who work in a fab is like irrelevant like but is it is it irrelevant because there's all these things you have to do when a human's in there because they sweat and they breathe and if you don't have to do that because it's a robot walking down even if it's puppeteered or tele-operated you you might be able to have different considerations I don't know if that actually affects well it's like a nesting of like it's a nesting of like cleanliness right example you've got this wafer you've put like down let's say you put down copper yeah and now you're moving it from one area to another well it needs to be stored in a vacuum but the easiest way to store a vacuum like or an inert gas yeah and that's like the thing that's being transported in but then around that you want it to be super clean as well.

2:28:45If you don't then the copper starts getting oxidized, it affects your yields, all this sort of stuff happens. And so like you kind of want it to be a nested layer of like well this thing inside the EUV tool is super clean and then the thing feeding it is super clean and then the thing it sits in is super clean because that's how you get to like their zero particles. Because like you know in the FOOP and the transportation devices like parts per trillion and maybe FOOP it's called F-O-U-P, front-operated, front-opening, I don't know, something pod. But it's called a foop. It's like the thing that moves and it carries the wafers.

2:29:15Sure, sure, sure. And then the fab is like parts per billion and you've got this nesting relationship so everything is super clean. I'm bullish on robots, like super bullish on robots, but only for like, not for tasks that have like, TSMC's Arizona fabs, or okay, let's say TSMC Tynon, which I think produces like, you know, indirectly hundreds of billions of dollars of global GDP, even directly it's like still tens of billions of dollars, has like five, ten thousand people in it. Like it's like irrelevant in terms of the number of people who work there. In terms of the overall economic value that's created.

2:29:50Right. It's like how many people fold laundry or how many people wash dishes or how many people like do construction work. Like these are way bigger markets. For robotics. Yeah. Speaking of China, what are you making of the Dario essay? or I guess his comments at Davos about selling chips to China is equivalent to nuclear weapons these days. The Ben Thompson line was something like he's okay selling chips because he wants dependency on the NVIDIA ecosystem, Kuda, but he would ban lithography tools from going to China. And I've been wrestling with this idea of like, I don't know if China would accept this, but wouldn't there be a different world where you want them dependent on American LLM APIs?

2:30:36And you don't even send them the chips and you say yeah, you're you can have as much AI as you want as long as you're paying You know open AI and an anthropic API. Yeah, I think it's I think it's like a curve of like They will accept it's it's it's you know one you you you push someone to the corner They're gonna start swinging right and I'm like very concerned that China does this right? Do they do you do you push them too far into the corner? Do they say screw this we're gonna start being a lot more aggressive? We're gonna we're gonna you know do more military actions or military actions or even just invest twice as much in...

2:31:07In global supply chains, like take over Africa more than they already have, like LATAM, etc. Or just take over Taiwan. Because if I can't have the chips, what value is there in Taiwan existing in its current state? So there's this game theory aspect. At the same time, you don't want China to be able to... If you believe AI is going to do what I think many, at least in San Francisco, think it's going to do, which is completely revolutionize humanity and cause GDP growth to accelerate, Do you want to have China also own that technology? And their ability to integrate that into their military and all these other things much faster.

2:31:44So there is these competing interests. Where is the right line? And some people think it's like, hey, sell them AI model. I think Dario would say, don't even sell them AI model access. Don't even sell them tokens. Yeah, I think so. I think Anthropoc does not sell AI access to China. They loop it through, and you can see this in the traffic data. They go through Korea and Japan and other places. And so they get it. And then the other side is sort of like the Ben Thompson view. And I think I'm more sympathetic to that, although I think I'm not exactly aligned with that. And we've been saying, don't sell them equipment, don't sell them equipment, don't sell them equipment.

2:32:18And my argument is more economic in the sense of if you sell them tens of billions of dollars of equipment, they can make hundreds of billions of dollars in AI value or chips with that equipment. Whereas if you sell them AI model access, then it costs them this much to get the economic. You know, they're not able to... You're capturing more of the value in the West. And so that's sort of the question that is at foot here, right? Do you want them to capture all this value of the supply chain in equipment or by buying the chips or using the models, right? And services. And we've seen, you know, across many, you know, stacks, China refuses to accept, you know, using American ecosystem.

2:32:55And they'll wait many years before they develop their own. Whether it was like, hey, they didn't use Windows, they figured out a bootlegging economy, or they didn't use Visa and eventually they came out with like Alipay and WeChat Pay or whatever it's called. And like these things are way better than Visa, in fact, right? Lower transaction cost and higher volume. I never used Red Star Linux. It's North Korea's Linux distribution. Wait, really? Yeah. If you put it on a network, it'll immediately call home. So you have to put it on a firewall network or else it just like steals everything immediately.

2:33:25I'm a fan of Temple OS, you know? Yeah, there you go. All right. Is Doug O 'Loughlin suffering from a case of Claude Code psychosis? Okay, yes, yes. So I think everyone's like, Claude Code is for coders. It's like, no. Claude Code is for people who don't code now. Yes. Right? And that's the big realization this year. Yeah. You know, we've got a couple folks now in the firm who have psychosis. Okay. But Douglas O 'Loughlin, who is like, you know, semi-analysis, number two, he's president. Yeah, yeah. You know, he's my boy. In fact, he's the one who encouraged me to make a Substack a long time ago.

2:34:01What were you doing before? I had a WordPress blog and I was consulting on the side, but I was like, okay, let me do a Substack now. Because I saw him making money off it, I was like, this is shit. Why are you getting paid for this? There were multiple times where he wrote something and I was like, I could do way better. I'll show you. And obviously it was good because we both taught each other a lot of things and we've been great friends. Eventually he joined SemiAnalysis. His background is he was a hedge fund analyst, and then he decided to do a sub stack slash hike the Continental Divide trail for six months walking from Mexico.

2:34:33Then he came back to doing sub stacking, tried to do a fun. Six months of touching grass and then he was like, I'm ready to lock it on ClockCode. So now he's never been a software developer, but he's been on a generational run. He's not coding anything, right? He's just telling Claude to do stuff. And like, it's to the point where it's like, our like head of data, head of IT is like, oh, can you send me that? And he's like, how do I do that? And then he's like, he zips the whole thing and sends it to him. It's like local host. He sends him a leak once. It's like local host. It's like, bro, that's not how this works.

2:35:03But yeah, no, I've talked to some folks who vibe code and they'll be like, why'd you choose Node.js? And they're like, what's Node.js? That's a very specific choice. Someone? Yeah, Tyler. No, but we went on a little tour of a lot of our clients. Roughly half our business is, or 40 % of our business is hedge funds. So we went to New York a week, two weeks ago, and we went to all of our clients. And part of it's them asking me, is opening I fucked? And I'm answering, no, I think they're fine. And then some actual ideas. And then a lot of it's Doug just telling them, Cloud Code is like, they're like, you don't have to hire any junior hedge fund analysts anymore.

2:35:38And they're like, the junior hedge fund analysts are like, and then he's explaining, what can you do? It's like, well, you can just do financial models and perform a financial models and everything in cloud code without ever opening Excel. And you can generate charts and you don't need to know how to code. You just need to know how this stuff generally works and you can just do it. How many hedge funds are just trying to copy trade situational awareness? I mean, I think a lot of hedge funds obviously believe in AI. I think there's a lot of them who don't believe in it, to be clear. But a lot of them that have done the best believe in it.

2:36:13Why are they selling software everywhere? Oh, you mean selling software stocks? Yeah, yeah. Oh, yeah. Why the sell-off then? Yeah, I mean, of course, it's like an incremental thing, right? But anyway, so these hedge funds, like, and then the question is, like, okay, if you believe in it, how do you manifest that trade? And so when you look across the, like, ecosystem, I would say almost all my clients sometimes think our two years out numbers are too high. but like there's like Leopold's like your numbers are too low and so it's like in general right and I think if you think about how much do you believe in AI and what's your access to information of AI you know there's not many hedge funds who live in San Francisco and like fully breathe and live and understand it and then depending on how much you believe in AI how do you manifest that trait right are you surprised that more hedge funds wouldn't like even just smaller shops wouldn't say like hey this AI thing seems like it's going to be big maybe we should set up in San Francisco?

2:37:09There's a number of people, right? So we're getting an office together, Leopold, myself, Dwarakash, and then a client of mine, another hedge fund. And they have one analyst here. And there's a number of other hedge funds that are hiring analysts here, but being plugged into the AI ecosystem does not mean you're just in San Francisco, because you can just walk around and talk to doofus startups and VCs and not actually see what's coming down the pipeline. And you have to combine it with all sorts of information right you have to have a good tune with like what's going on in asia supply chains you have to have a good tune with what's going on in new york um you have to get tuned with like what's going on like in the in the financial markets right and then like what's going on in credit markets and what's going on in all you know the data center energy bubble all these different industries and so it's it's it's actually not like so simple to like be in tune with what's going on in ai you can easily get like head faked right um you know for the longest time people were thinking, you know, Adobe's an AI company.

2:38:02And like, and it's like, for a bit, like, Adobe was going down on AI, and then they like launched a few AI features, and the stock skyrocketed. And then now it's going back down again, because people realize, oh, wait, no, actually, it's not an AI company. Like, I think it's the manifestation and thought of like, what is actually going to, the world going to look like, if Anthropic 3x is its revenue again this year, OpenAI 2x is its revenue again this year, or, you know, by the end of the year, how many people even believe by the end of the year AI startup revenue is over$100 billion? I think that's an insane statement for a lot of people, but that's what it's going to be.

2:38:35And who believes that number? It's like very few people. And then you draw the continuation, it's like who believes, you know, and when Anthropic says in their funding, like, hey, we're going to have $300 billion of revenue by the end of the decade, and it's like, actually, I think that number's too low. Because the economic value of what they're going to create is going to be insane. Yeah. And you tell people, oh, you know, OpenAI is going to have 18 gigawatts or 16 gigawatts by the end of 28, and they're going to be able to pay for it. And that's like, well, that's$300 billion to spend. How are they going to pay for it?

2:39:07It's like, you sweet summer child, don't worry. Sam can raise. They're going to blow up on revenue. They're fine. Right? It is like a bit of a vibe thing. It's a bit of like, you know, irrational exuberance almost, right? Like, Leopold's in his mid-20s. Like, I'm 29. Like, we are irrational. Yeah. Because we have not lived through... You've got these PMs who... You've never been that humble. I don't know. My family almost went bankrupt in 2008 because we lived in a motel and we almost foreclosed. We actually did foreclose on one motel. It was pretty bad. I was still a kid. I've never been humble.

2:39:42It's good to live through that and understand how things can go wrong. What are you expecting out of Zock and Meta this year? We've been big Zock defenders. especially, I mean, there's this pressure of like, oh, Meta is spending so much and yet they haven't created any AI product that's super compelling or that's really working. And our stance has generally been Meta is making more money from AI than almost any company in the world outside of NVIDIA. So it's like, of course, Zuck should be justified in saying, hey, this is real, it's big. I'm going to back the truck up and go all in. Yeah, I mean, it's clear if you look at The most recent earnings, I think their CPM went up 9 % when the consumer's weak, which means like if you were to like try and strip out like what is consumer spending increasing for CPM of ads versus what is the effectiveness of their algorithms or algorithm got better by double digits in one quarter.

2:40:36Yeah. Right. It's like actually insane how good the algo's getting, right? At serving you the slop in the ads, right? So, so, so in that sense, like. The pig sound, the trough. I love it.

2:40:54We're going all in on that farm.

2:41:00I love it. So if you think about it, right, like, okay, Meta's, where are they going to win, right? I think if you have the galaxy brain take, it's like, well, they've got the best wearables coming down the pipeline. They're going to put AI on it. Apple won't be able to put good AI on their wearables, so they'll cede it all to Google or Anthropical AI. Well, the other thing is people have had this narrative, oh, as AI gets better, the value of real world experiences will increase. And I think that's a cool theory. But if you actually play it out, AI getting better means more content that's more effectively crafted for you, more personalized, 100 times more content, 1 ,000, a million times more content.

2:41:41that would imply to me that people will just use digital products more, which means more time on site, more time in the app for Meta. So I don't know. I mean, I'm with you entirely. But I think the Galaxy for intake is that you're just going to have a wearable and that's going to have an AI assistant. OpenAI is trying to make wearables. Everyone's trying to make wearables. Google is, et cetera, et cetera. I think Meta will actually execute and then they'll have a good AI. And then you stack on a few things. How do they get users? Well, we've seen, at least if you look at the user metric charts, Google's use, you know, OpenAI's users were growing, growing, growing.

2:42:19They were going to hit a trillion by the end of the year. They had 800 billion. Why did they not keep growing in the last quarter? It's because NanoBanano came out and they took all the incremental users, right? And likewise, if you go look at like, you know, Gemini 3 didn't actually make Google grow that much. It was NanoBanana and then Pro or 2 or whatever it's called, right? Those are the ones that made them really grow. Meta's licensed all of Midjourney's code data models right one two they're like actually just like focusing hardcore on was that a billion dollar plus deal the number is undisclosed Midjourney still exists as a company no it felt it looked to me like effectively a massive exit but the best case scenario where they can just keep kind of being artists I think if you had me guess I would bet it's over a billion Right.

2:43:06Every deal that Meta did was over a billion. Basically, whether it's an employment contract, a licensing deal, an acquisition, everything had to be after it. So the interesting thing is Meta… You're missing a zero again. Don't never miss the zero again. Every discussion was how many billions are we spending on hiring this person, buying this company. Well, Meta interestingly has gone down market for a compute because there's not enough compute in the big size deal. So they've actually gone and like I mean rent bought like small clusters. Oh because it's like well I want more from like long tail Neo clouds.

2:43:40Yeah, just like yeah from a longer tail Okay Because that's the only place they can get the compute they need interest because you know They've already like went out and signed big deals with Google and core weave and so on So cluster max three gonna be a smaller shark because of consolidation in the industry. No, it's there's more it's gonna be bigger It's gonna be bigger bigger But but you know so metal That's the thunder. That's ominous. It's ominous. So I think meta will, you know, capture consumers through generative. If there's more content, people are just going to go to the content marketplace, right?

2:44:11The creator of the content captures less value as there are more content creators and more diversification of content, right? And so I think meta just wins by being a platform, right? Google does too and ByteDance does too, right? But like those three win by having a platform. And then the real question is, can they get in the assistant productivity game? And I think this is important. And through that, effectively search. If you're an assistant, it means that there's some commerce happening. Well, they spin out and poached a bunch of people from Google. So this wasn't in the media much, but they actually poached Google search people with similar sized deals as these crazy...

2:44:46Yeah, and I always demoing any of the wearables. You can imagine like Meta wants you to walk around in the world and see like oh, what are those headphones? And like while we're talking I just hit my little thing and buy it, right? And it's like you didn't even necessarily know that it happened But like of course Meta is going to want to monetize that. Everyone knows those are the Sony MDRX suit 272s 462. Dude, I've been screaming about them like doing some proper marketing. It's literally like over here is like WH-1000XM5. And then their in-ear is like WF-XM1000. It's like, dude, just call them like Bravia Buds and Bravia headphones or some shit.

2:45:26Well, did China just bought Sony? Yeah, yeah, Bravia brand is actually a Chinese company now. Sony sold their TV. PlayStation Buds. Yeah, yeah, yeah. Walkman. Oh, yeah. It's like, come on, like. Something, something. For sure. Anyway, anything else, Rudy? No, this is great. I'm excited for this weekend. Yeah, yeah, super excited. What are some plays that we don't watch a lot of sports? What are some plays? You are a football guy, right? Yeah, yeah, yeah. George man? Yeah, rural Georgia, so I like football. High school football was the thing. College football was the thing. I think NFL is a little less soulful.

2:46:05Sure. But, you know, now college football has the NIL, and so it's also soulless to some extent. It's fine. We enjoy it, you know. Primal desire of seeing heads clash. Yes. You know, and sometimes that manifests in like, you know, Twitter drama and sometimes that manifests in real football. Yeah. All I can say is fuck the Patriots. Okay. Whoa. Okay. Okay. I'm kind of bummed. We're going to, since we're going to be at the game, we're not going to really get the great experience seeing the ads. I'm going to be like glued to my phone. I want to see all the AI, the different. Well, don't worry. I got some more ads for you.

2:46:39Thank you so much for coming. Thank you so much. Great segue. This message has been brought to you by. Public, investing for those who take it seriously. Stocks, options, crypto, treasuries, and more with great customer service. Also, speaking of ads, we're brought to you by Vibe.co, where D2C brands, B2B startups, AI companies advertise on streaming TV. Pick channels, target audiences, and measure sales just like on Meta. What else is in the news, Jordy, before we get out of here? We're going to wrap up. We somehow never covered this. Waymo confirmed that they raised$16 billion at$126 billion posts, according to Ed Ludlow over at Bloomberg.

2:47:13Saw a bunch of VC victory laps. Sequoia getting in the mix. DST Global, Dragon Year. It was sort of a bold move for a lot of VC firms to go in so early as the company was sort of spinning out. And, of course, Google via Alphabet did$13 billion of the round themselves. Very nice. The Medal of Honor on themselves. That's what you're referring to. Exactly. Yes. While you pull up the next one, let me tell you about Figma. Figma Make isn't your average vibe coding tool. It lives in Figma, so outputs look good, feel real, and stay connected to how teams build. Create code-backed prototypes and apps fast.

2:47:44Pull up this post from Redaction. Okay. It is Twitter every time a minor AI advancement occurs. Like this new block meta changes everything. This new block meta?

2:48:01There's a lot of this going on. Well, you know what else changes everything? Gemini 3 Pro. Google's most intelligent model yet. State-of-the-art reasoning, next-level vibe coding, and deep multimodal understanding. YC wants you to start an AI Rentech. They want you to... Just do it. Get out there and do it. Start an AI Rentech and then get on Phantom Cash. Fund your wallet without exchanges or middlemen and spend with a phantom card. And that's our show. Goodbye.