Alex Karp LIVE at Palantir AIPCon | David Glazer, Ben Harvatine, Danny Lutkus, Jonathan Webb, Nancy Cable, Ryan Asdourian, Drew Cukor, Zack Porter, Kyle Kirkwood, Matthew Jacoby

34:02Oh, no way beer company for a year That was a great sort of transition from your technology. What were you doing at a Anheuser-Busch? It was a it was a management training program location-based. So yeah After that, ran a hardware startup for a bit. Okay. Went to another hardware startup. But I had some buddies from college who had worked here. Cool. And the thing about Palantir seemed like everybody had just kind of like more autonomy and authority than I saw anywhere else. Yeah. Yeah. Amazing. So what do you want to show us today? Can you give us a little tour of what's going on? I've got a little.

34:34Brought a robot. Yeah, one robot. Bringing a robot is a great sign of respect in our culture. Yeah. Thank you. Well, you can imagine when we have events like this, there are a lot of demos. It's pretty screen heavy with software stuff. And we've seen a lot of, I'd say, increasing demand for our edge offerings and hardware offerings. We're really trying to push the technology further and further down to the shop floor and into the field. And so I wanted to put together something, just a little kind of toy demo that made that a little bit more tangible for people who are here. Yep. So walk me from my understanding to how we get to the edge, how we get to robotics.

35:14Because my famous, like the case study that comes to my mind for Palantir in terms of like making things in the physical world is like I think the Airbus example. So I, and whenever somebody says, oh, what does Palantir do? I'm like, okay, imagine a plane. There's a bunch of different parts. You gotta have a certain amount of seat belts. You gotta have a certain amount of engines. You gotta have a certain amount of fuel lines. You gotta have a certain amount of chairs. and all those come from different places and they all have different lead times and strengths and they need different safety requirements.

35:41Did they get checked off? And so you put all of that instead of just in a loose database, you put it in a database, but then you have Palantir that's actually tying everything together. So you know if there's a lead time on engines, you need to order more seat belts in three weeks instead of two weeks. And that's kind of how I explain Palantir in terms of like make a big thing that's complex. Is that roughly right? And then how do you walk from that to like, we need Palantir to somehow interface with a robotic arm. Yep, yeah, I mean that's roughly right. Like the way I think about it, it's like anywhere you go, people have data scattered all over the place.

36:10So the first step is can we get that all into one place? Got it. Then can we model that data so it's as easy to work with it as it is to talk about the concepts that it represents? Right, just like make it kind of - So there's this big meme in Silicon Valley and defense tech right now that like, there's a whole host of manufacturing guys, they're all aging out, they're 65, and everything that they know about how to make a widget, whether it's a chair or a rocket motor, it's in their head. It's going to retire with them. They haven't written it down. Maybe it's in some loose notebooks. And so this is kind of a way to jump and start getting more data online, right?

36:43We're actually not throwing out the data, we're capturing it. Correct. Yeah. And really, like the whole point of any of these data exercises is you just want to put the right data in front of the right person at the right time to make the right decision. Yep. And then just be able to close the loop and learn from it. And so if you're looking across the supply chain, that's how you do it. if you go down to a factory floor, the process is there, that's how you do it. And so when it comes to this robot, we're basically just like pushing that edge further. So instead of popping up an alert on a screen that tells somebody to go do something, what if you could actually just tell the robot to go do it?

37:12So again, sort of a simple like toy example here, but the basic idea is that this is a little work cell that we made with a robot arm and a camera on it. 3D printed, right? Yeah, it's all 3D printed. Wait, even the arms are 3D? Oh wow, okay, yeah, I didn't realize that. Cool. And so it's kind of set up to be a dumb terminal that kind of works and looks like the robot arms you'd see on a factory floor. You can give it moves to take, maybe you can ask it for a picture, but past that it's not doing any heavy computation on board. But then you can push that data to an edge hub that can run embedded models, can run embedded ontology.

37:48So you can actually take that kind of model of the world in terms of objects, relationships, actions and models, and you can push that down to the edge. And even if you have, say, like a network sparse environment where you don't have that real-time uplink to the cloud, you can continue to run off of that ontology. Yeah, we were looking at semi-analysis. They put the five levels of robotics. I forget exactly how many levels there were, but they were trying to map the self-driving car analogy to physical robotics. And I believe like level zero or level one, like the most basic was you have a pre-programmed robotic arm that's doing the exact same move.

38:24it's taking the windshield and putting it on the F-150. And it's this huge arm and you can't go near it because there's no cameras on it whatsoever. And if you step in that work cell, it will kill you if you're not careful. And this seems like a step towards level two. We're able to actually understand what different products mean. If there's, oh, this type of product shows up, there's gonna be more likely that there's a defect or you need to adjust what the robot is doing. How can you actually get that data into something that's actionable? Yeah, and even in this simple demo, we've got, it'll trigger alerts on, it tries to execute a move and you end up with a block jammed up up there.

39:00It'll realize that. It'll say, hey, you got a jammed hopper, you need to declare it, that sort of stuff. Okay, interesting. Where does this play in the stack of other software? I know when we talked to, what was it, Dirac, our buddy Phil, he was saying that he's working with automotive companies, but then they also have a lot of, there's a lot of like lower level control software on machine lines. Some of that's from German companies that I think we just talked about with Dr. Karp. But like where do you see Palantir playing in the stack? You have a bunch of data, the database, you put Palantir on top, but then at a certain point there might be some robotics company that makes the robot and then they also might have some control software with kind of a messy API or something like that.

39:44Yeah, I think we can be pre-agnostic about how far up or down the stack we go. So we've got this box. This is this is the node that goes on the edge, right? So this is an example of an edge node that one of our partners at scale makes. So this is that box that you can stick in the closet. Network to those existing machines that you have on the floor. If you just need a turnkey solution. And I think at the other end of the extreme, that's where we've got something like this, where this really at the end of the day is an ontology defined piece of hardware in that the machine itself, its entire configuration, the state machine is running, everything about it is defined in the ontology, lives in the ontology, and it's really just a bespoke piece of hardware running that ontology-native software.

40:29It's a monument. So if you've got more nascent operations, more greenfield operations, you think about some of the companies we work with in defense tech. It's like they can go all the way down the stack if they want to. For some of the larger, more established customers that we're working with, the plug and play solution might be a good starting point. What's the sweet spot for the specs on an edge scale, like edge node, like something on the edge? Do you need to be running like a large language model that feels like something that you could do on a 4090? I'd say it depends on the application.

41:00Like we've done some examples of that, even like previous AIP cons, where it's like, do we need the local app served up with a chat bot for the line operator who can just be like, what's going on? And it just talks to you. and it's not just purely deterministic. Okay, if the block is blocked, then send the error message. Instead, it's actually interpreting a bunch of data in kind of a non-deterministic way. So I'd say it's like, I think like anything, it really depends on the application and the users. Because again, there are a lot of guys that are working on these lines, guys and girls, where they don't need another screen in their life.

41:36And so it's really finding what's the right way to interface with those operators to ultimately just drive the better decision making. So how much is, like, how much is, what is the role of the FDE in this kind of new era, new territory? Because it feels like. Yeah. Are you graduated from being an FDE yet? Or is it once an FDE, always an FDE? Yeah. I think it's once an FDE, always an FDE. I try to keep my hands on keyboard as often as I can still, you know, still flying out to whoever axle factories in rural Kentucky or whatever. That's awesome. Yeah. I think the closer you can stay to that stuff, the better.

42:09I think really like the role of the FD is like just like it always has been go on site with customer Don't just understand but internalize their problems their challenges, you know, and so go create some value Yeah, well, thank you so much for hopping on the stream. We appreciate that really congratulations on everything. Thanks for bringing your baby Yeah, yeah, you can definitely take this out here I will grab this and we will have our next guest Danny Lucas from Palantir coming in. He also has a demo So do you guys know if the demo is going to need the HDMI cable? Is that right? Okay. So we will bring in Danny whenever you get a chance.

42:47Yeah. Let's bring in our next guest. Here he is. What's going on? Welcome to the show. All right. Great to have you. Do you need HDMI cable? Yes. Do a live demo? Always. That is bold. Doing a demo is on a live stream. This is live. Not for the week. So literally anything you share on your screen potentially will go out to the internet forever to be baked into the future super intelligence Yeah, baked into the training models in the future into the pre-training data So be very careful don't leak anything, but but but introduce yourself. Tell us what you're gonna show us

43:25What's going on guys, my name is Danny. Yeah, let's see here I'm an engineer at Palantir. I've been a palantir for about 12 years in terms of like my role to describe like i'm sure everyone at palantir said that uh i guess like if i had a role or a title i i do a lot of our business in the midwest at this point so first six years at palantir i was on the government side i did work with department of justice yeah u.s special operations cia national counterterrorism center sure after my wife and i had our first kid she was like hey could you not go to weird places in the world anymore and i was like totally reasonable reasonable request we we moved back to the Midwest and I switched over the commercial side and that's kind of like what I do now is like grow our business in the Midwest yeah what's like a what's a like just line drive solution that you like just total wheelhouse solution for you know I imagine like a large enterprise customer in the Midwest yeah what I focus on a lot is manufacturing sure in the Midwest so you can like there's huge manufacturers in the Midwest whether that's like Johnson Controls or Eaton or Molson Coors, Cummins Engine.

44:38So it's a widgets factory. They're making widgets, they're buying parts, they're assembling them and you have to understand the flow rate, where's the rate limiting factor, how can we increase flow. This is where I think we have the most differentiation from a product perspective because it's like I can actually affect the physical world and then I can measure how I affect it and then I can learn and improve how I affect the physical world the next time right whether that's like hey i'm in supply chain and i'm short on inventory like how do i solve that problem in the most effective and optimized way versus like i'm trying to manufacture something and like how do i make sure my machines are running i have the right labor i'm trying to do the right thing so like the the real magic behind all this too is like these yes they start off as like singular use cases that are like pretty great like straight shot but then like when you start to connect these workflows together and it's like oh the machine's down like and I have this material like what do I do and how do I go do it what do you want us to show us today I can kind of hold this for you if you want so we can get sound on this okay cool yeah walk us through it what I was gonna demo is I think like one of the interesting things and I'm sure you've like talked to a lot of different palantirians today is like we are never gonna purport to be like a strategy consulting type of thing when we engage with customers like we're never going to purport to be like oh like a hey we're experts in x y or z and the great thing about that right is like we're true to like who we are the bad thing about that right is like companies will identify and the organizations that we like that we work with will identify like hey i know this is a problem right but like there's a huge amount of time between like hey there's a problem and then let's go like implement a solution and the dependencies on actually getting to that faster are like I have the internal SMEs that can actually like understand the problem and come up with the right solution and do the feasibility and all that great stuff or I go work with like strategy consulting I pay millions and millions of dollars to get a deck that tells me like hey this is the solution that we think you should employ with the right like ROI in this approach and we've done this feasibility study and we think that you should go do that and so like we find that as a huge impediment to like our own growth right like why should i wait months yeah you don't want them to go spend millions of dollars with some random group to then recommend a palantir product that's 100 right and so like what we've been exploring more is just like well why can't i use ai to do that like why can't i like give a fairly haphazard business like a description of business problem and use agents essentially to like structure that into a better business problem description to do the necessary research about like what are the potential solutions of things that I could and should deploy to go solve this problem.

47:34Can I generate ideas with all the requisites of how I actually employ those ideas and actually generate a proposal where then I also have like agents as critiques on that proposal to be like, is this technologically feasible? Is this like financially feasible? All the things that you would expect, like strategy consultants to do for you, like, I should just be able to do that in a day and come up with a proposal. But then like, I don't know if you guys have talked to anyone about AI FTE. But then like, I should just then be able to use the output of like, this to then go build it. Like, I should just be able to say like, cool, here's the solution, I need to go build input into AI FTE, build it right and go from like you know what would have taken six or nine months until we ever get engaged to like well i think this is a problem like let's just go do it like in the next week right does that make sense yeah yeah it makes sense um i have some follow-up questions but maybe maybe jump into the demo first cool i think like like my immediate i guess question maybe it's relevant uh it's like how do you ensure kind of quality right because like you didn't say this but like someone else in another context might call this like vibe coding right sort of like generating like a deep research report on like a problem and a potential solution and then like you know sort of prompting your way to an implementation and uh today uh you know just like code quality and product quality ends up popping up but i'm sure that totally i think about that my take on this is like when you start doing anything with ai or large language models like it there has to be a human in the loop right not only to make sure that quality is coming out of the other side but also to ensure feedback loops are occurring and right and then and then you can take that context and start getting closer and closer to a jesus take the wheel moment where like um where like you actually have built trust because like part of this is not actually like I think a technology problem it's like a people in process problem where like people actually build trust in it and also you get all the tribal knowledge that's not in any system actually incorporated in some knowledge context that you can start to build off of over time but I think that's like that's the that's the trick is like humans always have to be in the loop right to begin but then like you build trust until you actually do the Jesus take the wheel moment.

50:06Yeah, so yeah, with this demo, what is the, uh, is the design as like an internal tool or something that you would actually sort of out? A lot of our customers are starting to use this to start to shorten the, the, the cycle time of going from like initial problem identification to implementation. So like, and is that for, is that for customers that are already using Palantir? Yeah. Um, so like we, we've started using this primarily with like a lot of existing customers, right? But then the cool thing about it is, I don't know if you guys have heard where like, All of the things I'm going to show you are kind of like native components of the platform But then we've developed this capability where we can say like hey, this is actually a really repeatable workflow What if we package this up and then just it's way easier to deploy where we can just like deploy there deploy there deploy anywhere basically?

50:53Cool. Yeah, so walk us through pull it up and maybe bring it a little bit closer. So oh, yes Oh, yeah, yeah, go ahead. Yeah let's do it no here we go saying text messages or anything like that all right cool um i used to work in the aviation space a lot and i fly in and out of newark um which like if you guys do that you know that's a real pain in the ass yeah so let's let's start there let's just say like um redesign hey i'm a um oh yeah for sure go ahead so like the problem the problem that i'll type in basically is like hey I'm an aviation expert like we're seeing significant delays around like Newark Airport because there's not enough runways and the runways are too short like what should I do to optimize my flow okay basically to solve this problem sure so like you know you guys get to see me type yeah it's always fun yeah this is interesting I yeah but a ton of questions I've always wanted to Redesign the LAX like streets.

52:02Yeah, the U. Like the flow of traffic. Yeah, that is a wild choice by LAX. Just constant, constant traffic. Wasn't too bad this morning, fortunately. But we did have a funny incident with a member of our team who first day. John arrived, got through security. Oh yeah. And almost managed to miss his flight because he was getting breakfast. By a former guest and friend of the show. I would call and text and said, you know, this is no time to take shots at the dyslexic. He had missed, he had made a mistake and confused gate nine for, for gate six. Right. And, and there is no gate six in this particular terminal.

52:44Anyways, we've headed to a different terminal. I have mine. Thank you for covering. Of course. Yeah. Yeah. So it doesn't have to be. So right now I just, I typed in, I like pretty rough problem statement. I'm an aviation expert. I want to solve problems around EWR airport. Yep. there are too few runways and the runways are too short how do I optimize traffic flow around it to minimize disruption okay so that's kind of like the first point and what's happening here is like the first set of agents is basically taking that as a problem description and actually like putting more structure around it so it's not like my you know my like misspelled problem like cleaning it up it's like a prompt engineer effectively on the left side of the screen you can actually see some of the logic of like what happened the train of thought here of like hey here's the problem statement i can see the system prompt like what the task prompt is what the lm like responded to when they saw this to them actually then creating and structuring this problem which is like hey the core objective is i want to optimize air traffic flow around newark liberty international airport to minimize disruptions to delays and efficiencies it puts out like key requirements yep like prioritize aviation safety standards It gives out restraints.

53:56Nathan Fielder would be happy to hear that. Yeah, yeah, yeah, right. It gives out constraints like limited number of existing runways, restrict simultaneous operations, et cetera, et cetera. So this looks pretty good to me as the initial problem description. Way better than the garbly cook two sentence thing that I did. So now I want to start to get into the phase of actually starting to do research on this to say like what are potential tools, what are potential approaches to actually solve this problem. And so what's happening right now is like now we're going into kicking off into more of like an agent.

54:31Yeah, just branching a bunch of agents to go do deep research. Exactly. So like now on this screen, I can see that same like core objective function over on the left. What it's working towards. And then I can start to see as it's running on the left, like research topics as it's doing research pop up and modeling. This is all built in like native foundry tooling. Sure. How inference heavy is this? Because it feels like it's going to town right now. I'll show you kind of like the under of how we're actually doing the research. Yeah. It is a unique, I don't know, like problem set because it's like going to town is something we worry about when we're talking about like, oh yeah, you have a billion consumers and $10 really adds up.

55:16Yeah. But if it's like. A problem as important as this. If you're talking about optimizing an airport, I think I can deal with a$100 inference bill. I'm gonna be okay with that. For sure. So the other thing that I think is interesting here is that I think agent is a very, there are a lot of definitions for what an agent is, I think at this point in time. One definition is like, and this was kind of our first approach, was like, hey, let's build a set of logic that an LLM actually orchestrates different parts of that logic between, and it can use tools like deterministic tools or it can write back or it can access and query things to ultimately do some type of automation.

55:59I think the other definition of like what an agent right now is like more of a chat interface. And then in that regard, right, like I want to be able to give that chat interface like access to tools, right? And so in this case, like what I've given the agent access to is a bunch of different tools. First, I can see the model that I'm using behind the screen here. And from our perspective, we think the models are mostly commoditized at this point. There might be certain models that are better at different things. And you actually probably want to use these things interchangeably and actually have an evaluation framework that, based on the tasks that you're asking it to do, will select the right model for that particular task.

56:41But in this case, I'm using Grok 4. And then for the tools in particular, I've given it access to conduct research. So I've given it some ways in which it can actually reach out and use different, either internal or proprietary information of the organization that we're working with, or reach out and use something like perplexity to do more AI-based search. I've given it the ability to generate, create code blocks. If it's coming up with an ROI and it needs to do napkin math, I want you to allow you to actually generate the code, but also then run the code to see what the result is. And then, I mean, it seems like all of this is kind of like frontier level, but available broadly, but the palantir value is that you actually have data that isn't just available on the web.

57:29And so if I'm actually an airport and I actually have specific data about... Well, the thing that stands out to me is if you're a large enterprise, you want to work with Foundry and have that ability to be model agnostic. And where does the leverage flow in that situation when Foundry can just sort of decide on the fly, what form of intelligence do I want to use for this problem set? Very cool. So I can see kind of like the train of thought on the right, like what it's doing. And so it's going to go, it's already using the research kind of tool. and you can already see the research topics starting to like pop up here.

58:08So like this is an example of an application, right? That like a user would use. They know nothing about Foundry, right? They're logging into an application. Their job is like, go do this thing, right? But then behind the scenes, you have a lot of different options for how you're setting up this logic. I don't know how much you guys have seen Foundry, but this is an example of what we call AIP logic. I could write all of this orchestration and code if I wanted to. I'm fairly lazy. so I use the lower code tool, which is AIP logic. And so here I can just like set up a bunch of different orchestrations for how I want to function to run.

58:40In this case, I'm putting in inputs for what I want the query to be, which is around like that problem statement we talked about. And I'm setting up functions for how it can like reach out to different types of sources. So like the first one is like if I had an internal kind of like proprietary information on schematics of a runway or planes or what types of runways planes can land on things like that like that's all information that then i can make available to the llm to go to a combination of like semantic and keyword search against it to find the right information to go do research against but then like as a backfall then i'm just like also giving it access to go and query perplexity right and go say like hey go find what's out what else is out on the internet to actually go do this research about this particular problem right and then bring that back and then the last part of this is like an action then to like go capture all that information and store it back into the ontology layer in foundry awesome so this is kind of like what it's doing live is like it's still working it's working like and it's and it's writing as we like as it's doing research right so it's like what is the current runway configuration operational capacities and key limitations at ewr including details on runway lengths numbers and and how they impact aircraft operations.

59:55And so then it actually gives me like, this is pretty good information, it'll cite the sources where it's coming from and everything like that, right? What are effective non-infrastructure strategies for optimizing airport throughput, right? And so in this case, right, it's actually saying like, hey, there's this performance-based navigation as a cornerstone, right? Yeah, I remember hearing that if you have the plane board from the back to the front, it'll load way faster, but no one wants to do that because the - It's a business model thing. Yeah, because people pay to be in front of the plane and they want to get on the plane first.

1:00:27But there was another proposal that was like load all the passengers that have window seats, then all the passengers that have middle seats and then all the passengers that have aisle seats. And they all kind of just flow in. No one's quite figured that out. But yeah, I mean, I could imagine that it could come up with a bunch of different proposals for, you know, similar just kind of like rethinking of the flow. I think we're getting short on time here. one question one one let me like yeah yeah i'll show you kind of like an end product here please which is like let's go i already ran this today i was like hanging out with the american airlines guys because like we're making fun of ewr as one as one does not their hub um but yeah this is like an idea that it generates and then like i get a summary of what that idea is and then it automatically develops critique agents yeah that are like looking and evaluating on different type of like uh different criteria right which is like hey can i what's the risk assessment and mitigation evaluation what's the economic feasibility of actually doing this like what is the safety and regulatory compliance evaluation and then it's going to run like those evaluations using that agent as a like a task criteria yeah to actually then say like i can see the guidance that we gave the agent right and it's task and then it has to go evaluate to see if it makes sense from that perspective yeah right and it even like generates its own models and its own code to say like hey is this feasible from like it can i do basically nap uh like napkin math yeah and say like can i come up with like how i could calculate this and actually go and like run and how how close is this output do you think to what a larger yeah strategy pretty i think it's like pretty aligned right because like they're not they in in normal times like these strategy consulting firms aren't getting access to all the data and so they're like being like okay come up with the idea do the research generate the idea guess a little bit then like i need to do some napkin math on like how i would think about actually like critiquing this idea and then ultimately like i need to come up with a proposal right and here's like the end proposal for what i think you should go do same framework where i have agents then writing portions of that proposal and then from there right it's just like copy paste that proposal in the ai fde and like start building right last uh last quick question are you feeling the re-industrialization yet are you seeing new entrants into the midwest building things or is it more legacy players just trying to trying to increase i think it's like a lot of what i work with are companies like eaton which are like hundred euro companies or like johnson controls hundred euro companies um that are saying like How do I actually use this as an advantage to do to do better?

1:03:18right like and that's like where i think is interesting is that like maybe five years ago this was really hard like people were like yeah i don't trust it or i don't believe in it i think now what's interesting is they're like i trust it let's go like it's just you can give them you can sit down and give them a demo that's right well thank you so much for coming on thanks so much for joining thanks for having me guys we brave to do a live demo next takes guts yeah great uh any great work i'm a listener thank you yeah love it have a great rest of the conference Thanks for tuning in. You're the man.

1:03:49And we will bring in our next guest, John Cunelov from the nuclear. The man himself. Welcome. Sorry to keep you waiting. Good to meet you. I'm John. Today is a great name to have a company that starts with The. I don't know if you saw The Browser Company. So The Free Press sold for$200 million. The Browser Company sold for$620 million. Everyone is all in on companies that start with The today. There we go. But give us the intro on the nuclear company. What's the plan, and where are you in that plan? What's the plan? So to my understanding, we're the only company in the Western world focused on the deployment of new nuclear.

1:04:32What does that mean? I assume some of your communities probably followed the nuclear industry a little bit. I mean, there's no AI without power. I just talked in that talk earlier about China is about to pass the U.S. as the largest nuclear power in the world. Our thesis is the reactor is not the problem. There's a lot of legacy reactors that are operating in the U.S. They're some of the best performing reactors on planet Earth. There's a lot of startups, dozens, designing new reactors that are all going to be great reactors. The problem is being able to deploy those reactors on time on budget.

1:05:08We have the safest operating nuclear fleet, the highest performing operating nuclear fleet. You're talking about the Navy? I'm talking about the U.S. We have about 100 operating plants. I mean, today, 20 % of the power in the U.S. comes from nuclear. That's nuclear that was built in the 60s and 70s. We've built two reactors in 30 years. So what are we? We're the deployment arm. And what does that mean? So think of if you're American Airlines or Delta, you don't call GE or Rolls-Royce. you don't just call to buy a jet engine. You call Boeing or Airbus. If I handed you a jet engine or a Ferrari engine or a Bugatti engine, no matter how great that engine is, you're going to be like, what are we doing?

1:05:48So we want to be the full solution to deliver that power plant to either a hyperscaler, to a utility, to a foreign government, or potentially to operate those on our own. And the good thing is we're not competing with any of those reactor companies in the market. We're a partner of them. So once they go from R &D to manufacturing, to design, to implementation, there's a big difference between white lab coats designing projects in an R &D lab to living in a construction site where I've done, much of our team's done. I mean, I've built 8 million square feet of stuff at the last thing. Got a team of builders that work for Elon, building gigafactories, built the last nuclear power plants here.

1:06:27We want to be that team that when you're ready to go deploy your reactor, we can partner with you, get that reactor in the field, and get it up and operating faster. Your partners on the reactor side, how much of what they're doing is just remembering how we used to build reactors as a country versus doing that new innovation? So there's really only two incumbents in the U.S., and that's Westinghouse and GE. And, you know, obviously we're talking to them. And they built Vodal, the most recent nuclear power plants, to come online. that were successful, but over budget and over time, correct? Oh, man.

1:07:01It was, yeah, I hired everybody off that team. So Georgia, Vogel 3 and 4, first down. The first Zuckerberg of nuclear. Yeah, what we wanted. No, no, no. We wanted to hire. Like, if people look at that and go, abject failure, I go, no, no, no. These are lessons learned. This is like what went wrong. Guys, it's nuts, man. Like, it took 10 ,000 people at the peak of construction on that construction site. guys go to a rock concert look at 10 000 people and think they're showing up to work every day you don't want an amphitheater just to meet your team 10 000 people managing the project with paper no way dude guys the last decade construction we're not talking 40 years ago i'm talking in the last this thing finished last year with wheel barrels and wagons of paper so you're looking at 10 to 20 efficiency for the people working and you know the audience and the larger viewership might go, ah, lazy Americans.

1:07:59No, I'm not buying it. We are not giving our teams and people the advantages to win. The American spirit and fight alone, God, I'm believing it as much as anyone. It's not enough. We got to bring technology tools capability. That's where we're partnering with Palantir. So I'm taking hundreds of thousands of pages of documents, which is what it takes to build one of these power plants, putting into a data lake, segmenting that data out. So if certain parties want to secure their data, they can't. then having LLMs and AI on top of that, giving predictive analytics. So when the supply chain's delayed the night before, a construction man or woman's waking up in an RV in a trailer at 3 a.m.

1:08:33Okay, I'm going to be redirected at 3.15. I go there at 3.45. I go there. Giving our frontline teams all the tools, technology, and information, we can do it. We're not splitting an atom. We're not going to Mars. We're just building the most dominant AI enabled platform on planet earth. And we're gonna slash that 10 ,000 down to 5 ,000. We're gonna go to seven years instead of 12 years China's building these one gigawatt reactors for 5 billion in five years. There's no reason we can't do it in five or four years I'm not gonna name the number of my team will get really upset with me on the price side, but There's no reason these two reactors took 12 years and 30.

1:09:10So let's talk about timelines in the industry broadly Yeah, because there's some recent I guess I don't know if I can't remember if it was an EO or just a broad directive from the white house saying like we want new nuclear breaking ground in the us in the next 12 months is that is that brother it could be us so we are imminently close to a recovery project that i'm not supposed to talk about so i'm not going to name the state and uh but it's a 20 billion dollar recovery recovery bringing old capacity back online yeah yeah so uh nine billion dollar walk away they spent nine billion dollars on this nuclear two gigawatt nuclear power plant didn't finish it walked away so we are getting brought in we're eminently close if we win that you all should definitely come this tiny little team that's two years old that partnered with palantir to go recover this animal and finish it uh i would love to have you all yeah yeah when when you think about what they spent what is the value that's just sitting there on the dirt certainly not nine billion but you're picking up a couple million legal fees yeah no it's hopefully they poured some concrete no it looks like i mean if you walk on it we're on uh i'm not allowed to say we're right right yeah oh god i almost did um so we're in america we're in america we're playing that american sound effect we are in america we're not afraid to say it we're in America but the when you walk this site and you look at it it looks like you know aliens landed and just left because it's in rural America where this big infrastructure so there's a lot of value there there's been some value that's you know not not quite where it should be but we're gonna go we're gonna get that thing hopefully later this year early next year construction we have an author Dan Wang on the show maybe last week he wrote a book called breakneck and he and and he compares and contrasts China to the United States.

1:11:09And he calls China the engineering empire driven by an engineering mindset. The solution to everything in China is just more engineering. Build a train to nowhere, build a bridge, just build housing, build everything, build, build, build, build, build. And in the United States he calls us the lawyerly society. And we are, everyone in politics is lawyerly or a lawyer lineage. And so one of the problems that I've heard in nuclear is that oftentimes you go to build something, you think, okay, I got a plan. It's compliant with all the laws, and then the laws change. And all of a sudden you're back to square one.

1:11:40You gotta rip out all the pipes because they said no copper. Now you gotta use lead pipes again or whatever. How much of that do you think is real? Or how much do you think, because that feels like something that you can speed up by analyzing all of the legal code constantly and the regulatory filing, speeding that up. But some of it also has to happen on the other side, right? Like it's not just enough for you to be using AI to submit documents fast, you need review fast. So what's gonna happen on the other side? Oh God, I have so many comments on this subject. Just rant. So how long do we have?

1:12:11Yeah, we got a couple minutes. Five minutes, something. So yeah, I mean, this is the hot button issue for me. We have the safest operating nuclear fleet in the world and the highest operating capacity. This industry, don't get me wrong, the legal BS, yes, we all agree, but the victim mentality of the industry, the victim mentality of entrepreneurs in San Francisco acting like high school kids blaming the regulator brother it ain't that hard we hired the number two at the NRC Laura dudes she's on our team we're walking into the NRC going what do you need we're going to be fully transparent we're going to be fully compliant they should be incredibly critical it's nuclear for God's sakes if there is one and here's the other one big misnomer and it's working right the fleet's safe We have had in decades, 100 operating nuclear power plants, not one person in this country has died from radiation fallout.

1:13:070.0. That is perfection. So the private sector needs to stop being a victim and just start doing what we're doing and figure out how to partner with the regulator. We're seeing no problem. So the other kids that want to cry on Twitter, go for it. You want to sue the regulator, go for it. we're just going to go in and partner with them and figure out how to how to build bigger faster lower cost safer higher quality than ever before and i will say what we're doing with palantir well here's the good news to the to the people designing reactors and you're ready to go deploy them what you're doing and what i'm doing have nothing in common i have a team again we let me and my wife were living in an rv got got engaged on the last construction site i've got guys that were building Vogel 3 and 4, had heart attacks on the construction site, had people living at the gigafactories.

1:13:56That is a totally different world. Let us take your drawings, your great R &D, drag it into reality, and we're going to build that trust with the regulator with you. But I do think we got to go pencils down, swords down on blaming the regulator. Now, the legal, you know, that's a whole verse engineer thing. That's a whole nother topic we could take on. But we need the regulator to challenge us to be safe and we just as an industry have to figure out how to comply and get the job done yeah what great rant I would love to see you and carp rant together yeah yeah what what did Palantir show you that made you go with them do was there was there a key case study yeah so we started we're a two-year-old company that's about to be the first the only company in the US with commercial nuclear under our watch I'm like what did we do right what are others doing we're just building a team to go build and and kind of reactor tech and agnostic is the other is the other stuff managed by the government is that what you mean like or is it just older companies that there's no one that's actually focused on building everyone's designing new reactors i just want to go build stuff so i could build a westinghouse a ge reactor you know any one of the new advanced reactors we just want to build so then the last year what we did is we looked at everything i hired somebody over here a lot smarter than me was at tesla was at microsoft um looked at all the different ai platforms what can we do we knew what we wanted nuclear os so nuclear os is the you know again all all aspects of data related to the project into a data lake predictive analytics to our frontline teams no one's even close man yeah this is it i'm not trying to be like a sales job i would like to get like a commission yeah i was going to guess that there's not another great alternative that this would have been nice to at least look at a couple options and decide.

1:15:42Well, here's the good thing. I mean, it's just the most secure platform, the way it is configured. You know, we're going to go build the most dominant AI-enabled nuclear platform, and we're doing it with Palantir. So it took us about a year of study. It took us a couple months of planning, and now we're just racing right now to go kind of build those solutions, and it's working. Yeah, what's the structure of the financial milestones for you? Because I imagine that a lot of this doesn't look just like fund everything with venture capital there's probably some project finance there's actually a customer who may not totally it's paying you just to manage so for our business model so Topco you know the nuclear company you're investing your VC dollars into technology and team which this town knows yeah that big you know buckets of capital project capital I hired a big boy CFO that's raised 10 billion in his life he was CFO with JB at Redwood.

1:16:36Similar to how like the Neo clouds will go and build new data centers, but then there's project finance on the data center. So debt and equity on the project, you know, we're the ones getting it to completion. We can get an equity earn out in the project. We can get a fee during construction. Sure. And then there's multiple, either we could build on transfer to a large utility. We could build on operate for a hyperscaler. We could build on transfer to a foreign government or we could operate it ourselves. So, you know, there's a few ways we get there, but the debt and equity is going on the project not through us now i mean our valuation's not to a point to where i could put 20 billion on our balance sheet yeah uh but i don't know maybe maybe in a couple years let's talk let's see how this goes um so you know we're you know again very just bullish on palantir and i don't know whoever listened to that talk earlier it's i mean the binary outcome is it's us first china and all the tech bros and the badass ceos and the badass five fortune 500 tech executive here's what i would say we got to leave our ego at the door china is kicking our ass that i hope was not recorded everything's recorded we're live so the look it is look the reality is it's not even a competition we're losing so bad and we've got to work together so i would say to the community watching you know push me be hard on me critical on me that's fine but let's figure out how to challenge each other and work together because it's a binary outcome right now it's us first china It's not even close.

1:18:01They're winning at so many categories and we've got to figure out how to work together. And that's what I think Palantir and a unique framework they're bringing, not only the technology, but the mentality of how do we work together and win. And, you know, now it's all going to be about performance on that construction site on time, on budget, high safety. Well, and I love your position in the in the nuclear kind of market broadly, and that if somebody can build great reactors, you can help them actually become a real business based on it and not have to worry about every single point. We got a partner, man.

1:18:34That's the thing, right? This is where China is going into the Middle East, fully vertically integrated going in MBS. We will do it all. One shop stop. They don't want to work with three constructors and somebody selling a reactor. no so like how do we partner together go as a coalition we're going to deliver power globally we're going to deliver power in the in here in the u.s but i do think figuring out how we you know bring down this ego of like there's so many silos and we need to challenge each other but that's what i would say to you all because there's a lot more people in this listen to you than listen to me um how do we bring our tech community together our big ceos who are important and great but if you compare them to China we're not winning so it's like how do we do that and go win collectively fantastic well I think we have our next guest here we're gonna take a look at some rocket motor so thank you for thanks for joining us thank you for doing this work have a good rest of your day up next we have Nancy Cable from Ursa major we will bring her in and do you want us to try and bring that in here what are you thinking I'm happy to bring it in bring Bring in the engine.

1:19:42Bring in the engine. It's an engine, right? Okay. It's a device. We got an engine coming on the show. It's shocking that it was clear through security. When we do these remote shows, we sometimes have to bring very, very suspicious looking Wi-Fi hotspots. Ben and the boys brought a Wi-Fi hotspot through the... Actually, I think I had to walk it into the Capitol through a very odd place. Here, maybe pick up the microphone and we'll throw it on the table. Can you set it gently on the table? Yep, we can throw it on the table. I think we'll be okay. Yep, you set your own gently down. Okay. Oh, wow.

1:20:18Incredible. This is a wild demo. First rocket engine we've had on the show. I'm John. John, I'm Nancy. Great to meet you. Nancy, a pleasure. Thanks for coming on. We're going to hold this as much as you can. We've had people bring fish to the show, sushi, that was extracted, or the fish was killed with a robot, Shinkai. That was a fun one. We had somebody promise us a SpaceX engine, too. Yeah, oh, yeah. We've got to follow up on that. But this is the best demo we've gotten so far. We beat SpaceX in something. This is a good day for us. Yes, so explain to us, what is this, and what's your business, and introduce yourself.

1:20:54Yeah, absolutely. So I'm Nancy Cable. I am the Director of Operations for URSA Major, and we are an aerospace and defense company. So we are deploying primarily right now hypersonic rocket technology, which is what this is. This is our Hadley engine. So a 5 ,000 pound thrust class, proven hypersonic flight capability. So this thing right here has a flown Mach 5. Really critical in the defense space right now. We must field technology and we must do it faster. And that's what Hadley and some of our next gen products are enabling. Now, correct me if I'm wrong. The value of the hypersonic missile is that it has the maneuverability of a cruise missile with like the speed of an ICBM and it's not and so is maneuverability a piece of this is this like maneuverability is a piece of this for our customers so a lot of interceptor technology is what uh current applications and for our next gen products um the maneuverability and the storability of the fuels are also front of mind yeah and uh help me understand where ursa major fits in the overall stack of like the primes and the different supply chain like uh are you developing whole weapons systems that sell directly to the DoD?

1:22:04Are you partnering with other companies that we might be familiar with? Where does our major fit in? Yeah, absolutely. So we're doing, we aim to be disruptive. And disruptive means that we want to break the mold of what some of the primes in the government have traditionally done, which is these years or even decades long deployment cycles of development and qualification. And to do that, we do want to push the industry. So that does mean not necessarily fielding the weapon system ourselves, although that is on the horizon, but putting ourselves in the position where we're partnering with the government, partnering with the primes and forcing them to push the envelope on how fast we can get these products into the spaces that they need to be.

1:22:43So right now, huge focus on just manufacturing excellence, cost, speed, reliability. Absolutely. And that is most of my role is on the manufacturing side and making sure that I can take this excellent technology that our rocket scientists have developed and scale it so that it's available to market. Right now we're on, you know, looking at the order of tens to hundreds of units a year. That needs to be tens of thousands of units a year. And that's really where the Palantir partnership comes in. How does Palantir fit in this? Yeah, absolutely. You might think that engineers are great at data flow, but if we were to look at this rocket engine here, different engineers designed the turbo machinery and the injector and the chamber and all of them came up with a unique way to process their data a unique test system you know a different network drive a different place to store the information and different network drive that is well and I think you know what I think about well you know we have a small company here maybe 10 people and we probably do have like six different like Google drives and different folders for different data Well, that's the interesting thing, right?

1:23:51It's just the natural chaos of things. I think everyone in every industry, rocket propulsion included, ends up feeling like, man, I'm 15 years behind. How could anyone possibly store something on a C drive? But when you're focused on getting the hardware to work, you're not necessarily focused on the efficiency. And so putting the data efficiencies front and center, even before Palantir, our aim was right data, right people, right time, right decisions. I loved what Dr. Karp was saying about people happiness. people are not happy when they feel behind. They are happy when they feel ahead, when they can make real-time decisions.

1:24:25And leveraging Palantir out onto the shop floor and into the back end of our data structures means that we can get the information to people so they can be real-time and then even predictive about how we're doing manufacturing. Yeah, so how does someone at Ursa Major actually interact with Palantir? Is it on the iPad, on a phone, on a computer, while they're working on a test bench? in every phase? Is it everywhere? Yeah, great question. So we've been with Palantir about three months now. Okay, so early, yeah. And right now, the daily interactions are mostly with our engineering and programmatic teams.

1:24:58We've built some inventory modules. We've built in, looking at our engineering line of balance, our change management systems. But like we were hearing from our nuclear, you know, from nuclear, the people on the floor doing the work are actually the most important people in the factory. If my technicians can't build an engine, we cannot deliver to our customers. So that is the next endeavor that we are a few weeks into with amazing results so far is to actually make Palantir a manufacturing execution system. Make it the shop floor portal. One data source, one source of truth, one program from raw material ordering, ordering all of the parts, producing all of the parts internal through fielded data at our customers.

1:25:42Yeah, you almost call it like an ERP almost. Yeah, so we actually, we have an ERP. This is what everyone does. Everyone has, they have an ERP, for an intercise resource plan. Yep, accounting function, all of your work orders. The PLM, Product Lifecycle Management. And then an MES is the traditional thing, a manufacturing execution system. And we have said, why not use Palantir? It's already integrated. I don't want one more monolithic software. Connect it with the ERP, actually pull some of the functions out of the ERP. Because Palantir's better at them. Yeah, I remember hearing a story, I don't know how true it is, but something about like SpaceX built like a ton of custom software for everything they need to do and then eventually I think the team like spun out and built a business around that.

1:26:22Yeah. Yeah, well SpaceX Actually, so they they have a product and it's kind of the gold standard everyone who's worked at space. That's like I want that one That one and that really is the you know, the magic of that software is everything in one place Which is what ontology brings. Yep everything we need in one place. Very cool Yeah. What's it going to take to go from making tens or hundreds of these to tens of thousands? The physical process matters, of course, right? We are a hardware company. You look at the complexity of this and you can understand why we're not going to be forward with a robotic automation line.

1:26:57So making sure we have the right tools, the right fixtures, the right machines, you know 3d printing is critical to what we do here yes 80 % of the rocket all of these metallic components Wow our metal 3d printed yeah developing some of our own unique alloys so scaling the machines is probably the longest lead time for us and then setting up the correct tools fixtures as you can imagine test and infrastructure is really big yeah but not having the data around that in silos so when we need to build hundreds of these I need to know where every piece part is at every moment so that we can make the best real-time decisions possible for quality for the customers so the the physical infrastructure is really what we're most familiar with and now Palantir is helping us with that digital infrastructure side of things I've been in manufacturing my whole career yeah 80 % of the line down scenarios I've ever had we stopped building product you Wanna guess what they're from?

1:27:59Lacking inventory or? It's lacking inventory. It is not having a component. And so we think about like, yeah, a rocket engine is really physically complex. That's actually the hard part. The hard part is getting all the pieces where they need to be to build a 1200 component rocket engine. And it's things like that, that the ontology is helping us solve. A couple of years ago, I said. It's funny, I don't know. I don't know if this is hubris, but I feel like you could put this together, John. Well, that's kinda the point. That's the point, right? I mean, but it's just like so actually putting the pieces together is the easy part But it's like making the parts and making sure you have them at the right time.

1:28:33Yeah is the real challenge So it's like doing a puzzle over like, you know 20 days type of thing. Yeah I mean we joke it's like right Lego Legos for adults, but you can see it really just is a collection of fittings Fittings and fasteners I and that's kind of the point How can we have a system that makes it so easy and so obvious how we manufacture these that I could pull the two of you in and say they build a rocket engine, and you could do it with confidence. That's what we're after. We've got young kids, I think they would enjoy putting one of these together. Yeah, a couple years ago, I sat next to somebody on a plane who was selling, what was it, pipe bending, pipe fitting?

1:29:07Whatever this is. Tube bending? Tube bending, yeah. He said, I'm in, my business is tube bending. And I was like, what? And he was like, yeah, he was going to SpaceX, specifically to sell tube bending machines to them. I didn't realize it was a whole industry. But he has to be there. He made his money in bending tubes. But he has to be there in person to make sure that they don't run out. Absolutely. Because it's a whole industry. If the tube isn't bent, you can't make the rocket. If the tube isn't bent, you can't make the rocket. Crazy. And tubes actually carry some risk. They're some of the thinnest walled components on the rocket, right?

1:29:34This has a lot of mass to it. Tubes are often, can be, where failures happen. Sure, sure. So in an ecosystem, we need to test them, but also where did this tube come from? What day was it bent? What was the lot of stock material? What revision was I on in my CAD model? Sure, sure, sure. What testing did this engine undergo? all of that currently I could find in our systems. And it would take me hours, but - But if it's all in one place. If it's all in one place and we have a consolidated tool, it's that traceability. That's incredibly cool. Fantastic. Anything else, Gruden? Thank you so much for bringing your baby on the track.

1:30:07It's a great sign of respect. Yeah, absolutely. I mean, what's cooler than carrying around a hypersonic rocket engine? Everyone loves it but the TSA. They don't love it so much. It's a rough one to travel with. Yeah. Anyway, thank you so much for coming on this stream. Yeah, absolutely. Thanks for coming on. Thank you so much. Thank you. We have our next guest ready or should I talk? We have a couple minutes. Why don't you tell us about some ads? Do you have some ads you can run? I'd love to hear some ads. You want to talk about ramp.com? Ramp. Oh, you want to hear the ramp song? Ramp. Ramp. Ramp.

1:30:39Let's go through some of the web supporters. I did want to, while you pull that up, I did want to talk about Matt Huang. Oh, yeah. Paradigm and the Stripe team introducing a new payments-first blockchain called Tempo. Matt says, as stablecoins go mainstream, there's a need for optimized infrastructure. Tempo is purpose-built for stablecoins and real-world payments, born from Stripe's experience in global payments and Paradigm's expertise in crypto. To ensure Tempo serves a broad array of needs, we're excited to be working with an incredible group of initial design partners, including Anthropic, Coupang, Deutsche Bank, DoorDash, LeadBank, Mercury, NewBank, OpenAI, Revolut, Shopify, Standard Charter, Visa, and more.

1:31:21Tempo's payment-first design includes predictable low fees, payments, gas, and any stablecoin, payments-first UX, opt-in privacy, scale, 100 ,000 transactions per second, and EVM-compatible built on Reith. Tempo eases the path to bring real world flows on chain such as global payouts pay-ins and payroll embedded financial products and accounts fast and cheap remittances tokenized deposits for 24 7 settlement microtransactions agentic payments and more matt says we're building tempo with principles of decentralization and neutrality that includes stablecoin neutrality anyone can issue a stablecoin we might be able to have a tbpn coin that sounds exciting and any oh yeah yeah That was clearly a joke.

1:32:10No, but I was talking about a USDT VPN. Yeah, one for one stable coin that that that we issued to does not move. It does not move. You can't make it move. It won't budge Independent and diverse validator set with a roadmap toward a permissionless model. So apparently they're already in a private test net and anyways to two power players Paradigm and Stripe coming together. It sounds like they're positioning, I guess Matt is running Tempo, but they're positioning this as they're both investors in Tempo. So I think they really do want to take a decentralized approach. So this is not downstream of like the Stripe acquisitions directly, Privy and Bridge?

1:32:56Yeah, so I have a post here from Zach Abrams, founder of Bridge. He says, Bridge was one of the first companies to use blockchains to solve core payments problems. During our journey, we've seen how even the most performant blockchain struggle with basic financial services use cases. A few examples, a payroll transaction consistently failing when Trump launched. That's interesting. So when the Trump coin launched, apparently people that were running payroll couldn't get... On Bridge with stable coins? No, no, no. He's not talking about Bridge specifically, but he's saying if you were trying to pay employees at the time the Trump coin launched...

1:33:31Who's paying employees in Trump coin? No, no, no. Not in Trump coin. Like that that day, I think it was like a Saturday or Friday, I forget exactly. But when it launched, if you tried to pay, there was so much activity on chain at that moment. Okay. Like good luck, you know, paying like a freelancer. So yeah, the example would be like, I'm trying to pay a freelancer in stable coins, like on chain, because obviously like your default payroll providers are just using like, you know, web to rails or whatever. And that wasn't brought down by the launch, right? Okay, got it. um aid disbursements taking days due to low transactions per second and projects to later cancel due to six-figure upfront gas costs um tempo is new l1 built specifically for payments and so um anyways uh quite the team they've put together here yeah we got to get some of the folks on the on the show and and have them break it down because um i'm very interested in why not solana why not circle?

1:34:28It feels like there's a stable competitor. The other question is why not another L, like why not an L2? Exactly. Built on. Built on. But this is something unique and they must have put a lot of time and effort into it. So congrats to them on the launch, but we will want to know more. Anyway, I believe we have our next guest. Welcome to the show.

1:34:48Ryan. I'm Johnny. Ryan. It's my pleasure. Welcome to the show. You hold this microphone. Why don't you kick us off with an introduction on yourself and what brought you here today? Perfect. I'm Ryan Estorian. I'm the Chief Marketing and Strategy Officer for Lumen. And we're here at AIPcon talking about all the great things we're doing together to modernize telecom. Lumen's a telecom company. Let's give it up for modernizing telecom. Yeah, exactly. Finally, finally. It's fun because it's decades of complex operational. Palantir is helping us modernize into this new world that you need for AI ready multi-cloud world that is what everyone's here talking about.

1:35:30Yeah, how do you define, break down more of what you do in telecom specifically? Yeah, so Lumen is, for decades, we have basically been connecting the world. It starts with connection and then in the last bit of time, the world has needed new ways of connecting. We're bringing that infrastructure, which we're bringing control. If you think about the way it was before, it was one port. Like fiber, cables in the ground? All fiber, right? Everything that's running across fiber. Those super fast connections you need, one port, one connection was the way of the world. We're changing that. We're getting it cloud ready, cloud enabled, remote controlled, all of those things that give you that redundancy, latency, all the things that power AI.

1:36:15That's what Lumen is doing and we're connecting the world. Okay, who's the customer right now? We have lots of customers. So we're really focused on the enterprise. The enterprises that are building these new capabilities, data center operators, hyperscalers, of course. And so we've announced some of the work we've done on the backbone, the infrastructure, backbone of the AI economy. But what we're really doing is enabling businesses new things, new technologies that they want to give them a technological advantage. we're disrupting this industry to help them disrupt their industry. Yeah, yeah, yeah.

1:36:51So, I mean, obviously there's like an immense amount of money flowing into data centers. Is a lot of that actually going into like new bandwidth requirements between data centers? Like the basic narrative is like, yeah, they might spend a billion dollars training something, but it's all happening within one data center. Well, so the thing you hear about a lot, and you guys have talked about a lot as well, is compute, storage, cooling, all those things that are needed. the missing link is connectivity. And realistically, it's something that has really emerged as of recent to say there are new types of connectivity, new next-gen fiber that has way more capacity than the world has ever needed before.

1:37:30We're growing leaps and bounds. By 2028, we'll have about 66 million route miles of fiber. And that is growing 3 to 5x what we've had before. Okay. And that is the capacity the world needs. Yeah. So there's some sort of - And is that capacity being used inefficiently today or is demand still way outstripping supply? The demand is completely maxing out. It's why we are putting these investments in the ground. And we're not only - The hyperscalers are, I'd say, the tip of the spear. They're consuming a lot of this. they're looking for a lot of this data center to data center connectivity, but it's really enterprises everywhere that are now saying, you know what, we also need that type of bandwidth.

1:38:18And some will take it dedicated, some will take it shared, but the need is completely outpacing what the needs of the last couple of decades have been. Yeah. Try and make that more concrete for me. Because I feel like most people's interaction with AI is I send the most condensed packets it's possible across the internet, just a couple lines of text, and then a bunch of GPUs light on fire at the AWS data center, Azure, if I'm using GPT-5, and then it sends back text. This is not rich video, this is not VR. I buy, I immediately intuitively understand, if we're in the metaverse world, and we're streaming 4K, stereoscopic, that's super bandwidth heavy.

1:39:00How is AI bandwidth heavy? So it's actually great listening to the customers that have been here at AIPcon because you hear American Airlines, you hear BP, you hear some of these customers that are talking about their infrastructure. All of the scheduling, the inferencing, the planning that is happening in real time and adjusting, that is not just people typing in their prompts into the text. It is systems talking to systems. And this is where the data explosion has come from. It's all happening in the background. Okay, yeah, yeah, yeah. So even though I fire off one query to GPT-5, if it's doing deep research, it might be pinging 75 different websites and that's driving up total internet use.

1:39:42Yes, and the systems are also creating their own queries. Yes. Yeah, we saw that with the demo from Palantir. He typed one line of text to help optimize this airport. And then it was like 20 employees were doing... That's right. Okay, yeah. That's right. And so this is where the disruption in telecom comes. And if you really think about what has changed in telecom over the last 25 years, the answer is not much. When you can take one port and you can put lots of services on that port and put the control in the customer's hands, you've changed the way people interact. It's cloudifying telecom. And in this new world of what is happening with cloud, like cloud 2.0, that is the necessary bandwidth control and precision that you need in connectivity.

1:40:32What does cloudifying telecom mean? Does that mean more like multi-tenant on the actual fiber lines? Like instead of a hyperscaler owning one route, then they're bidding it out and spot rates or something? Yeah, multi-tenant is a good way to think about some of the services on top. In the past, you've literally, if you think even back to old telephone switches, you've had the one wire to one wire, it's been one port to one service. You add a service, you add a port, it's a truck roll, it's a person coming out. Cloudifying it is bringing all of that technology to the users, giving them that interface, that portal where they can say, I need these services, I need them in these locations, I need this speed, I need the bandwidth turned up.

1:41:14It's network as a service. Yeah, so higher level of abstraction. Yes. And yeah, more like almost like a virtual machine on top of the telecom infrastructure so it can be provisioned like on an ad hoc basis. Yeah. And one of the biggest changes I think in the economics, this AI economy is also if you think about a network subscription, if you will, of the past, you sign up, you get a certain amount of bandwidth. But if you look at the companies of today, if you look at the sports industry, manufacturing industry, healthcare industry, they have these spikes that are massive. And so we're providing that network as a service where it turns up, turns down, and then customers are paying for what they, it's a consumption model.

1:41:58And again, that's part of this cloudifying model, which has not hit telecom till what we're looking to transform. So yeah, help me understand the new shape of the the telecom industry in your business, like I imagine that there's some genius scientist that comes up with a faster fiber optic cable that is manufactured somewhere, then someone purchases that, they buy some land, they bury it in the ground, maybe they get some rights, and then at a certain point someone's leasing or essentially charging a toll along that toll road. Do you sit all, are we completely vertically integrated? So we sit vertically integrated, but I think what - So you do R &D on new fiber optic technology?

1:42:38We work with a number of partners on that. And then we're also thinking about the AI optimizations on that fiber. So if you think about intelligent routing, if you think about redundancy, if you think about all those things where you could have something as simple as a fiber cut in the ground. Maybe it's on purpose, maybe it's not on purpose, but something happens. You need to be aware of that and then you need to dispatch someone to go fix it. You can't have any interruption to the services you're running. So we have to have that redundancy. Yep. On top of that, our customers and enterprises everywhere, I think they started mostly building with one cloud.

1:43:11Now if you think about this multi-cloud world where they're hitting Azure, GDC, AWS, they're hitting all of them at the same time with the same applications in different regions across the US. They have to seamlessly let those systems talk to each other. Yep. And they don't want a direct connection to each of them. That's where we started. But now they want to be able to live in this fabric where their systems can talk to all of these in all the regions, get all the data and process faster because that's part of the disruption they want. Last question for me. How does Palantir fit into that? Yeah.

1:43:48So if you think of the operational complexity of the decades of past, you know, you've built all these networks. We talked about fiber in the ground. Think about the systems over those decades that have been built up. One of the things Palantir is helping us with is managing this operational complexity. You sort of see an abstraction of this in LA when there's the fire and the boxes with the telephone lines just explode. You're like, why didn't they build a box that doesn't explode? And so you imagine that, okay, that's where the power lines work. The fiber optic lines, yeah, they're newer, but there's probably still some stuff that that might go wrong if it was installed 30 years ago.

1:44:28So you gotta identify that early. There's that and there's the software layer that is running all of those. Gotta make sure that that's up to date and not crashing. And Palantir's helping us optimize those, helping us bring them together. And what we are building for customers is then a system that they don't have to think about the optimization they need in their network. We're gonna help automate that, we're going to help bring AI to that network. And that's part of this partnership. And it's also, frankly, the most exciting part about disrupting telco. It's not an industry that too many have talked about disrupting for a while.

1:45:06It's ripe for it. It's needed. And this AI multi-cloud era, Lumen's here for it. That's very exciting. Anything else, Jordy? Love it. We're running late, so thank you so much for wrapping up. Thanks for having me. All right. I'll grab this. Thank you. we have our next guest coming into the studio drew Cucor I think we actually have multiple we might need to pull up an extra chair we have a little hands we have lads coming in if we want to bring everyone in we can will we can pass the mic around whatever you whatever you guys want to do we have multiple oh okay hey oh how you doing welcome how you doing good how's the day Could you kick us off with an introduction for those who don't know?

1:45:54Okay. I'm Dave. Dave Glazer. I've been a Palantir for 12 years. And I'm a CFO. Pre-IPO. Pre-IPO. Yeah. Or DPO, right? DPO, yeah. Since when our prior CFO retired, he was actually on the show recently. Yeah, Colin. I talked to him. He retired in 2017 and since then. Cool. I've been leading the finance team. Yeah. So my big question for you. gross margins for the Fortune 500 in the AI era, are we gonna see a structural shift? You know, the inference bills are skyrocketing, inference per token is dropping, but then Jevons Paradox, and we're doing more token inference than ever before. Reasoning models are kind of staying expensive, and we saw in the journal earlier this week, maybe last week, a software company called Notion said that they saw their gross margins drop from 90 to 80%, not bad still.

1:46:47But there is, does seem to be some sort of impact. And I'm wondering how you think it might play out for the really big companies. Yeah, look, I think this is one of the things that we've been sort of saying is like LMs are commodity cognition, right? And so like, essentially it's like, it's getting, they're getting better and better, right? ELO scores better and better, tokens are getting cheaper. Right? And as Alex said, I don't know if you've watched, you know, but like, you know, he's talking about, okay, like how do you actually derive value from that raw output of an lm and so it's like i think it's like the raw output it is getting cheaper we're still like very early days on these models and you're seeing them just sort of like up into the right in eloscore and so these like things combined i think are going to make it cheaper and cheaper over time and i think we'll see sort of on on gross margin i think you look at some of the other things like hyperscaler costs right from a lot of places i think like people their people's gross margins have survived right they're more efficient that they're all this.

1:47:40And so I think like we will see, but like I think that is, it's gonna be much more about like, how are you deriving value from them? Then like, well, the cost is gonna be so overwhelming, but they're super, like, totally, it's like focused on the value. And I do think over time, it's like people are gonna be able to manage those costs. Yeah, yeah, it feels like, it feels like higher costs potentially, but so much more value. And it's pretty easy to tell, yeah, I'm spending a lot on inferencing a certain LLM API, but obviously I'm delivering more value. And so I'm charging - Also, you have to think about the position that Palantir sits in.

1:48:11We got a product demo earlier. Hivemind was leveraging a bunch of different models. And that position of having leverage and being like, we are the product. We have the data. We have the customer relationship. And we can vend in whatever intelligence sources we need in order to accomplish the task. That's a better position than being if you're a GBT wrapper and your product is really 4.0 and you're just kind of reselling that. Yeah, yeah. Yeah, sorry. Yeah, and I do think it's like, yeah, I think it's going to be all about the value rather than like, well, the value is there, but the cost is super derivative.

1:48:45Yeah. How are you thinking about positioning Palinger's story in commercial in the United States over the next couple of years? Like, what is the right framework? People have always had the wrong mindset. It's a consulting shop. What do they even do? Blah, blah, blah. Like, what is the right frame of mind to be in? Look, I think the right frame of mind is like we're delivering a tremendous amount of value. Yeah. to these customers, with these customers, right? It's like, and they're needed too in this, right? And it's like, you deliver that value and we're just at the beginning. And so you look at our US commercial business, like grew over 90 % last quarter.

1:49:19It's still relatively small, right? And it's like, we have, there's so much runway there. Yeah. Right? It's like, just that business has like sub 400 customers. Yeah. Right? Like that is, when you look sort of across a lot of other companies, it's like, that's, you know, and so it's like, we're doing all this with such a, like a small customer base. And obviously it's rapidly growing, but you know, it's like it just shows the amount of runway that's ahead. Yeah. Do you do you think that people should be thinking about the commercial business as like a bundle, like a competitor to a bundle of products that already exist or something that's entirely net new or displacing an entirely different class of spend in the enterprise?

1:50:00Like how can people even wrap their mind around some version of all of the above? So it's like when you think about, you know, you're not like head to head who are we competing with, right? And then everyone's like, but I don't get it. It's like, is this a combination of, right? We're not really, we're competing against like the Frankenstein monster that almost every large corporation has. And then you're also competing like particularly in government. But it also applies in, you know, particularly large corporations is like custom built software. So it's like those two, you're competing against that.

1:50:26And over time, you're obviously going to sort of eat into a lot of the spend. but it's like only because of the value that's being delivered and then it's like you don't maybe need some of these Yeah, yeah, it feels like the it's like it's like Transformation new net new technology that would not get built in the enterprise otherwise Correct And then once you've built that once you've got a data asset then perhaps you don't need Yeah, so the other practice. Yeah, how is your framework or philosophy? Approaching the finance function at Palantir change because I feel like there's like very distinct eras where you know Yeah, it changes that every day.

1:51:01Do you feel like you have to update it every day? Because in some ways, when we talked to Karp earlier, it's like he's bringing that same energy and philosophy. It feels like it's somewhat consistent, even though numbers go up and down and all that good stuff. Yeah, look, I challenge any CFO working for Karp to have hair. Right? So look. I think you've got to step back and say, okay, how do we approach finance? And it's like, this is a company, and people have said it a lot, we don't have a playbook. And obviously, there's a way that's run, the company's been built over the last 20-ish years. I've been lucky enough to be here for 12 of them.

1:51:44And because of that, it's like, we're very unique. And what that means is we are constantly changing what we're doing. And so a lot of things, you talk about forward-deployment engineers in the early days, that's consulting, that's this, it obviously helped us build the product that we have today. right and so what you weren't what you weren't optimizing on in those days was financial statements that wall street would want right because it's and and then it's like but because of what we built today not because or because of what we built we have financial statements wall street loves but it wasn't built for that purpose right and which is crazy valuable right because it means we're we're so differentiated and we're doing things the way that like we want to do them and the company was built that way.

1:52:26Can you tell me the story of how the COVID era changed Palantir's financials? I remember seeing that T &E fell off a cliff and it never really came back. And that was, at the time I was talking to some people who were looking at the company, they were pretty excited about what that meant. And it felt like it was almost like a structural shift for the company. But is that a reasonable story to tell? Is that apocryphal? Look, it's part of the story, right? And so I think like what happened with COVID, it was we could no longer, like you just couldn't be as much at a customer site, right? And so then it's like, well, we gotta extend the product further, right?

1:53:04And like, and this is a story that keeps happening in Palantir. It's like, well, you know, we only have, you know, around 4 ,000 people, right? And, or you look at sort of our headcount growth, like if you go back two years, it's up 12 % from two years ago, revenue is up 88%. It's like, well, how do you do that? It's like, well, the product's got to be better. Right? And you have to have products like AI, FTE, like all these things that are constantly evolving. And like that is the story of Palantir. It's like you're trying to do something, you're either resource constrained or somehow constrained.

1:53:33It's like, what do you do to meet that? And almost always is product-led. Yeah, that makes a ton of sense. I know you have a busy day, so we'll let you go. Awesome. Thanks so much for helping. Thanks for joining. We'll talk to you soon. We will bring in our next guests in a minute. Jordy, do you have any breaking news? I got a post here from Skooks. Skooks says, Alex Karp trying his best to get TBPN banned from YouTube. I will say, I think it was like the least family-friendly 10-minute segment of the hundreds of hours that we put out. It was some of the best. Some of the best. Some of the best.

1:54:05It was a lot of fun. I'm glad that Skooks enjoyed the stream. And thank you for YouTube for keeping us up. Keeping us up. We might be... Stream's going strong. Thank you to Restream for keeping the stream live. Thank you. couldn't do without them we'll bring in our next guest guests we are ready to keep rocking and rolling here we got at two chairs two chairs coming in come on in come on in

1:54:36very cool fantastic how you doing good to meet you I'm John Hey, Kyle. Pleasure. I'm John. Nice to meet you. How are you doing? Good to be here. Lads. We got the lads. Take a seat. Take a seat. Do you guys want to share? Yeah, we'll share. We'll share. Great. So, yeah, why don't you two kick us off with the introductions. Let us know who you are. I'm sorry you got stuck with a rough chair. I couldn't figure out how to get the chair to sit up properly. I should know. Don't even try. It's not going to work. I already tried it. Anyway, introduce yourselves. So I'm Zach Porter, a senior simulation engineer with Andretti Global on the IndyCar program.

1:55:17Cool. And I'm Kyle Kirkwood, driver of the number 27 Honda for Andretti Global. Fantastic. Yeah, and I'm Drew from TWG. Fantastic. How do all of you fit together? We're all under the TWG umbrella. Okay. Basically a bunch of different businesses within that. Drew can probably speak to it a little better than I can. Yeah, please. Yeah, I mean, it's a family. It's a great holding company. We have tons of businesses from insurance to asset management, investment banking, and sports, media, entertainment, Western lifestyle. And of course, the crown jewel of just about everything is the awesomeness of motorsports and the Andretti team and IndyCar.

1:55:56How long have you been involved with Andretti? It's my fourth season at Andretti. Fourth season? Yeah. Third season. Losing track of time here. I think it's my fourth. No, it's my third. It's my third season with them, but I've also, I've been a part of the family for longer than that. I was with them in Indie Lights, and then I joined back with them in IndyCar. So really five seasons, actually, if you combined it all. Yeah. And, you know, I get to be this suit guy. So I sit and watch this, but I've been here a year. Oh, fantastic. Yeah. And, yeah, and walk me through the flow of, like, why you're here specifically at AIP Con.

1:56:30Why are you working with Palantir? Yeah. So in IndyCar, we have a ton of data. Yeah. in a ton of different siloed places. It sits from stuff that we control, like our car setup database and stuff, but it also sits in databases from IndyCar that we don't control. We have to consume all these things, and they're all connected. They all represent performance. They all represent the pieces of the car and how they go around the track and how we get faster and how we're relatively performing against the competitors. And so we came to Palantir and worked down this path to try and connect all these disparate data sets into one place where our engineers can make better decisions faster, sooner, because in the end, you know, from practice one to practice two or practice two to qualifying, whatever it is, there's this limited amount of time that we have to make a decision.

1:57:12The practice is coming whether you're ready or not. So the more informed we can be, the better decision we can make in theory, the faster we can iterate and be more competitive. So yeah, it feels like the, maybe we're just in the era of like, you know, small micro optimizations just add up to greatness. Are there any stories from your career or just racing in general that stand out to you, where someone just discovered some secret that just gave them a massive advantage. I'm thinking of in sailing, there was this, maybe it's a fake story, I don't know, but this idea that there was in the, what's the big sailing cup that Ellison races in?

1:57:50America's Cup? Yeah, it's all catamarans now. And the story goes that they were all racing monoholes, and someone looked in the rule book and said, There's nothing that says you can't bring a catamaran and then in one day somebody brought a catamaran and just beat everyone And it was just one of the most fantastic stories. Have there been any eras? That you've studied where someone's just figured out something that just rewrote. I mean it would never be like this again But you had the the fan car and f1 right? Yeah, tell me about this. Yeah. Yeah, tell me the full story I don't know the full story We're in an era of motorsport now that things are super tightly regulated sure really hard to find these big gains But what he's referencing back in the day, there was an era where aerodynamics were kind of king.

1:58:35And the guys did a similar thing. They looked at the rulebook and said, hey, there's nothing that says we can't power the air inside the car on our own. So they built a car that had big fans at the back of it and skirts that ran down the side. And the car literally sucked itself. Sucked its way down. So just so much extra downforce. I don't remember exactly how long it existed, but it wasn't very long. I'm sure it got banned. It's amazing. But it was fundamentally dominant. And there's been a lot of those kind of things now and over time. But now we're kind of in this era of fighting for these hundreds of seconds, these little micro moments.

1:59:02And that's where being able to drill down through big data is powerful for us. Yeah, yeah. I can imagine we'll be like, we do a live show, right? So speed and timing is important. And sometimes we're like, oh, this document isn't here. We don't have this link and things like that. You guys are racing around a track where every millisecond matters. And so if you're jumping between different data sets and systems of record, I can imagine that can be a disaster. Yeah, and it's not just while Kyle's on track. Yes, he's doing all of that. But then as soon as he's back, it's between sessions as well.

1:59:38The clock's always ticking. We're competing on the track and off the track. Yeah, I mean, we just have such little time to go through so much data. And to be able to piece it all together and understand a full picture, you have to do a lot of different things, which our engineers are very good at. But it's time consuming. So if there's a way to actually consolidate it, simplify it, and make things more efficient, then it's gonna allow our engineers to make better decisions down the road, which is optimizing performance on the racetrack. Okay, talk about the tension between the three of you. I imagine that you only care about speed, you care about speed and manufacturing.

2:00:10Can we make it? And you care about speed, manufacturing capability, and cost, maybe? Cost. Cost? So what are the trade-offs? Obviously, everyone cares about speed and winning, but there are layers to the trade-offs because you can't just always turn every dial to 11, right? Well, I mean, look, I spent 30 years in the Marines. Yeah. And we got tired of fighting wars on PowerPoint. And for business, we're getting tired of making decisions off of rudimentary and incomplete systems that provide only partial solutions, and it just takes forever to get data together. And so from a business perspective, we have to look at it and basically say, look, we want to transition to something better.

2:00:53And the cost of that is not just material like dollars. It's also change. It's changing mindset. And as you can see from Andretti, like they're all into this. Like this team is ready to make that transformation, but it'll still come at a cost, right? There's people who are stuck in their ways. Look, I like to do things this way. I'm not used to that much data coming at me. I can't make decisions that fast. Like this is transformational and really fundamentally it's people, money, it's organizational, and obviously when you've got a great team, like it's just gonna go like a hot knife through butter.

2:01:25It's gonna be amazing. That's great. Yeah, walk me through some of the benefits and try and give me some anecdotes about where gains have come from throughout your career. Yeah, I mean, like for us, we take in so much time series data on the car specifically, that's the representation of what Kyle's doing on the track, and what the car's doing and all of that, and being able to connect that data to his feedback and ensure also that that data is clean and it is correct. You know, it's not like a car that's just rolling down the road and it's hanging around and putting some sensor data out. Like he's flogging the thing around the racetrack and occasionally touching walls and other cars.

2:02:04More than touching. It's really difficult sometimes to make sure every system is working perfectly. It's a never-ending battle of trying to do that. And so, you know, we're working really hard with some ML models and some stuff to pick out sensor anomalies and flag them automatically so that our systems engineers don't miss them and they can go drill down and figure out why that sensor's failed or where and what their knock-on effects are and in the end just get that part replaced immediately so that the next outing, the next time we're on track, we know the data's going to be as good as it can be.

2:02:30That's been the earliest, easiest wins for us is kind of in that space. Yeah, yeah. Is there a, how do you think about budgetary constraints? Is that something that's just set internally? Like how do you work? I'm happy that I don't have to worry about Even zooming out for those who might not be familiar like I mean we saw some we saw some drama earlier this week about salary caps and And different ways to get around things like how do you think about setting the budget for the team? And then actually executing against that because that's got to be the last the last phase of against How do you actually deliver something that you can deliver on race day every single day with reliability and not need to cut the cost later let me let's talk like this is innovation yeah okay so we got to be careful here yeah right so if you come in I mean obviously there's dollar budgets right because it's not a constraint yeah but at the end of the day like what we want to do is we're talking about a fully connected business here sure so they've got an HR shop they've got a tech team they've got engineering they've got a ton of groups that all need to be brought together yeah so apart from just the car and the magnificence of what we're doing, you've got to bring it all together.

2:03:40And so we need room and space to be able to build out a complete connected business. Because frankly, every signal across the business is value. And by squeezing and optimizing and making things run more efficiently, we end up with a better sport. And I think at this point, we're in that journey. And so costs are going to be not giant, but constrained. And we're going to deliver. And we're going to watch and see as this evolves until we land somewhere where we can finally say this is it, this is the benchmark, and this is what we should manage off of. For us, we're going to ask for every tool we possibly can to make the car better.

2:04:16He's expecting us to do that job, and in turn we turn around to the commercial side of our business and look at them and say, hey, it's your best job to go out and find that sponsorship, find those things. Because if we don't use this tool, our competitors will. And we're in the business of winning. If we're not going to try to do that then why are we here? Take us through the next few months on the calendar, the rest of the year, the next year. So we literally just ended the last race of the season like three days ago, four days ago. So we officially start our offseason, and this is where we sort of take some of our use cases and our ideas that we've sort of half-baked and trialed some stuff and look at it and productionize it.

2:04:50Sure. And in the end, try and get all of these, or at least the first initial use cases ready to go for St. Pete 2026. Yep. That's kind of the target, and there's a ton of prep from here to there. Yeah. And I'd say in the off season, racing is so expensive that you you're limited on how much testing you can actually do on a racetrack. Right. So it's very important that all the data that we collect and we utilize is is actually making a difference. And we're actually able to progress with with with the data that we have. So that's where the engineers come in. Right. We've got a massive group of engineers that take a lot of pride in their work.

2:05:28and they have five, six months from now until the start of the next season that they dig in through maybe one or two tests that we get, maybe some wind tunnel stuff, maybe some various other things, shaker rigs, we call it. But we can't really get on track that much because of how expensive it is. So a lot of what we do is in the sim world and it is very data-driven. Yeah, what does the rest of your off-season look like? Are you training, running? I saw the F1 movie and Brad Pitt's always running around. Are you running? Are you a technical guy or both? uh you know it training is important right uh yeah i mean you you have to be as a racing driver you got to be like a certain weight certain size you have to be um you got to have good endurance but you also need to have some strength to be able to wheel the car around right right we don't have power steering you're hitting the brake pedal as hard as you possibly can and we're pulling up to four or five g's for an hour and forty to two hours at a time so it can get very physical very fast no power steering no in the car and the car makes over five thousand six thousand pounds of downforce so um imagine driving your road car that weighs eight thousand pounds or something like that um around without power steering flash that on the screen when they're when you got the driver view so that you guys get a little credit yeah people assume it's like turning the wheel of you know a tesla or whatever yeah no it's uh it's much tougher than people tend to realize i that's specific to IndyCar racing though.

2:06:50IndyCar racing, we don't have power steering. F1 does. A lot of sports cars that you see, they do have power steering, but IndyCar itself, they do it for the sport and they've kept it that way for many years. So it's a little bit old style, but at the same time, it's good because it really translates it. A little bit, right? Yeah. It creates a sport out of it, right? It's a little bit more physical. People don't look at it as much as like, oh, you're just driving a car around some roads, right? Pushing pedals, turning wheels. No, there's actually a physical side to it so um the off season is a lot of training uh preparation we do a lot of sim work and uh driver in the loop simulators and um yeah it's just being ready for for the next race that comes up it's hard it's hard though because you don't have g-forces you can't you can't simulate g-forces for a driver so um having that involved is um is something that you get acquired to as the season progresses if i'm being honest yeah uh what's your daily i'm sorry What's your daily driver when you're not on the track?

2:07:45My daily driver. So that is one is that is one of the great things about being a racing driver is you don't have to own a car. Oh, you don't. Yeah. You. So I race for loaners or something. Yeah, exactly. So I race for Honda. Okay. And then the car and I have a. No word with. You have glow on the S2000. I have a Acura MDX. Very cool. They're their sister companies. Right. And then I also. They're not sending you an NSX. They don't make the NSX anymore. They still got them laying around. Give them a call. We'll talk to them. We'll see. We need them ripping around in NSX. And then I also race sports cars for Lexus as well.

2:08:24Okay, cool. For LFA every day, obviously. They also don't make an LFA anymore. Yeah, just a million,$2 car. I can just go rip and depreciate real quick. I have an IS 500 at home. So that's the other car. Fantastic. Well, thank you guys for coming on. This is fantastic. Anything else worth sharing before you get out of here? Okay. Enjoy the rest of the conference. Thank you so much for helping on. Thank you. We will talk to you soon. Cheers, guys. Have a good one. Thanks. Goodbye. Jordy, any other breaking news going on? We have our next guest coming into the studio in just a minute. I believe we have.

2:09:01What do we got? Who do we have? We have someone else coming on? Okay. Okay. Cool. Yeah, yeah, yeah. We're good whenever. We kind of ran late. now we're now we're running a couple minutes early we will keep it going oh yeah Palantir CEO Alex carp thinks the value of skilled workers is spiking even as big tech companies possibly his own may shrink our revenue is going up our sales force is going down he said on TVP and the number of people we plan to have in the future is less than now very cool we scoop we're scoop maxing we're Newsmaxing everybody. What else? I think we're ready for our next guest if you want to.

2:09:43This is from the timeline. Looking good. Lots of posts. Having fun. Welcome to the stream. If you're ready, we're good. We can we're happy to have you. How you doing? What's happening? Nice to meet you. Thank you so much for taking the time. Yeah. Welcome to the show. Thank you. Any relation to Brandon Jacoby with an I? I don't think so. I think you guys know that we have a buddy who works. He's a designer and we like to we like to put fun of him because he is. We call him Jacoby. And whenever we have a design problem, we always call him. The last name sticks with that one. Yeah. Anyway, please introduce yourself for the streamer.

2:10:13Who are you? What do you do? Happy to. Sorry, I'm out of breath. That's a lot of serious. You're good. You're good. So, Matt Jacobi. I'm the head of data science and analytics at Racetrack. Southeast-based fuel and convenience retailer. And shout out to my wife for letting me come up here, because we're technically on vacation this week. I heard this. This is crazy. The grind never stops. You couldn't miss AIP. I got the memo about lock-in season. Well, it's you, gentlemen. I couldn't pass up the chance to participate. We really appreciate it. It's great to have you. Yeah, so break down the business a little bit more.

2:10:43Give me a sense of the scale, what the day-to-day is like, customers. Obviously, we have a general idea, but give us more. Yeah, yeah, happy to share. So roughly 700 retail locations across our family of brands of Racetrack, Raceway, and Golf. A lot of people don't realize that we own Golf. Oh, we own Golf. Yep, yep. 10 ,000 employees, associates in our stores, and people at our store support center in Atlanta. A lot of people don't know either. We're top five largest privately held company in the state of Georgia, and we are top 15 in the United States. Thank you. So walk me through a little bit of the history of the company, because I imagine that what we're going to talk about in terms of software, where artificial intelligence is a revision to the way it was done years ago, right?

2:11:31So yeah, walk me through a little bit of the history. Get me up to speed. Oh, wow. Well, I can't speak to all of it. I've been there about two years. But what I can say is that we've done a really great job of focusing on transformation, specifically data-enabled transformation. Actually, I just wrapped up a conversation about this downstairs. But if you ask me, one of the purest use cases for transformation is converting from gut-based and tribal knowledge-based decision-making to data-driven, and therefore, after that, analytics and AI-based transformation. So we've really focused heavily, even before my time, on making the best decisions we can with data.

2:12:11And so our partnership with Palantir has really allowed us to take that to the next level, the proverbial next level. So I promised myself I would avoid buzzwords in this conversation, but it may not happen. They come up naturally. But yeah, it's been a conscious and concerted effort by our leadership top to bottom to really make that happen. And it's not easy at times, right? You're asking people to step out of what they've done in the past and to trust data and math that may or may not be right, if we're just being candid. And so we've really grown and focused and developed on building that muscle with the organization top to bottom.

2:12:49It's been a really, really interesting and impactful two years with our team thus far. Walk me through some of the concrete ways that you can use data to make a decision at racetrack. I remember there's this funny story. It might be might be apocryphal, but I heard that I always do this where I tell some story that might be entirely hallucinate So the story goes is that one is that McDonald's needed to figure out how to place a bunch of restaurants I'm sure that this is something somewhat related to what you have to do You decide where the restaurants go and they did a ton of analysis and they figured out this street corner was the best And that street corner was the best and they spent millions of dollars in consulting and they put them all there And then Burger King came along and said, yeah, just put one next to McDonald's.

2:13:39And there's some beauty there, there's some hilarity there. But you can imagine that that's the type of very tractable problem, where should I put a thing. Also, store layout, planograms, figuring out what goes on promotion, when pricing, dynamic pricing. There's a whole bunch of things that I could imagine you could do, but walk me through what you did last week. Or even at the individual store level, where it's like, hey, we're out of this product. Yeah, what are the problems? What's the most recent case study you did? Yeah, yeah, great question. Look at you talking about planograms. So, yeah, we like to say that we're always focused on the customer.

2:14:16At the end of the day, it's our customers and it's our associates that make this massive business continue to run and thrive. And so you're hitting on inventory. That's a really important use case. But even more important than that is making sure that we have the right levels of people at our stores to meet that customer demand. And there's nothing worse than when you go up to a gas station to fill up your gas tank and there's a yellow bag on the handle. Or I would actually argue it's even more painful when you put it into your... And then there's no gas. And then it's slow or... Slow or something, yeah.

2:14:48So there's that. And there's also the inside experience, right? We take pride in our food offering. So fresh pizza, sandwiches, breakfast sandwiches. And that takes people. That takes time. And that takes hours. and making sure that we have the right level of people in the store, right number of hours, and the right skill sets as well. It's not just, you can't just throw hours at these problems. You need to understand the skill set to meet that demand and meet those expectations of the customer because at the end of the day, it really is that customer that makes us continue to thrive. And, you know, we've got this pin on, we're celebrating 95 years.

2:15:24We've been here a long time and we expect to be here a lot longer. 95 years ago, software didn't exist. It truly did not exist. And now you're sitting here implementing AI in the largest enterprise software platform possible. Switching gears, a little bit of a hot take. Have you been surprised by the developments in just how the electric car has rolled out? Like there was a moment when everyone was like, do not get in the gas station business at all. It's gonna be all electric, all these companies are cooked. And then we saw the consumer kind of pull back from that and want a different experience.

2:16:03And maybe they have a daily that's a Tesla and it's great. But then they also still are in the gas world in some ways. Have you, has there been optimism inside the company for the future? Well, we are certainly investing in the future. I was going to say people that are charging EVs, they still want to get fresh pizza, right? They do. Exactly. Yeah, and we're actually taking a unique approach where we're developing that infrastructure and those customer venues on our own. So we've chosen to really understand the customer and do it in a way that meets their expectations because we can't predict what the future is going to hold 100%.

2:16:43It's also a different experience right now because you might be stopping for 20 minutes instead of two minutes or five minutes. That's a great point, too. So you have a more captive audience for a longer period of time. Throw an arcade in there. Exactly. throw something else. Come get some racetrack swag in the gas station. Yeah, or anything. Fresh pizza or what have you. But yeah, we're certainly not turning a blind eye to what lays ahead. That's cool. You know, we have certain strategies and things that we're talking about to make sure that we stay ahead. It does feel like it's a unique opportunity now to actually take that seriously.

2:17:14You've seen where this market stabilizes. And there's also just the standardization around NACS now, like the actual charging port is standardizing. So that probably makes the infrastructure cost a lot less or a lot less risky, I guess, for you. Yeah, very exciting. So walk me through the actual scale of the Palantir implementation. Are you early days? Are you trying to roll this out to all the employees? You said 10 ,000, wasn't it? Something like that? Do you want everyone to interface with this or is this more of a managerial tool that would be used to make decisions about how to run the business?

2:17:44Yeah, that's a great question. I think right now we've really focused in on use cases that are driven at the managerial level or the head kind of the store support center level. But that's certainly not to say that there aren't implications at our stores because there certainly are. And I think as we progress and as we deploy more and more use cases, I very easily could see getting the technology in our frontline associates hands as a real value add and frankly, a differentiator. Yeah. Have you had any problems with different enterprise software companies not playing nicely together. You don't have to name names, but we've just been tracking this story that there's now some AI companies that come out and say, hey, we wanna take your Google Docs and get it to talk to your Slack, and Slack is owned by Salesforce, so they don't wanna talk to each other.

2:18:34And I'm wondering in the retail context if a POS system and an inventory management system, there might be some similar sharp elbows, or is it all pretty copacetic? Yeah, I think it's fairly copacetic, but mostly because of our IT team and the really great work that they've done from a data architecture standpoint and consolidating everything centrally and really removing the need for kind of call it peer-to-peer communication of those platforms. Because everything goes into data. Exactly, exactly. And again, I think that that team really deserves a shout out too. So while our team is in the business, the IT and the data team has really been an enabler for us.

2:19:12We have a wealth of information and data that we can make some of these really complex decisions with. and without it we would be severely hamstrung and would be working on challenges like pulling out of POS systems or what have you. And so we've kind of, we're past that level and we have a really strong data lake and infrastructure and architecture to support all of the nerdy math that my team loves to do. Yeah. Awesome. Yeah, what else are you trying to identify going forward? I mean, I imagine that like the base case is just like, I want to know what stores are overperforming, underperforming.

2:19:47But then ideally, you want to be able to predict which stores are going to start underperforming and intervene beforehand. Is that roughly the main way? Yeah, roughly. I think it depends on the use cases. And again, not to throw buzzwords out there again, but we break down analytics into four main types. There is the descriptive, so the old school reporting and dashboarding, Tableau, Power BI. The diagnostic, which explains the descriptive. And then my team really steps in on the predictive and the prescriptive front. So, you know, think about predictive maintenance or, hey, this fuel pump is predicted to go down in the next two or three weeks.

2:20:24That predictive and prescriptive approach allows us to pivot, again, transformationally, away from being reactive to being proactive with things that really impact our customers. So we like to really focus on, hey, where are the customer pain points? How can we peel that onion? How can we solve some of those? so they have a better experience and that that drives a lot of it too so so yeah it's um there's a world of use cases out there and we're really just scratching the surface very cool one last question for me are there bad actors in the gas station business that intentionally pump the gas slow to drive people into the convenience store oh my gosh that flies in the face of everything that we think.

2:21:10That is no front. That is because you want me. So there's, we like to joke a lot about, you know, on my team and maybe others share this sentiment or don't, but is it worse if a pump isn't working or is it actually worse if a pump is slow? It's slow. And I actually think my experience are the most painful when I go up to a pump and it just, it's slowly ticking. At least when you see a bag and you see the yellow handle. Then you know, just don't even try. Don't go there. I just remember maybe it was because when I was a kid and I was broke and I'd put like$20 on pump five and it just felt like it'd go fast.

2:21:45And now as an adult, I just get, but I'm getting like five times the amount of gas. See, you weren't going to racetracks. It'd be very off brand for a racetrack to do anything slowly. Speed is in the DNA of this company for 93 years, 95 years. I can't wait for 100. You'll have to come back on. How was that? 100 years of racetrack data analysis. Break it down. We'll do 100-hour streams straight. Year by year. I mean, it must be fascinating. Name every data point. I mean, just pulling the revenue over a 93-year ramp. That's got to be fascinating. That'd be interesting. Fascinating. Anyway, thank you so much for coming on and interrupting your vacation.

2:22:27Yeah, this is great. I'll talk to you soon. Enjoy the conference. Have a great rest of your day. Enjoy the conference. And that's our last guest for the day, right? That's our last guest for the day. This was fun. started out with a bang we should run out we should run what run through a thank you to all the sponsors that make this possible we told you about ramp calm time is money save both of course powered by restream one live stream 30 plus destinations of course we won't need to tell you about figma think bigger build faster go to figma.com for all your design needs and get compliant on vanta.com risk trust continuously we also got graphite dev supporting us code review for the age of AI.

2:23:09Polymarket, of course. Some big news out of Polymarket. There was a major trade deal. We'll talk about that tomorrow. Julius, what analysis do you want to run? You can chat with your data and get expert level insights in seconds. TurboPuffer, our newest sponsor. Search Every Byte, serverless vector and full text search. Built for first principles on object storage. Profound, if you want to get your brand mentioned in chat, you can do it. the profound. Linear, of course, is a purpose-built tool for planning and building products. Big day for linear. Big day for linear. Getting lots of shout-outs in the Atlassian column.

2:23:46If Atlassian is paying$610 for the browser company, they should get ready to pay$6 trillion for linear. I think so. We are, of course, supported by Numeral, numeralhq.com, sales tax on autopilot. Fin.ai, the number one AI agent for customer service. Adio, customer relationship magic. Adio is the AI native CRM that builds, scales, and grows your company to the next level. We, of course, sleep on eight sleeps. You can go to eightsleep.com. We are also supported by public.com, investing for those that take it seriously. And we always tell you about adquick.com. Adquick advertising made easy and measurable.

2:24:29Adquick forever. And if you noticed, Dr. Karp was wearing a fantastic Patek Philippe Aquanaut with an orange strap. And if you want one for yourself, you can go to bezel, get bezel.com. Your bezel concierge is available now to source you any watch on the planet. Seriously, any watch. And they would love to find you a orange band, an orange band Aquanaut for sure. Business Insider has a scoop here that says Palantir CEO Alex Karp says top tech talent is about to get crazy valuable. Alex Karp, CEO of Palantir, said on quote unquote TV. Again in the quotes. Why do they put us in quotes? This is the dividing line.

2:25:08Why do they put us in quotes? This is the dividing line. Close the laptop. Okay, so Business Insider. The website Business Insider says that top tech talent. I think we just got to put just one of the words in quotes. It can't be quote business. Business. Business insider. Business insider. That is the way we talk. I actually have to look into this company because I love business. And I love insider trading. Insiders in business. Isn't that the lore? Isn't that the lore? Henry Boggett, the guy who started Business Insider. Loved insider trading. I think he lost his license. I'm not kidding. I'm not kidding.

2:25:56Okay, look this up. Business, insider, insider history, history. And more breaking news, Justin Bieber is launching Swag 2 tonight, the new album. What does that mean? And Meek Mill posted two hours ago. Meek Mill becomes a AI founder. So according to Wikipedia, according to Wikipedia, Henry Bloggett was charged with civil securities fraud by the US SEC, settled the charges, paid for$4 million. He was permanently banned, barred from the securities industry by the SEC and NYSE. The charges rose during the dot-com boom, Merrill Lynch, which included issuing materially misleading research reports on internet companies and making exaggerated or unwarranted claims about them to customers.

2:26:47And then in 2007, four years later, he co-founded Business Insider, which is a fantastic pun. It's so funny. It's so funny. He's leaning into it. He was in the business of insider trading, and he said, why don't I combine? They didn't say insider trading. They said civil securities fraud. Okay. It doesn't sound great. But, you know, after a seven-year run, Jeff Bezos purchased a stake in Business Insider. and he had a great run 2007 to 2023. Anyway. There's so many great quotes from the carb segment. This one, I would say, he says, I would say modestly, I'm the most humble I've ever been. You would never build a software company downstream from value creation.

2:27:33It's all, how do I make the client feel like they're getting laid while they're getting F'd? So good. The founder, Adam, who introduced AI Key, a small device that lets AI control your entire phone, just plug it in and ask it to complete a task. He's saying all of this, all of this, and still no TVP and invite. We should probably have him on. A lot of people were said, no thanks, because I guess he previously worked in military intelligence and people didn't feel inclined to plug a hardware device into their phone. But we're in the capital of military intelligence right now. It looks like he sold out the initial batch.

2:28:18Let's have him on. Adam. Put the timeline in turmoil. Anyone who puts the timeline in turmoil is welcome on the show. I'll give him a follow right now, and we will make it happen. A lot of people are having fun with the stream. This is a great reaction. Anyway, that's our show. We got to get out of the United States and back to the United States. We do. Last thing, this just because it is breaking and it's funny. OpenAI plans to launch an AI-powered hiring platform by mid-2026, putting the outfit in close competition with LinkedIn. With LinkedIn? The company also wants to start certifying people for AI fluency.

2:28:59Are you AI fluent? How many Mdashes? Yeah, this seems like more of a Mercor competitor than LinkedIn maybe. I don't know. like, yeah, we need to dig in more to that. But the other odd thing is that wouldn't Microsoft get a copy of whatever they build? So wouldn't Microsoft get access, like if they build a new, I mean, that's the deal, that's the nature of the deal is that they get the rights to open AI's IP. So if they build something that's valuable, but if they build a network, then that's a separate thing, right, because the IP doesn't matter as much. like like the weights to gbt5 are not as valuable as the platform as the chat gbt app so yeah maybe maybe there's something there i don't know people have been complaining about linkedin for a long time so maybe maybe there's breaking news what is this donald boat says that he has art for the ultra dome oh yeah yeah i was i was talking to him about that i'm very excited great he made something so well i wish we could keep streaming but we got to get back to uh we do we gotta go okay let's go all right folks anyway thank you we'll see you tomorrow today we love you back to a regular show tomorrow.

2:30:05Have a great afternoon. Bye.