SPEAKER_01: This is the biggest story of the year.

SPEAKER_00: NVIDIA taking over all of open source.

SPEAKER_03: They've agreed to acquire Hugging Face for $12.9 billion. U.S. open source equals NVIDIA. Chinese open source equals Kimmy. Now we have our U.S. champion.

SPEAKER_04: If you want to know how Anthropic is hitting $50 billion, if you want to know how OpenAI is hitting whatever it is, $30, $40 billion, customers like Uber. Unmetered is the word we're going to be talking about in 2027. That means that you could see NVIDIA add a third of the market cap of the top two frontier models.

SPEAKER_07: One third of their revenue moving over to NVIDIA. If it works, this is the biggest news ever.

David Friedberg: All right, everybody, welcome back to Twist. It's Friday, August 28th, summer's over. Lots on the docket. Sam Waltman talking about cyber threats again,

SPEAKER_12: scaring the bejesus out of everybody, saying, hey, we're going to need to have a coordinated effort here.

SPEAKER_14: NVIDIA bought Hugging Face and we got some great guests. We want to start with a story or go right to a guest? What's the lady's choice, Lon? What do you want to do here?

SPEAKER_16: Oh, oh, misgendering me again. How dare you? No, I think we should start with this. I think we've got a top story.

SPEAKER_18: AI's Oppenheimer moment.

SPEAKER_19: It was the beard that made me misgender you. That's right. Bearded lady. It's my old circus career.

SPEAKER_00: We might be in AI's Oppenheimer moment, Jason. OpenAI, Anthropix, some of the other companies have issued a warning about coming cyber threats. Yesterday, over 100 companies signed an open letter warning that we only have months to prepare for AI-enabled cyber attacks. It includes OpenAI. It includes Anthropik, Google, Microsoft, all the usual suspects. They're warning hospitals, water systems, critical infrastructure, all under threat. They want security standards raised, including shared threat intel, and they want this defense funded at every level. All that said, the letter doesn't have any actual commitments. There's no deadlines. There's no mandated funding. It's just sort of floating, floating the suggestion. So, you know, I mean, in your perspective, what do you think the best way for us to actually, I think everybody would agree we should do this.

SPEAKER_16: What do you think the best way to organize this effort might be? Okay, we did an open letter.

SPEAKER_14: This is an open letter.

SPEAKER_15: Correct.

SPEAKER_14: Written by people like OpenAI, Anthropik, Google, Microsoft, et cetera. They passed around and said, hey, we're entering dangerous territory, and we need a coordinated effort.

SPEAKER_00: Essentially, yeah, 100 companies signed on to this open letter.

SPEAKER_02: The hyperscaler's obviously sort of leading the pack.

David Friedberg: Why now is always the question. Why now? And you have to deconstruct people's behavior.

SPEAKER_07: So let's pause for a second and say, why would the top 100 folks take this time now to write a coordinated letter? And the letter is addressed to who?

SPEAKER_31: Who is the letter addressed to? And let's pull the letter up here on the screen. These are important things when an open letter happens.

SPEAKER_00: It's called a call for collective action on cyber defense. It's not directed at any one person, but I think it's the public. It's PR on one level. And it's aimed at sort of politicians.

SPEAKER_07: Well, it exists on the OpenAI website as we're showing it. It says an open letter for a global surge in cyber defense. So it's very important to first principles anytime there's an open letter or somebody makes a statement to slow down and say, why now? First question, why is this happening now? And then who's driving this? So who's driving this is pretty obvious. There's a bunch of signatories on the left. We assume that they're driving it. It was released on all of their websites

SPEAKER_12: or it was released on OpenAI's websites.

SPEAKER_35: OpenAI is the leader of this pack. Is it like every single person releases at the same time? Or do we not know?

SPEAKER_36: I don't know. I'm looking it up. I mean, when you Google this title,

SPEAKER_00: it does come up on several different sites. So I think OpenAI, again, leading the charge, that's the number one link when you look it up. But I think other companies have also posted it. Well, that could be just SEO.

SPEAKER_12: So every company is, so the translation here from first principles, and we're trying to get to first principles always when we have these kind of announcements, because you want to be able to media deconstruct. I'm going to help your media deconstructor. Every company is releasing this. The top SEO, based on what you're telling me, went to OpenAI. Okay, fine. And I saw Sam quoted it. So then if Microsoft did this, did Satya Nadella write about this? And then did Dario or somebody from Anthropic, like a leader, actually tweet about it? So those are the things you want to know. Because we want to understand who's really behind this, or if this is when you have 100 people or 50 people write these things, that's the other thing you want to find out. And it's like, I get asked to sign these things sometimes. So let's take a look at Dario's account. Is he talking about it? Let's take a look at Satya Nadella's. Is he talking about it? And let's pull up Sam's and how Sam is framing this. And then we have to ask ourselves, why is this happening now? So we'll pull up those three individuals' Twitter. We'll see if they're even talking about it.

SPEAKER_40: I know Sam is talking about it.

SPEAKER_15: Satya has not tweeted since August 21st. He's not talking about this today. Microsoft signed it.

SPEAKER_14: He's not talking about it. Now we go to Dario and maybe the Anthropic official account. Let's check that. And this is the way you do proper media deconstruction, is you go figure out who's talking about this and who talked about it when. So here, August 27th, yesterday, Sam Waltman writes, this is a critically important moment for cyber defense with AI.

SPEAKER_07: There is not much time to act. Okay, this is like a very charged statement. We are happy if you want to work with us or any of our competitors or partners, but please take this moment seriously. Again, he's being very dramatic here. And we'll read into that in a moment. Only an urgent and intense collective response will work. So this is critical.

SPEAKER_14: He is coming out very hard about this. Did Dario talk about it?

SPEAKER_00: No, I'm looking at Dario's page right now. He actually hasn't tweeted since the Gavin Baker kerfuffle of last week.

SPEAKER_14: And then here's OpenAI's tweet from August 26th. We have conducted a thorough investigation into the Hugging Face incident. That's when their AI bot went and tried to steal credentials on Hugging Face's site and kind of hacked their site to accomplish goals. We are releasing a technical report and accompanying blog posts that reconstructs the agent's activities. That's OpenAI's agent that hacked Hugging Face.

SPEAKER_07: Explain why existing safeguards failed and detail how we're preventing reoccurrence. So here we go. Sam is leaning into this. OpenAI has egg on their face

SPEAKER_48: or they have liability potentially from having hacked Hugging Face's site. If Hugging Face's site was hacked, then you have to ask yourself, is OpenAI's, are OpenAI's other agents during the building of models hacking other people's sites? It is happening here. I think this is coming out today. Always when these letters come out, always when people make statements, there's usually an incident that people know about, which is the Hugging Face one. There's also sometimes other incidents to come and they are doing law and what we call getting ahead of it. Sure. So somebody says, let's get ahead of this. I think what's happened is there's probably two or three hacks that have occurred that are acute. Acute is a fancy word for important intents. And we might have a hack that's occurred already that the government is aware of, OpenAI, Anthropic, Microsoft, et cetera, aware of that could be of the nature of major hacks we talked about previously on the program, the Ashley Madison hack.

SPEAKER_12: Sure. There was an iCloud hack. There might be one of those that's going to drop in the next couple of days. It's Friday. This is coming out Thursday, Friday. Over the weekend, just predicting here, you might hear about a significant hack

SPEAKER_14: of our government or of your sensitive information. They are prepping us the same way last week I told you, or maybe earlier this week, and it went viral, that the Chinese are doing a psyops

SPEAKER_12: on America and their own country to make their people accept robotic overlords that will be taking their jobs,

SPEAKER_14: that will be integrated into society, that will be observing and recording their citizens to control them. That's what they do in China.

SPEAKER_12: They are prepping them to accept robotic companions, robotic workers, robotics, you know, in their lives. That's their PR campaign right there. We are doing some other campaign right now, which is, hey, your privacy is over.

SPEAKER_48: You're going to get hacked. It's not our fault. And maybe trying to spread the blame around here. So there are psyops happening everywhere, folks. That's what I think this is. This is a prep for a coming hack that they know about

SPEAKER_02: or that they know is imminent. Yeah, we had Uga D'Amour from Expo on This Week in AI this week,

SPEAKER_00: and he said it's been a year. He's like, for a year, the West has been totally vulnerable to this. We're getting hacked all the time. People might not realize it. It's not life and death for most people, but we're already in the danger zone,

SPEAKER_28: and we have been for some time. That was a... I didn't realize that, I think, as clearly as he stated it.

SPEAKER_54: The truth is, most of the big, popular AI models can handle your workloads just fine. And your performance and your cost, that comes down to your infrastructure more than your model selection. And the best way to keep your system running fast while staying affordable is with DigitalOcean's Inference Engine. DigitalOcean built their entire cloud platform from the ground up, specifically with AI in mind. That means you get a serious performance upgrade without stitching together services from a dozen different vendors. Or getting hit with a surprise bill at the end of the month, which keeps happening to me. DigitalOcean's system is already powering some of the biggest and most popular AI applications. Character AI runs over a billion queries every single day on DigitalOcean, and they cut their cost per token in half while doing so. So, here's your call to action. If you want to understand what building on a true AI-native platform looks like, go to do.co.twist. Start building on the DigitalOcean AI-native cloud today, and cut your AI workload costs by up to 50%. That's do.co.twist.

SPEAKER_07: What is at the core of this is that if you build a great model for writing code

SPEAKER_12: and you are incredible at incentivizing AI to do recursive, I'm sorry, doing reinforcement learning or creating loops, you are then, by definition,

SPEAKER_07: allowing for or enabling hacking at a scale that is essentially infinite and unconstrained. So, let's just say this clearly. The frontier models

SPEAKER_14: have enabled inadvertently by making incredible tools to vibe, code, whatever you want.

SPEAKER_07: The counter to that is making a tool that can infinitely, depending on how much compute you have access to, is infinitely better at hacking than the White Hats and is relentless and has an unlimited amount of time and effort and intelligence

SPEAKER_48: to pair hacks together that humans simply do not have the time. There's no one human who can put all of GitHub, every open source code, into their brain at once and then say, here is the history of every hack done since 1970. Here's all code that's being used, sorted by the last update to its code and how often it's updated and how many people are using it, the stars, the impact of it, and then it can put those things together and say, what hasn't been updated? What's not being vetted and cleaned up and tightened up? Which projects, which software is not tight, is not updated regularly, and what are the latest hacks? What's the history of hacks? And it can just figure out brilliantly in the magic AI box, it can, just like it can guess your next word, it can guess the next hack. Yeah. It can guess the next big hack. Now,

SPEAKER_16: I wanted to talk about this, Jason. Do you remember this from earlier this week? It's also, it's not just the availability

SPEAKER_00: of this intelligence, it's getting cheaper. So hackers are budget constrained. They can't just run Fable 5 all day and do all the things you're saying, but they can now afford to use DeepSeq. And as those models get not just cheaper, but much more powerful, we're going to see this just continue to explode.

SPEAKER_16: Or people can just run exactly what you're saying overnight for weeks at a time and find every vulnerability there is.

SPEAKER_68: And so this is going to be a boom for Palo Alto networks. It's going to be incredible.

SPEAKER_70: For CrowdStrike, those are the two leaders. I might put a J-Trade on this weekend for those two stocks. I'm going to have to do an analysis of them.

SPEAKER_14: I don't have time to do it right now on the air. We've got a big docket. But I think that those firms are now going to be fully, fully engaged with

SPEAKER_70: by the frontier models who are going to go to Palo Alto networks, they're going to go to CrowdStrike and others and say,

SPEAKER_14: let's build. As you build frontier models, we're going to need access to them early. We're going to need to build the sure it up tools, you know, like check the locks. Yeah. And somebody is going to just have to go check the locks, check the alarm system, check the perimeter of every piece of software. And I think we're going to see a hack

SPEAKER_07: before the end of the year. I think it's possible we'll see a hack that will make the public distrust AI even more.

SPEAKER_12: A cataclysmic hack of people's banking information or crypto, wallets, or personal, you know, iMessages, Signal, WhatsApp. one of those gets hacked and AI dumps it. People start getting extorted. The extortion by Russian, North Korean hackers, et cetera, Chinese hackers, or some big dump, or the stealing of crypto, this is going to make people hate AI much more than they hate data centers or they hate writing AI slop and passing it off as your words in the Wall Street Journal editorial page. Shout out Wall Street Journal editorial page for phoning it in. Those things are going to be minor compared with losing your Bitcoin or having your iMessages dumped. It's going to be really an interesting moment in time and I think that's what the industry is trying to get ahead of. They're warming us up to the fact that we've already

SPEAKER_75: probably been hacked by AI.

SPEAKER_00: And they are. I think you're absolutely correct. It is a little blame shifty. It's like, hey man, this is all of our problem. It's like, is it all of our problem or is this your problem?

SPEAKER_75: I think what you'll see is the majority of the hacks will be done by the frontier models.

SPEAKER_12: So the frontier models then who create these, it's going to be essentially like producing a car without three-point seatbelts,

SPEAKER_79: producing a car without airbags,

SPEAKER_80: you know, selling cigarettes without, and saying nicotine is not an addictive drug.

SPEAKER_14: Have you ever seen the congressional hearing? I'll have one of the team where they go down the line and ask everybody working at Big Tadako.

SPEAKER_15: Who thinks, yeah, nicotine is addictive. Right, yeah. And the audacity of them

SPEAKER_14: is while they were architecting cigarettes to be more addictive through any number of chemical sequences and, you know, product innovations to have the audacity to say, not addictive. This is going to be, I think, what happens to the frontier models, which is you made something

SPEAKER_07: that you knew could hack everything and you released it knowing that, oh, congratulations, you wrote a letter, but you gave people a car, you left out the airbags and the three-point seatbelts and you made it go over 150 miles an hour and you made it, you know, watch this to just understand this is what we're going to see in about 18 months when the world gets hacked that people lose their Bitcoin.

SPEAKER_87: Unsafe at any latency, we'll call it.

SPEAKER_88: Begin my questioning on the matter of whether or not nicotine is addictive.

SPEAKER_89: Let me ask you first, and I'd like to just go down the row, whether each of you believes that nicotine is not addictive. I heard virtually

SPEAKER_90: all of you touch on it and just yes or no, do you believe nicotine is not addictive?

SPEAKER_91: I believe nicotine is not addictive, yes. Mr. Johnston.

SPEAKER_92: Congressman, cigarettes and nicotine clearly do not meet the classic definitions of addiction. There is no toxicity. We'll take that

SPEAKER_94: as a no and...

SPEAKER_93: Wow.

SPEAKER_97: Wow.

SPEAKER_81: Unbelievable. Shout out to Michael Mann's classic 1999 film The Insider, which is about behind the scenes. If you haven't seen it, please watch it. Incredible movie about this. I'm just giving you a highlight.

SPEAKER_100: This is literally what's going to happen

SPEAKER_12: in under 18 months. Clip it now. Bookmark it. J. Cal was right about everything. Here's your clip, folks. Yeah, the clip. Get on this. Sam and Dario are going to be on a stand there and they're going to say,

SPEAKER_14: did you release this? And then everybody's cryptocurrency wallet got hacked, everybody's signal, whatever, cryptid thing got hacked. That might not be that they, you know, hack encryption. That will probably be quantum computing's legacy, but it could be that somebody programs a way to build a backdoor into everybody's phones, get them to download some, you know, exploit as the next, you know, phishing attack, let's say. And an at-scale phishing attack that was just so good that 2% of society falls for it and then it expands and 10% of society falls for it and now you got 10% of people, you know, having their Bitcoin sucked out of their wallet. This is like a completely possible thing to do amongst many things

SPEAKER_07: and that's just a human engineering device, right? There could be other things and now everybody can create those things.

SPEAKER_48: It used to be if you wanted to learn how to do a phishing attack, you had to get into some dark web stuff, you had to go trade some wares.

SPEAKER_77: Well, you're spoofing somebody else's domain. There are technical elements to it.

SPEAKER_48: To get those, you had to get into a community with other hackers, you had to have somebody send you the code. It wasn't like freely available inside a large language model. Right. And when you take the rails off of these and you have an open source one, oh my lord.

SPEAKER_15: This is going to get spicy. There's your media deconstruction for the day. We should bring in our first guest, I think. Let's get started.

SPEAKER_00: Let's do it. So we've got Craig Piggott. He's the founder and CEO of Halter, Jason. They're a livestock management technology company centered around virtual fencing systems. Craig, thanks for being here.

SPEAKER_111: My pleasure. Thanks for having me. Yeah, so tell us or show us what you're building

SPEAKER_112: cool.

SPEAKER_113: We built a tech platform for ranchers, ranchers and farmers. At the center of that is a collar. Collar goes on a cow or goes on cattle and yeah, we use that collar to train animals to respond to cues and what that means for ranchers is that they're able to virtually fence and shift those cattles around their land. That means the land's more productive. They can run more cattle. We do a bunch of other stuff as well. We monitor health,

SPEAKER_114: fertility, calving, we measure grass. It's like an operating system.

SPEAKER_70: And here it is. If you're watching, you can see the collar there. It looks like a Fitbit with solar panels for cattle. This goes over their neck like a harness of some type like I have on my wrist here from my whoop. Solar powered GPS.

SPEAKER_07: I'm assuming it has accelerometers, heat sensors, heart rate. What else is in the sensor pack here? And how much does it cost to build something like that? Like what does it cost

SPEAKER_14: per cow to make one of those and then how do you charge for them? I'm very curious about the business model but let's start with what sensors did you choose to put in there for cattle ranchers?

SPEAKER_54: Starting a company is hard enough. You need to focus on running your business and building a great product and you need to clear all that back office work, all those chores off of your plate and that's why Everi exists. E-V-E-R-Y is the all-in-one platform for incorporation, banking, payroll, benefits, taxes, basically your entire back office and Everi has worked with over 1,000 startups. If you're just getting started they'll handle your Delaware C Corp, EIN and getting yourself a registered agent all with no legal fees and no delays and if you're further along in your founder journey and need to scale up your team Everi will help you hire contractors in over 200 countries plus you're going to get 3% cash back on every dollar when you use your Everi corporate card. So head over to Everi.io

SPEAKER_120: and stop wasting time on stuff that isn't growing your company. That's E-V-E-R-Y dot I-O Everi dot I-O.

SPEAKER_122: Yes. The core sensor that powers

SPEAKER_123: a lot of the analytics is the IMU so accelerometer gyro, magnetometer.

SPEAKER_113: We do track temperature and humidity and other things but you can infer most health related things and fertility and calving from behaviors which is what you we run models to kind of understand what's happening from the IMU. So that's kind of the core sensor in terms of the model it's a subscription so you pay per cow per month these multiple tiers and packages it kind of depends on your operation and you don't buy the hardware we give the hardware away for free you just pay the subscription so we've got about one and a half

SPEAKER_123: million collars out there across New Zealand Australia US, Canada etc.

SPEAKER_14: Awesome. Now there is an audio queue as the fence explain how you train cattle to I don't know go to the left go to the right don't go forward

SPEAKER_129: how does that work you're not shocking them you're giving them a vibration explain what's going on here.

SPEAKER_123: The best analogy is when you're parking your car you have the audio feedback from the sensors as you like approach walls and other cars so it beeps at you very similar concept left and right so animals know if they're approaching a virtual fence based on that audio queue we also use a vibration OQ we do train them with a

SPEAKER_113: very low energy pulse which gives that sound meaning most cows or cattle today are trained to an electric fence and the primary queue is vision and we just turn that into audio so once they're trained runs on sound and vibration what does it

SPEAKER_131: cost per cow

SPEAKER_132: per month what's my per seat per stake what's my per stake cost here

SPEAKER_113: the like base starting tier is around nine dollars per cow a month but it depends on all the packages so you can

SPEAKER_07: spend 200 bucks a year per cow they're paying 200 dollars a year

SPEAKER_134: how many years does the cow wear the collar for before it is processed

SPEAKER_123: so the collar imagine it kind of like a seat you can scan it

SPEAKER_113: with your phone and tag it on or change any animal that it's on so the collar will sometimes go on many cows but most cattle in terms of the cows will be say 10 years if you're a dairy cow sometimes it's longer than that if you're a beef cow it's often we collar them the mums like the calves that you raise often for like the end product you want

SPEAKER_114: them to kind of creep grows ahead and so we don't collar them so we collar the cows most

SPEAKER_100: so what is the number

SPEAKER_14: one reason people use this what is the

SPEAKER_139: selling point when you come in they say hey this feature is why we're buying this because obviously it's got a range of features but what is the number one to keep the cattle in a

SPEAKER_113: that are running twice as many cattle on the same piece of land

SPEAKER_146: because

SPEAKER_113: they're able to actually control it not waste any grass things like that so that's the big value prop but every customer

SPEAKER_114: is obviously different some of them are chasing genetics and metrics and things like that so it kind of changes

SPEAKER_148: okay so I can more

SPEAKER_14: so talk to me just and I'm sure Alon's got a couple of questions here but how does this work you know when we're out in ranch land here

SPEAKER_70: we don't have internet nor do we have power so obviously it's got solar on it but nor do we have like 5G so you know

SPEAKER_14: if you're in ranch country

SPEAKER_70: there's

SPEAKER_14: no way to get these are you hooked up to Starlink or do people have to put

SPEAKER_150: repeaters on their property how does this actually reach and deal with that connectivity issue

SPEAKER_133: exactly in terms of the Starlink the Starlink point so it used to

SPEAKER_123: be base station or tower based and that was right on some of the less mountainous country but like you say through Montana Wyoming Colorado mountains the more off-grid you get the more you need that direct connection with Starlink so each colonel can talk to Starlink and they also talk to each

SPEAKER_114: other so they kind of coordinate a mesh they best leader to then talk to Starlink that kind of saves power and cost and things like that

SPEAKER_151: fascinating and now it's obviously doesn't have a dish or you didn't put

SPEAKER_14: a Starlink mini on the back of the leader of the back here but how does a device that is the size of it looks like a bread box or loaf of bread how does a device the size

SPEAKER_12: of a loaf of bread actually hit the Starlink what is the technology there because it's much bigger than a smartphone but it looks like

SPEAKER_150: it's much smaller than

SPEAKER_113: your phone can talk to Starlink some of them can depending on who you work with it's the same technology it's the direct to sell technology what's

SPEAKER_158: the speed of the direct to sell now how much

SPEAKER_14: throughput do you get obviously you can store the information and when it gets a

SPEAKER_113: for sure it kind of depends when we coordinate all the information between them and compress it and send it and it takes a few seconds every minute to transfer that data we do run a lot of machine learning models on the device itself so when we need like 100 hertz access to certain sensors we actually run those models on the

SPEAKER_114: collar we train them offline and then we put them onto the collar that gives it you don't have a bandwidth problem so

SPEAKER_16: I'm curious I mean in the pre-interview I was calling this collar

SPEAKER_00: essentially like whoop for cows because you're you're monitoring so much about each animal aside from just where it is can you give us a little insight what are the most useful pieces of data a farmer needs like

SPEAKER_28: when they

SPEAKER_113: that's just the monitoring side compared to say the fencing and shifting but on that side obviously they care about animal health they care about are they fertile are they is this cow and calf is she calving is she

SPEAKER_123: in trouble do I need to go help a lot of these animals are sometimes hours away and they get on a horseback and they spend four hours trying to find them and so being able to know wherever you are hey this animal is healthy and you know behaving normally is a really big unlock we're working on some other cool stuff like measuring intake

SPEAKER_114: through the collar by counting bites and profiling bites and then you can rank cows on efficiency and things like that so

SPEAKER_14: you could

SPEAKER_07: literally know their jaw speed how much they're eating cattle we on my ranch

SPEAKER_12: here my neighbors got cattle he rents his ranch out for cattle and had a fence go down we had two cattle show up lawn wife says to me those are ours now right let me explain to you one thing about Texas and how cattle

SPEAKER_07: works stealing cattle means they can murder you that's like the one thing you don't do they're literally branded and they got an ear tag for a reason

SPEAKER_173: people take cattle rustling

SPEAKER_62: that's what John Wayne would have called it that's like

SPEAKER_175: the number one no no we don't get to

SPEAKER_12: keep the cows we do get to walk them over the fence and then we get to rebuild the fence with our neighbor and that's how ranches work you rebuild the fence together that's like when you and your neighbor get together but what about cameras video etc like is there a case for that and then because you could actually see what they're eating you could process that and then you could then look at the output hey what's the marbling on this cowl I'm just thinking from a selfish perspective on my wagyu but that is the

SPEAKER_123: weather stations and data off the

SPEAKER_113: collar as well we have a model that tells ranchers or farmers that information so we don't have cameras on the device obviously your cattle aren't they're always in like one place on your ranch so if you want to know hey should I shift them over here and they're not there then you wouldn't have that information so we get asked about it a lot though you also in the halter app you can take photos of your land and it'll tell you everything you need to know so yeah no cameras on the collars yet at least

SPEAKER_07: founders we've got a huge offer for you today from my firm launch and our friends at PayPal eligible startups will

SPEAKER_39: that we use it we love it but there is so much more than just that yellow button they've built a global payment infrastructure that's designed to scale along with your company and if you're an AI and SaaS startup PayPal is already set up for the way you monetize that means seats subscriptions credits token usage API consumption and your entire revenue stack so if you're just starting to build or you're already scaling up pay zero processing fees on your first hundred thousand

SPEAKER_35: dollars in eligible PayPal payment volume to find out more go to paypal.launch.com

SPEAKER_148: and how is the startup doing I realize you've raised a decent amount of money you got over a million harnesses out there a million callers out

SPEAKER_161: we launched the US market about two years ago we're about to get down to South America we're hiring in all the markets at the moment how many cattle

SPEAKER_70: are there in the world and if you have a million collars what percentage of the TAM do you

SPEAKER_04: have

SPEAKER_123: there is about 1.5 billion cattle in the world so it's a very depressingly small percentage of penetration but that's actually called

SPEAKER_188: an opportunity in my business the

SPEAKER_190: founder looks at it as long oh my god we've done so little the investor looks at it and says

SPEAKER_14: yum yum not for the stake but for the market opportunity so there's 1.5 billion cattle you have 1.5 1% would be 15 so you're at 10 basis points of

SPEAKER_79: cattle in harnesses to date you literally can go 1,000 X from here

SPEAKER_187: totally and obviously

SPEAKER_113: not all those cattle are going to be in the sweet spot for tan but you get an idea for the size of the market and I think we view

SPEAKER_123: ourselves more as also a tech platform for this industry like we a lot of the product now that our customers buy is almost separate to the satellite product that I talked about or we are heads down trying to scale yeah

SPEAKER_197: it sounds like scaling

SPEAKER_129: this is the easiest thing in the world if my competitor if a competing ranch

SPEAKER_07: has this product and you go and sell it are you doing founder led sales still or just the sales motion now

SPEAKER_04: tell me about your sales motion still founder led sales you go meet the ranchers or do you have a sales team yet and how many people in the sales team and then how did you instruct

SPEAKER_139: them like this is this week in startups explain the sales motion here so other founders can learn please

SPEAKER_123: totally the sales motion is

SPEAKER_07: and you also learn critically important that this is not an enterprise sale where you dm somebody like

SPEAKER_14: a login this is salt of the earth ranchers farmers vets you have to build relationships they have to trust you and

SPEAKER_79: it's not going to be you know two 20 minute zoom calls and you got a customer

SPEAKER_113: yes we it took us a while to find this model like we we tried lots of different things like how do you sell to

SPEAKER_123: you know ranchers and farmers and I think toast became our inspiration like they run an outbound hyper local there's a lot of parallels right like people running restaurants aren't thinking about point of sale all the time people running ranchers aren't thinking about collars like you

SPEAKER_122: so we probably have 200 or 250 reps

SPEAKER_14: this is critically important one 600 people at the company a third of them are doing the go to market so once you

SPEAKER_07: get product market fit which probably took you

SPEAKER_123: even we started the business and so we spent I'd say a good part of

SPEAKER_113: five years wow deeply technical working out how to keep the collars alive in winter how to talk to them how to build the hardware how

SPEAKER_214: yes

SPEAKER_213: tell us about that

SPEAKER_14: what did your parents think when you told them the idea where they because a lot of times when you have an innovative idea the market resists and this is the standard issue lawn founders have they have a vision but the closer you are to the problem the more resistance because it's a paradigm shift and they say this is never gonna work we know better but now initial reaction your parents had to your vision

SPEAKER_122: they've always been supportive but they definitely didn't believe well you know they were skeptical of

SPEAKER_113: why and you know so I think it took a while for them to be like oh this actually just looks like it's working and so a lot of my parents like they've helped us so much by just

SPEAKER_123: giving us free reign of their you know their land and their cattle we've been buying and selling trying to work out how to train different

SPEAKER_113: breeds and things like that so but yeah initially they would have been like Craig's just doing a fun little thing you know good on them this is critical

SPEAKER_139: I think just as we pause here for a moment every

SPEAKER_07: founder and also being successful in life you will face resistance nothing to do as you said your parents are great people industry filled with great people I know a lot of them now that I have more time you've spent in a vertical the more codified your belief system becomes

SPEAKER_14: and this is why outsiders typically do a better job reinventing the hotel industry or say the livery industry Airbnb and obviously Uber or stock market and

SPEAKER_07: how to buy and sell stocks Robinhood you will face resistance even from good people because they have codified their

SPEAKER_48: belief system and they have first hand experience of it working but then some

SPEAKER_14: new technology comes out they're not aware of that

SPEAKER_123: but I don't know I think you just try to build the best product right if you build a product that's so good they can't ignore it it's fun to

SPEAKER_14: so you had to lead the industry to adopt this technology great you got some first hand experience at rocket lab that's a rocket

SPEAKER_12: company out in New Zealand yeah with the very famous founder by the way what's his

SPEAKER_139: name sir Peter Beck yeah he's a sir Peter is Peter a sir

SPEAKER_14: in New Zealand or a sir in the UK

SPEAKER_224: we're part of the same sir people in New Zealand

SPEAKER_189: can become sirs same with Canada we haven't had our independence moment

SPEAKER_226: incredible

SPEAKER_227: that you can get that that's very nice

SPEAKER_07: so but then something happens in the next phase of a startup once you have some adoption and you've hit

SPEAKER_230: resistance to

SPEAKER_07: being all in and then their product suggestions happen because they're in that app so show us the app and then tell us what two or three things have your customers demanded of your product that you didn't think of or that they you didn't think maybe that was the top of the priority list so let's take a look at

SPEAKER_123: because they live in this product they use it every single day of the year they log in 50 times a day it is how they run their operations so they care so much about like hey this isn't right I want this feature can you add this to it and so our philosophy at halter is to try and have the best engineering team and the best scientists and designers and then listen to our customers and let them drive a lot of the feedback so the features I talked about before where we are like measuring grass and telling them how it's growing and how to graze it that was you know lots of our customers were asking us for the data hey she's clearly sick she's grazing less she's walking less how come you aren't telling us this build an alert to tell us that she's sick and so some of this we we have our own you know we have our own theories and run our own tests but a lot of it has been market led and so then they feel like the part of the journey they ask for something we ship it we can update the devices when they're on like we do that every couple of weeks obviously you can update an app so the whole platform is evolving all the

SPEAKER_14: awesome and you know it's very interesting when I got the whoop and I started playing with it I've played with all the platforms they're all great and have

SPEAKER_04: different pros and medicines but I got two alerts in the last like two years of wearing this and both times it was skin temperature related now you

SPEAKER_07: and now I'm like wait a second if I know it knows I'm in the early stages of a cold if everybody had whoop during the pandemic you would have been able to tell people 48 hours earlier that they had COVID you wouldn't need a COVID test I don't think you would just say you have these conditions your skin temperature is up your respiration is down that is indicative of

SPEAKER_12: people would take verse 24 that

SPEAKER_48: to me is you you might save cattle that are on the brink of getting a disease and then that's super power because each of those what is a cattle worth

SPEAKER_236: you know like three and a half grand at the moment wow so you

SPEAKER_12: save one or two per year that's the same as like 20 of these collars like you pay for itself yeah

SPEAKER_218: a lot of these

SPEAKER_123: ranches are running on public lands like forest service land and they'll turn out 500 cattle and they go to gather them and they spend three weeks up there with 10 cowboys and they

SPEAKER_172: even line has a spicy question okay Craig finish your thought and then we'll go to spicy question go ahead you go long

SPEAKER_16: oh it's not very spicy it is sort of a break off from our

SPEAKER_28: blue sky is falling we'll do reddit is falling anytime your company

SPEAKER_00: comes up on reddit I notice that there's a theme in the comments which is people are very worried you're going to put them in the collars this is to be used on humans cattle can be fenced in here's wild wild west where you can see them in the elaborate collars is he planning to use this on humans so I'm giving you your opportunity right now to assure us

SPEAKER_176: there's

SPEAKER_28: no plans to put this on

SPEAKER_252: this isn't this is on your desk when you get

SPEAKER_123: asked almost daily about like when you doing goats when you doing sheep when you doing horses and we say there's so many cattle in the world like we we're barely getting started on cattle we're doing nothing but cattle for a crazy conspiracies floating around it

SPEAKER_253: is conspiracy corner it is for sure anytime

SPEAKER_254: in the Peter Thiel universe the fieldverse just

SPEAKER_139: brings the

SPEAKER_256: conspiracy it's

SPEAKER_139: his name alone yeah I think what you're doing is amazing because number one I like premium meat

SPEAKER_14: but number two being able to be more efficient with the food supply and to reduce disease and to be more efficient in the land not to be a hippie here but it's important conservation and when you get a ranch you really start thinking about conservation and how to work the land properly I don't have cattle but I probably will rent to some cattle ranchers portion of my ranch and I think it's

SPEAKER_150: very important that we apply some science here I applaud it but you

SPEAKER_123: every morning when your customers love the product it's in the real world it makes a big difference whenever we're hiring we say to engineers come work on something that's actually in the real world go meet some of our

SPEAKER_70: founders

SPEAKER_14: have founders sometimes

SPEAKER_264: are ADHD ADD I

SPEAKER_144: don't necessarily believe in these diseases I think it's all a little bit of BS

SPEAKER_14: but putting that aside the condition of being excited about new ideas is something that founders do because they're like oh I started this cattle company and oh yeah I could put this on

SPEAKER_07: horses and what's critically important for founders is they keep drilling and they keep drilling down in the same spot and then they go adjacent these are the two strategies and I just want to make it very clear because I don't know if I have a clip of this to send to the 100 founders we

SPEAKER_48: adjacency of going to horses means you're starting over people who have horses are completely different they don't slaughter the horses they're not food the horses are pets and they have incredibly high values but people have a very small number of them people don't have a thousand horses on their ranch they have three completely different markets so if you're going to go adjacent make sure

SPEAKER_07: you've drilled straight down first that's what do uber for boats

SPEAKER_48: you should do uber for planes you will start an airline start this start that and he was like you know what i'm going to start vegas london australia we're going to take china same product keep drilling straight down when you drill straight down and like i'm going to build a farm or i'm going to build a hotel next to my oil rig still has to be the same concept drilling great job craig does that resonate with you craig do you have people in the company now trying to distract you to do the vertical

SPEAKER_218: yeah i think focus is

SPEAKER_123: there's 10 million cattle in new zealand maybe we're on i don't know five percent of them like long way to run so even as we do geographic

SPEAKER_260: expansion within the same product we try to be very focused and you got to write that

SPEAKER_276: playbook that was the other thing travis did really well he had josh doing the playbook in new york he had william

SPEAKER_07: do the playbook in california then they compare playbooks and it was like wait these things are the same recruiting drivers are the same oh okay but recruiting the the door dashers and the uber eat folks completely different in new york because in new york they're on bikes and in california they're in cars and none of them are on bikes two different groups of people two completely different strategies okay now you're going to singapore okay how do these lessons apply to another geography it's going to be different it's always going to be different like and when you come to the u.s

SPEAKER_14: they're going to be much bigger i suspect ranches and there are

SPEAKER_113: some of them are probably bigger than half

SPEAKER_123: of new zealand is a country so totally like the scale is huge

SPEAKER_70: all right listen continued success sorry i missed this investment when you do your next round maybe save a slice

SPEAKER_12: for jcal and the syndicate dot com all right craig great job if you ever make it to austin i'll take you for west home west home all right is the

SPEAKER_35: greatest new york trip out of australia west home that's my favorite all right

SPEAKER_285: i'm in texas quite a lot as you can probably guess every

SPEAKER_12: thursday i have a little supper club we do have a little steak and we'll invite you

SPEAKER_00: once once every protester in hong kong has one imagine the tam on that i'm kidding they're never gonna

SPEAKER_79: to be honest you don't need to put a call or everybody has a smartphone and they've opted in

SPEAKER_14: just been thinking like i gotta get this set up when i travel international like i'm a target you know for information about the startups we invest in like there's literally corporate espionage that occurs in our industry in a way we don't really talk about it

SPEAKER_12: on sunday which i think is live at 12 30 texas time which would be 10 30 california 1 30 so if you're it's very sad for me and a lot of folks he was on this week in tech

SPEAKER_75: he was one of the anchors there in the early days with Leo Laporte which was the big inspiration literally the inspiration for this week in startups

SPEAKER_12: was both and I just like to give credit to my big brothers in this industry I'm 55 Adams like columns I would eat them up I would earmark them I would underline them and say someday I would love to get paid to write about computers I literally had this dream when I was 13 14 15 and I got my PC junior man someday I could be like John Dvorak then they all did CNET and all these incredible things I did Silicon Alley Reporter Cyber Surfer my magazines my dream opinion that was spicy like Howard Cern like him that people would pay attention to and here we are you know X number of years later but to give Adam Curry his due and Dave Weiner Dave Weiner created RSS I started getting into blogging because my

SPEAKER_14: employees had left Silicon Alley Reporter Rafat Ali created paid content and then Skift I'm an investor

SPEAKER_04: at Shani Jardin went to Boing Boing and I was watching this RSS movement you could have this open standard of RSS really simple syndication and it was like wait you

SPEAKER_39: don't have to have people come to your website they can subscribe to this RSS feed and then the news winds up on their desktop so I was so impressed

SPEAKER_07: by this I started Weblogs Inc I Gawker Gizmodo no I had two Gawker and Gizmodo and I was like he has two I'm going to make

SPEAKER_301: 100 blocks

SPEAKER_02: and that was the Dave Weiner influence it's hard to explain how

SPEAKER_00: revolutionary RSS was to people today because now so much media is decentralized anybody can start a medium or a sub stack in their publishing back then to publish yourself and send it out to anybody who has this link and they can just grab

SPEAKER_02: it it was it really changed everything so John C Dvorak an incredible inspiration

SPEAKER_07: then Leo Lepore oh then Adam Curry says to Dave Weiner I owe a lot to Dave Weiner who is

SPEAKER_04: an incredibly just an incredible innovator he created HyperCard he created RSS

SPEAKER_07: and OPML and then eventually Adam Curry said can you add to the RSS feed an attachment an enclosure and he said yes it's built just that easy because he was running the RSS

SPEAKER_04: standard then Adam Curry started taking MP3 files and attaching them to Adam Curry the famous MTV VJ

SPEAKER_12: he was another inspiration for me because he started buying domains and I was like wow this guy bought MTV.com I'm going to start buying some domains that's how I got into premium domains because of Adam Curry then podcasting started happening I said we have to do podcasts at Weblogs Inc so John C. Dvorak Adam Curry Dave Weiner then Leo Laporte started this week in tech

SPEAKER_39: with Howard Stern the curmudgeon the Roger Ebert of technology

SPEAKER_07: I'm sitting next to them Leo Laporte and John C. Dvorak mind blowing for me 20 years ago and I said to Leo hey I'm doing startups would you be okay with me doing this week in startups he said yes go for it I large portion of my career and Sunday

SPEAKER_48: I get to be on no agenda and just I want to be able to say that to Adam and

SPEAKER_12: then thank John C. Dvorak in memoriam and so it's going

SPEAKER_315: we'll put a link in the show notes

SPEAKER_00: Mattia Milanovich from Hop Arrow he's the founder and CEO of Hop Arrow they are a defense tech startup they're developing Rook Jason it's

SPEAKER_16: an autonomous hypersonic cargo rocket can't wait to hear more about that Mattia thanks so much for being here

SPEAKER_158: cargo rocket I've heard my friend

SPEAKER_07: Elon talk a little bit about this because he was saying someday you could take one of his rockets land them

SPEAKER_241: tell us a bit about Hop Arrow and where you're at with this welcome to the show

SPEAKER_322: yeah firstly thank you guys for having me really excited to be here so yeah

SPEAKER_324: we're Hop Arrow we're building cargo rockets that basically deliver time sensitive or high critical assets pretty much anywhere on earth in minutes we're starting with Rook which is our first vehicle that has a range on the order of hundreds are able to carry more payload

SPEAKER_158: how big is Rook how big is it in terms of height and capacity what can I put in there

SPEAKER_328: I obviously can't put a tank but I could put a bunch of guns and ammunition I'm guessing

SPEAKER_329: yeah so we're not publicly disposing the exact

SPEAKER_324: dimensions of Rook but very roughly 20 foot long and so inside of that we're planning on having a payload capacity of about 550 pounds so the biggest use case the one that we advertise publicly is being able to deploy autonomous systems so being able to go from one place in power things like that

SPEAKER_139: amazing so if I'm envisioning a scenario I don't

SPEAKER_12: know a random one in the Pacific Ocean maybe there's an island maybe that island is a couple hundred miles away from another island maybe that island needs some defense and we could send some defensive droids clankers you you know drones pick what you want on one side of the island and then hey maybe there's something going on the

SPEAKER_331: beginning of time it's been we've sent out

SPEAKER_324: 18 year old boys into the front line initially we had spears thrown at them then we had bullets shot at them and now increasingly other things and so where warfare is clearly trending towards is getting these 18 year old boys out of harm's way and ultimately the war fighter and so the way it's clearly trending is whoever is able to deploy autonomous capacity at range and at scale is ultimately going to be able to deter the next major conflict and

SPEAKER_70: you've been working on this for three or four years now I believe and no just under two years just under two okay so you're under two years how

SPEAKER_14: investors getting comfortable with it but you go to Ycombin and I understand and then you got to pitch investors on this if you're pitching it 10 years ago it would be no go but

SPEAKER_07: last two years we've seen some great military tech companies and some great space companies

SPEAKER_04: so the lane is there you know who to pitch as an investment how much do you need to raise to get from concept to reality how much did you raise and how close are you to reality

SPEAKER_324: you made a really good point we all stand on the shoulders of giants we would not be here without spacex without all of now in the previous conversation with Craig a lot of the why now is that AI is able to augment a lot of the engineering development so it's seeping into hardware development not just software development and so we're a very small team of seven and we've been able to move at the speed and ferociousness of a team of probably 50 or 70 and

SPEAKER_340: so this is a really great time to be building at the intersection of hypersonics and defense

SPEAKER_342: yeah

SPEAKER_14: and the landing and I'll tell you what you need 30 million dollars in this like series a if you're doing it like probably 20 30 million dollars is the bogey here in my mind 200 million dollar valuation 100 million dollar valuation whatever it is dilute 15 that would be what advisors would say to do here just taking broad strokes but the opportunity is you get a contract with Emil Michael and the department of defense slash war

SPEAKER_35: I prefer defense I love the war thing was a little too aggressive it's kind of like you're trying to pick a fight just

SPEAKER_343: defense is the perfect name for that

SPEAKER_35: department of defense is so much better like we don't want to be threatening people and

SPEAKER_12: provoking wars department of defense was so much of a better branding but let's put that aside we'll probably change it in two department of war we could have philosophical debates on that but this administration for the first time has said start up start up start ups lower our cost and we're going to support start ups with early contracts

SPEAKER_48: so how do you get in on the early contracts how does that work today game on the field for startups in

SPEAKER_324: weeks that speaks volumes compared to any other thing that you can do and so that's been the biggest thing that we've seen that prototypes only get you so far it's actually showing clear tangible progress a lot of our conversations do start off with again like how do you see this thing being like you talk about an OV1 in the military you look at how do you envision this fitting in and that's great that'll warm up the conversation but at some point you just got to whip out the iPad

SPEAKER_14: I will put this piece of advice we always have tactical practical advice here make it real what we're trying to do as

SPEAKER_07: entrepreneurs when we do companies or investors putting in there is you're trying to take the story and the pitch and make it credible how do you move from I'm just a pitch deck I'm just a guy with an idea or a gal with an idea to we actually have credibility and we're all in a credibility race in life the credibility comes from a fire test the credibility comes from the team can you know if you get past the pitch deck anybody can make a pitch deck but getting a team together that builds credibility but

SPEAKER_14: as you're saying here getting to real world

SPEAKER_07: make it real you have to make it real for people and just firing off an engine and saying hey I can fire off an engine without blowing myself up that's like zero to

SPEAKER_14: one in terms of pitch deck

SPEAKER_12: to real world capabilities that's that's what you're saying here yeah

SPEAKER_352: definitely yeah

SPEAKER_70: all right listen I wish you great success in this there are a couple of hot spots in the world where this is going to be

SPEAKER_12: the wrong metaphor for islands and houses but I live

SPEAKER_173: in Long Island and I got a place in the Hamptons and a place up in Albany I

SPEAKER_354: want to move a bunch of stuff I don't know what circumstance that is but is there a dual use here eventually

SPEAKER_356: well not

SPEAKER_324: ranging from manufacturing to coal chain transportation automotive other industries like that where it's not necessarily the value of the part itself is almost irrelevant it's what's the cost of not having it and so those industries usually have very high costs of downtime and so that has been an amazing wedge for us that being said we've had far more pull from the Department of Defense or Department of War whichever you're choosing is and so we we've just focused a lot of our energy on that that being said yeah there's a huge opportunity for us on the commercial side

SPEAKER_70: unbelievable really appreciate you coming on the site where can people learn more I'm this one brilliant idea looks like great execution to date but who are you hiring for and where can people learn more

SPEAKER_324: so we'll be announcing roles most engineering roles we need we have a great engineering team so far that's been our

SPEAKER_158: from hop arrow gonna change the world this one

SPEAKER_70: thanks Mattia did I pronounce it correct

SPEAKER_365: Mattia yeah just great

SPEAKER_70: and if you ever make it to Texas or if you're based here let's have a steak I got

SPEAKER_35: two of you guys together yeah Thursday nights supper club we'll get a steak all right talk to you soon brother thanks for you I'm good

SPEAKER_370: thank you for having me

SPEAKER_35: what what a banger episode here two for two we today on founders for two I mean you know what makes a great episode of this week in startups a little

SPEAKER_04: bit of news some great founders and some great founder lessons there and of course maybe a little more news and even going off duty what do you got left for us Lon is there another news story that the audience is demanding I comment

SPEAKER_00: I think we should talk about Nvidia taking over all of open source potentially they've agreed to acquire hugging face for 12.9 billion that hugging face of course the public marketplace for open source AI models data sets the github of machine learning it's sometimes called so Nvidia tried to buy it in late 2025 but they were turned down meanwhile Nvidia's earnings reports show another monster quarter forecast 70% 2028 revenue growth while analysts were expecting just 44% so Jason I know you were talking about this we've been talking about this for the last few days can you unpack for us what you see as

SPEAKER_28: Jensen's sort of play here he's open maxing as you were putting it the other day

SPEAKER_70: yeah so this would be the big number one story on all in which will drop Saturday morning in all likelihood maybe late Friday

SPEAKER_14: Jensen had a great open source nemotron project he has a great project internally doing open source for self driving in partnership with Uber and other folks building the open source stack

SPEAKER_31: he in his heart believes

SPEAKER_14: in open source however his customer base was largely frontier models who needs the most compute open AI needs compute anthropic needs compute these guys are or were a year ago 90% of the token usage 90% of consumption consumer etc obviously friends at Google with Gemini also have a billion users using AI across their products they may not be going there saying I'm going to use AI but they wind up using AI so he with those three specific frontier model companies and of course SpaceX I they are big consumers of this big customers of his this record blowout quarter is directly because of those

SPEAKER_75: let's say four or five entities you probably put Microsoft in there as well however

SPEAKER_14: open source is surging

SPEAKER_48: and he is looking at it going wait a second there's no open source champion in America who's the open source champion in America that comes to mind Lon

SPEAKER_02: I say open source in America you say nobody sure that was why I was struggling you say nobody it's a trick

SPEAKER_07: question there is nobody in maybe Apple did something in images and you know we talked about it here there's no champion there's no anthropic of open source yet I'll tell you who the open source competitor is it's NVIDIA NVIDIA when you say US open source you will say NVIDIA US open source equals

SPEAKER_383: NVIDIA Chinese open source equals Kimmy XAI

SPEAKER_87: DeepSeek moonshots the big labs now we have our US champion

SPEAKER_48: while the king was looking down the jester stole his thorny crown this is NVIDIA's big play Sam Altman if you remember did big deals with NVIDIA Sam and Jensen were super involved remember yes but you might be able to find in my notes the history of these two and you could tee it up for us here Lon I

SPEAKER_12: commitment commitment was the first time we started having this

SPEAKER_47: discussion about circular revenue it was

SPEAKER_00: it was in September 22nd 2025 Altman did a huge hundred billion dollar deal with NVIDIA for you know they were chips compute all of that then just 14 days later Altman announced a major equity agreement with AMD giving them warrants to buy up to 10% of open AI another one of these big circular deals this was part of what set off that whole worry about circular deals Jensen clearly was not happy about this by January just a few months later he was non-binding sort of already talking about maybe walking back that deal and he was reportedly criticizing a and I'm quoting lack of discipline in Altman and open AI's approach in response to this Altman then brought in Google vet Richard Ho that was when they started working on their own inference chip now called jalapeno right

SPEAKER_07: so things got chippy the relationship went from 100 billion in commitments to I Jensen he just went into the AMD deal went and hired this person to create the jalapeno chip which is supposedly spectacular and is a really great inference chip of course Jensen buys Brock gets his own inference chip Cerebrus goes public with their amazing inference chip now Jensen has to say wait a second if we're going to have

SPEAKER_02: the one the one element we have not discussed yet

SPEAKER_00: is pool side also this week NVIDIA made this huge deal a sort of an aqua hire they're going to take like 100 engineers from pool side six billion dollars plus a billion dollar or one billion dollar deal plus a

SPEAKER_16: licensing fee of six billion dollars so they're basically taking over one of the largest coding AI labs left that was independent

SPEAKER_14: codex open AI coding or clock code I think it's called Laguna I believe Laguna as in the

SPEAKER_397: great hippie town in Southern California Laguna

SPEAKER_70: is a very weird place I was enamored with it but you probably don't stay for more than

SPEAKER_12: 48 hours gets a little weird but Laguna is a beautiful town if

SPEAKER_379: it

SPEAKER_18: used to be like in the 60s 70s it was very bohemian it was like an

SPEAKER_00: artist colony so it has that like like in the keep Austin weird there was like a funky Laguna vibe that

SPEAKER_16: communities it's

SPEAKER_04: kind of like what happened with Venice Venice was like a weird art community now it's like a yuppie community so yuppie all

SPEAKER_00: these things get because it's so expensive to live in any of these places now the hippies and the bohemians kind of got kicked out

SPEAKER_70: so here's what I'll say Jensen's doing Jensen is going to take open source he's going to roll a

SPEAKER_07: rack into companies he's going to roll a rack into the same companies Anthropic and Open AI his top two customers I believe he's going to go to the same customers

SPEAKER_04: Dario and Sam are meeting with and he's going to say oh you're doing a bunch of frontier stuff with them that's great we power 72%

SPEAKER_39: of what they do that's fantastic might I interest you in a rack

SPEAKER_07: Uber and remember Uber Andrew and the CFO and Dara were talking about their token usage now imagine Jensen says hey

SPEAKER_48: would love to stop by he's a great sales guy love to stop by Dara

SPEAKER_227: maybe we take a couple pictures have a little lunch we talk about AI with your team have all your developers come okay great and he says how much you spending on clock was a clock spend was it open AI spent 400 million I got an idea for you spending 400 million on tokens

SPEAKER_48: I think that's probably what they're spending like literally a half a billion dollars going into frontier

SPEAKER_04: models 500 million if you want to know how Anthropic is hitting 50 billion if you want to know how open AI is hitting

SPEAKER_48: and he slides in and says by the way I got this Laguna

SPEAKER_04: and you're using this you're spending 500 billion can we get maybe 100 billion of that we'll slide in some racks they'll be here tomorrow and nothing you're doing goes up to the frontier model everything stays on prem you got the AI sovereignty and it's unmetered and they say oh what do you mean I metered there there's no shocking bill at the end of the month and the

SPEAKER_39: CFO goes wait a second you're telling me I can set up a rack here and my developers can go to town and I don't have to get in the

SPEAKER_07: middle of an argument between Dara Andrew the CTO this person on our AI spend because we and depending on how many tokens we can squeeze out of it and you're telling me your open source products getting 30% better every six months and it's going to get 50% better oh and you're optimizing the hardware and that's going to get 50% better every 18 months and have unmetered unmetered is

SPEAKER_04: the word we're going to be talking about in 2027 it's not going to be token maxing and that

SPEAKER_07: means massive headwinds for Anthropic and their coding products and the coding products are toast the frontier model coding products are going to have really hard time competing with an unmetered option this

SPEAKER_48: is going to be in the S1 for Anthropic and for OpenAI as their number one risk factor the number one risk factor is not data centers getting shut down Laguna on an NVIDIA rack and them losing

SPEAKER_12: the developer customers which are the tip of the spear that's why he bought Hugging Face that's where the developers hang out that's why he bought Poolside that's why he has those 100 engineers and that's why he has Laguna he can now go after

SPEAKER_237: the two most profitable parts of OpenAI and Anthropic

SPEAKER_62: there

SPEAKER_02: they'll keep with that but

SPEAKER_07: it's going to be NVIDIA Laguna which means Poolside Laguna okay sounds a little hippy dippy will it be here in three years will it be here in three years so if Poolside comes and says hey we want to roll in a rack with Laguna they say okay you got to get the rack we'll put the software on it's not one product it's going to have a thousand developers on it and instead of you wondering will it exist in two years and am I investing in a startup that maybe doesn't exist in two years you're going to say oh the largest company in

SPEAKER_04: of the market cap of the top two frontier models so if those two companies are worth five trillion they have

SPEAKER_07: a right to 30% of it that's 1.5 trillion dollars in additional market cap moving from Anthropic if it works moving from Anthropic and OpenAI one third of their revenue moving over to NVIDIA if it works this is the biggest news ever this is the news of the week not NVIDIA's blowout quarter it's in Jensen going all in on open source open weight coding tools is the biggest story of the year this is the biggest story of the year and

SPEAKER_416: nobody's talking about it except you except you are and now we will be across our social media channels anything else in the news that I have to comment

SPEAKER_28: well do you want to talk about as long as we're going to make we had already touched on this Stan Druckenmiller AI slop in the Wall Street Journal I don't know if you want to go back into it

SPEAKER_421: I mean I'm curious you're an artist

SPEAKER_14: a writer a creative when you go to the editorial page of your favorite publication and I'm sure you read the New York Times editorial from time to Washington Post or Wall Street Journal you wind up on the editorial page explain what

SPEAKER_07: your belief is and your expectation of an editorial page at a premier publication

SPEAKER_423: and

SPEAKER_07: there is Wall Street Journal is the number one premier publication in finance the number two in news in the United

SPEAKER_00: Stanley Druckenmiller there are so many avenues you could call a reporter you could have a press conference you could post a tweet you could make a little video of yourself a pod do a podcast there's a million ways to get your message out if you're going to go to a newspaper one of the most respected newspapers of the country and say I want to write an op ed for you to me that's setting the expectation that this is going to be a piece of writing I the early days of America where they were writing pamphlets all the time Alexander Hamilton sharing his pamphlets that's the idea or Thomas Payne with common sense it was like this conversation is happening I need to be part of it people need to hear my take and I'm going to build that argument in a piece of writing to me that's a promise to your reader that that writing is your own that it's your argument that you've thought it out that you're advancing it carefully and that every word every sentence every line of that is your own original work any piece of writing that you publish I think that's the promise you're making it

SPEAKER_28: writing that I do for the company I've at least consulted with an AI or multiple AI bots on

SPEAKER_430: it for research or proofing

SPEAKER_00: everything you could think of from what should I write about to what you know like what would be the good ideation research and then helping me I can be such a crucial time saver is get something down on paper it doesn't have to be good it just has to be relatively accurate and mostly what I want and then I'll go through and I'll fix it and that's the essential part of the entire process to me is the AI produce this I'm going to tweak it and edit it and change it so that it reflects my own work and that's also when you learn about it and think more deeply about

SPEAKER_439: started 2026 with like 1.2 I think I want to say something like that what are you why are you people following me is it because I'm

SPEAKER_441: spicy or

SPEAKER_14: proper venue is to jump on CNBC or all in podcast or whatever it

SPEAKER_07: is and then just discuss your position in your own words but you don't believe an editorial page should have AI written copy on it because if I'm interpreting correctly in that venue there is an expectation that you took the time to

SPEAKER_14: construct your argument in the prose and that's exactly how I feel if you're going to write something with AI it should be labeled as such in the first sentence up front

SPEAKER_07: just like if you make an AI video of a shark attack and it's a recreation of it in a documentary people would make recreations a reenactment and they say a reenactment or you see artist

SPEAKER_16: interpretation or artist rendering yeah

SPEAKER_07: correct so why in documentaries did people start putting this is a recreation and then this is actual footage why because the expectation of the audience was if you're going to put some video in there it's real video this is a documentary it's very simple I think the documentary analogy which I didn't get to this is the expectation of the user similar to a documentary if you're going to do a reenactment if you're going to do an interpretation you're obligated you are super

SPEAKER_14: obligated and the mistake that the Wall Street Journal made and then subsequently Druckenmiller made this combination of not checking

SPEAKER_07: the work and not disclosing to the reader is why there's such a strong reaction to this now there's a group of but you shouldn't say here's my ideas write me a piece publish it and not disclose it if we're going to do slop I want and I will call it slop because I don't think it this is the kind of stuff that erodes trust if you've eroded the trust of the reader the consumer of the information you failed at your job Wall Street Journal failed at their job Druckenmiller failed at his job they could have solved this by just putting written with

SPEAKER_02: AI at the top that's what I

SPEAKER_00: he wrote it with if he and I wrote it together it would say Stanley F Druckenmiller and Lon Harris and if it didn't I would write in the Wall Street Journal and be like what the hell why wasn't I credited on my brilliant op-ed this morning with my

SPEAKER_02: friend Stan does this piece have a disclosure now after this whole brouhaha

SPEAKER_04: on the commentary

SPEAKER_07: journal is holding on to

SPEAKER_02: this is a fine thing to do correct their response is basically this represents

SPEAKER_00: Stanley Druckenmiller's point of view he approved it for publication so it's it's fair game it's fine you're allowed to use and I that we use but I think the fundamental difference is that those other tools are doing the writing for you they're fixing or adapting your writing and Claude or

SPEAKER_16: GPT or whatever model he used to compose this is actually doing it for him and you don't I'll tell you people are

SPEAKER_14: a little

SPEAKER_70: spicy with politicians and here if you look at this url if you pull it up Andrew Sir Bain I'm not sure who that is this is a republican strategist with 97,000 followers his avatar is the famous wrestler from the film They Live

SPEAKER_16: oh that's Roddy Piper Rowdy Roddy Piper from They Live

SPEAKER_70: which

SPEAKER_139: is a very

SPEAKER_07: and he says it appears the most recent screed from the pro Amnesty Wall Street Journal editorial page attacking J.D. Vance was 100%

SPEAKER_12: written by him so now we are going to be in the awkward position of using pangram to go through every single editorial from the last year and we're going to find out who's not wearing shorts because

SPEAKER_15: the tide is about to go out I think this is people being deliberately obtuse

SPEAKER_00: I feel like it's very obvious it's easy to understand the distinction between no human in the loop and human in the loop if there's a human in the loop that's fine you're just using a tool it's making you

SPEAKER_241: I think there was a lawsuit about them I'm trying to remember new

SPEAKER_458: kids on the block there

SPEAKER_07: were a couple of these lawsuits if I remember correctly from the 90s or 2000s

SPEAKER_04: and I don't know how they actually worked out but if you were paying for concert tickets here it is yes I'm aided by willing attorneys fans such of such groups as new kids on the block are turning grievances real or perceived into class action lawsuits in what some experts say is an abuse of an otherwise valuable legal remedy for large scale consumer fraud not

SPEAKER_463: the new kids

SPEAKER_04: and not it is the new kids June 7th 1992 I that is I wonder if there needs to be disclosure for lip

SPEAKER_383: syncing when you go to a concert is that a thing the

SPEAKER_81: presumption is that's not happening on a wide scale that's what got milly vanilli in all that trouble

SPEAKER_129: yeah all right let's go off duty anything in off duty that people should be aware of

SPEAKER_16: here in Austin the mosquitoes are an issue I have to go out like fully covered head to toe when I take Barry outside or else I get stung it it's

SPEAKER_00: impossible so you tweeted that you you actually get to pleasantly watch the blue laser flash when it has spotted and is taking out

SPEAKER_28: one of your insect rivals

SPEAKER_70: I want to invest in this

SPEAKER_139: discussion about killing mosquitoes and bugs

SPEAKER_48: on the ranch and then what happened was the entire family got taken out by mosquitoes inside the house outside the house you live on a ranch you live in Austin there's like three or four months where it is acute and

SPEAKER_07: insane they're everywhere I have two of these in the backyard these things supposedly do two acres I put two of them just in the backyard

SPEAKER_12: to you know nobody getting mosquito bites anymore we have little ones inside we have these two big ones outside we put these two big ones outside I highly recommend you get one of these for your backyard plug it in permanently

SPEAKER_474: well

SPEAKER_48: here's the other thing

SPEAKER_07: if you scratch goes into the exposed bite and then you get a rash so everybody and this is a new thing for me as a Texan and for you as well never got this many mosquito by 10 in a week

SPEAKER_475: I got

SPEAKER_07: three or four of them on my legs then I got got an infection I had to get mupricin from an emergency care clinic I had to literally go there want to scratch my

SPEAKER_04: legs 24 hours a day that I start wearing long pants I get these two things and now I got it under control kill all

SPEAKER_14: mosquitoes around

SPEAKER_07: her house there's dozens of other acres for them to exist on but no more around our house couldn't agree more enough it's it's a nightmare anything else that's my off duty take make sure we put the link in there people love when we go off duty on Friday I

SPEAKER_28: have a couple very quick streaming recommendations I haven't done one of these in a while I watched a real I thought this was a really great movie I didn't see a lot of other people

SPEAKER_00: talk about it it's the wizard of the kremlin it's called it is it's a true story it's a film about the rise to power of Putin Jude Law stars as Putin Paul Dano plays his propaganda minister Vadim Baranov orchestrated his rise from you know head of the KGB basically government bureaucrat how did he become the not only the leader of Russia but this autocratic extremely powerful dug in it's a really fascinating story it's a long movie that kind of takes you through the entire like how the rise of Putin happened obviously fictionalized it's not the exact it's not a documentary but it's very compelling great performances from Jude Law and Paul Dano that's on Hulu and Disney Plus right now and actually it's Oliver Asayas I don't know how exactly to pronounce his name is the same director as

SPEAKER_16: Carlos we've talked about how much we like that Carlos the jackal movie oh

SPEAKER_478: this is the same guy

SPEAKER_16: director as Carlos oh I'm watching this this week

SPEAKER_319: if you

SPEAKER_14: can get Carlos the jackal Carlos the jackal is a five

SPEAKER_70: six seven hour miniseries but there's like a three hour version as a movie if you

SPEAKER_00: also condense it tell us the name of this one more time this is called the wizard of

SPEAKER_16: the kremlin and it's just a movie it's not a series it's like a little over two hours but it's

SPEAKER_00: it's very good it's really well done and it's it's like a lot of details that I'd never even heard before you know you know the outlines he was KGB and then he's always had this obsession about reuniting Russia really really interesting the other one I wanted to quickly mention another historical period drama this one's a series it's from the UK it's on Prime video it's called King and Conqueror and it's about the events leading up to the Norman invasion of England and the Battle of Hastings in it ends up being this big battle for who gets to be King of England very good well done show

SPEAKER_16: I'm really into it awesome

SPEAKER_439: are you watching Lanterns two episodes I am watching Lanterns I'm all caught up with what's your

SPEAKER_490: take on Lanterns two episodes I'm

SPEAKER_00: really enjoying it I think it's a nice balance it feels like it gets the comic book you're getting references it doesn't feel totally alien from the DC comics but it is its own thing and that's what I really like and or like the penguin when it can feel like its own TV show with its own vibe that is

SPEAKER_15: look up that rebel ridge is a great

SPEAKER_07: popcorn movie like just starts out it literally is like Rambo where like oppressed individualist yeah he's a best to like manage

SPEAKER_02: this battle with this local sheriff Don

SPEAKER_28: Johnson it's really well done he's a great actor and he's also a perfect Jon Stewart so yeah the Jon Stewart Hal Jordan vibe in this is a lot of fun and I recommend I

SPEAKER_14: love this actor I would say for this Lantern series it's a slow burn it's a slow burn it's very evocative

SPEAKER_07: of Ozark which was a great series slow methodical the aesthetics of the show

SPEAKER_48: are unique like true detective or Ozark it has an aesthetic that is not comic book aesthetic

SPEAKER_497: necessarily I

SPEAKER_81: that

SPEAKER_14: I think true detective or Ozark because the people who are involved in it were involved in those series I understand anyway I don't think you have to watch it episode by episode because it's incredibly frustrating because it moves so slow as a detective kind of thing I think I want to just wait till it's over and watch it like in you know four or five days I want to give an update I love my man bags I had talked about this alpaca bravo sling max 8L RFID blocking sling laptop bag I bought for 160 bucks I love this thing it's constructed incredibly well I like to take it with me because it fits my MacBook pro

SPEAKER_07: 14 inch 160 bucks I an expensive price that's a reasonable price for a bag sure so I think I'm a genius right this thing's got all these great features then

SPEAKER_14: to like you know maybe I can be a value in some way I'd like to join that board and give me another excuse to go

SPEAKER_75: one for Jordan these guys one for Zach one for Niels and one for Lon I need six of these ordered stacked please

SPEAKER_80: I'll let Heidi know I want six of these ordered to my office look at this six

SPEAKER_07: for the price of five for the price of one here's my Uniqlo bag that I

SPEAKER_12: in the United States you can it's in the slack room I see it my lord slow on the trigger boys come on now here we go this bag is essentially

SPEAKER_07: the same bag I like each bag individually they make it in black they make it in green so I bought it in green since I have the other one in black I find myself equally reaching for both the Uniqlo one is a little bit lighter

SPEAKER_48: but it has all the same features it doesn't have a faraday bag and it doesn't have a pass through USB charger but it's essentially the same bag it fits my number one thing which is I want to take my 14 inch 48 gigs of

SPEAKER_12: man I listen you guys know I got a little bit of chatter here on the side I made a little bit of cash a little bit I'm doing okay I'm doing speaking gigs here now and again I'm doing just

SPEAKER_48: fine I love value and when I find something that one ups the

SPEAKER_35: that was what happened all right listen another amazing

SPEAKER_12: episode of this week in startups this week in venture capital is what we do every Wednesday here

SPEAKER_07: on the show we also

SPEAKER_522: I dot AI

SPEAKER_07: exists go sign up for

SPEAKER_14: it it's got 8000 subs Lon has been given the directive to make this

SPEAKER_07: week in startups this week in AI and this week in VC sub stacks this is my instruction to you and this is important the reason I'm putting this in the show is I'm teaching entrepreneurs constantly I think sub stack is its own unique ecosystem you told me we're not get a growth strategy going and then what is the least amount of work we can do on these newsletters that provides the maximum amount of value that's the game we're playing you made a great short for

SPEAKER_12: me on for this week in AI I don't think we emailed it

SPEAKER_14: we're now doing shorts on how to and demos here's our first one you can find these on Instagram TikTok LinkedIn

SPEAKER_07: and YouTube whatever and X here's all the places and then tell me how many people then subscribe to those but a dedicated email with just that video and then ask them which other how to would you like us to hackaday became our most profitable best blog at Weblogs Inc why it did one thing every day we featured somebody else's hack with one sentence describing the hack and a link to it this thing

SPEAKER_04: took I don't know a couple of hours to find the best hack and 10 minutes to write the one sentence so it was really the curation of it

SPEAKER_14: thing to AOL the legal department said hey and it was time one or AOL at the time they said yeah we need you

SPEAKER_139: can take $300,000 off of the purchase price and let me own it I will personally buy it I took some of my proceeds from the deal and I bought it then I wound up selling it for a million dollars or so I think 500k or million dollars I made like a couple hundred grand selling it to someone because I didn't want to manage it

SPEAKER_12: I should have never done that but anyway Hackaday was this perfect combination providing massive value with a very small

SPEAKER_07: amount of work but you had no choice but to open it every day that's my goal for you with this week in VC this week in AI and this week in startups newsletters I

SPEAKER_48: want you to report back here on your plans for those and how it's going so we can teach people my editorial trick most value for the least amount of work that's the exercise I want you Lon Harris editorial director Jason to do you're on it see you all next time bye bye bye everybody