Discord’s Power Move, Slate’s Modular EVs & the “Act for Private Gain” | E2117

24:21They're like, no. And they're like, oh, well, when you meet him, you'll know in the first five minutes. Yeah, it's like your version of being vegan. Yeah, exactly. Cosmic vegan, Uber angel investment. Absolutely. So, you know what? I'm still along the company. I have exposure to everything else. I got exposure to Waymo. I got exposure to Tesla. I got exposure to everything. So I'm not like talking my book here all that much. I mean, I am technically, but Volkswagen is one of the top car makers in the world. They came out with an incredible car. It's not the car for me, a Volkswagen bus right now, but I was, cause we have a Suburban, which is like the largest SUV you can have.

24:58So when we have like a big family trip with a lot of bags, like when we're going to Tahoe, when we used to go to Tahoe from California, we're going on a road trip. we needed to have like a big, big, big max size, eight, seven, eight people in the car, or in our case, seven, five family members, one or two friends, or, you know, a nanny. Tons of bags. Like it's the greatest car ever. Suburban, tank, love it. But the bus is so cute and it's all EV. And I've been looking at it and I'm like, huh, it's a very cute car. And it also has seven seats, six, seven seat configuration, four captain's chairs in the back row.

25:29So like Slate. It's a bit of a platform and they have self-driving technology. Uber has a network. Volkswagen. Tell me how many cars they produce. I think I remember hearing top five producer in the world, maybe even top three. It looks like number two. It looks like in 2024, Toyota, number one, Volkswagen, number two, Hyundai, three, GM, four. Yeah. So they're very, very large global auto manufacturer. Like if you go to South America, you go to Africa, like they're making stuff everywhere. And I don't know what all their labels are, but they got a big push here. They've been, you know, really trying to get into EV stuff and they're going to do robo taxis with Uber.

26:08I believe it's an exclusive, but even if it's not, I had said earlier, like, listen, there's like 20 people working on self-driving. Let's assume half get there. Now you got 10 people get there. Now, what timeframe do the 10 people get there? Well, Waymo is already there at a very high price. So Waymo's in the game. Tesla's almost there, imminently there. And then you have WeRide, you have Zoox, right? So you've got arguably three or four people, like one really going for it, and then maybe three with operators or something. So we're on the cusp of this. It's going to be a long 10-year rollout.

26:45But Volkswagen has a unique ability to produce a lot of cars if they're the number two car maker. And Uber has the largest ride-sharing network. You put those two things together, and they took the right approach. I've been saying over and over again, my belief is that the second, God forbid, a kid runs in the street, your dog or your cat runs in the street, and we get on video, somebody getting hit by a self-driving car, whether it's a Waymo or Uber had one. Elaine Herzberg was her name. This incident happened in 2018. The first recorded case of a pedestrian fatality involving a self-driving car.

27:20Yeah, and I believe that immediately shut down Uber and it was in Arizona. It was in Tempe, Arizona. It was an Uber test vehicle. And the woman was playing Candy Crush. So this was an operator caused problem in addition to self-driving, right? Because the self-driving was like, you know, level two or something at that time. You needed to have an operator. The backup driver of the vehicle was charged with negligent homicide, pled guilty to endangerment and was sentenced to three years probation. Uber was not criminally responsible in the crash. That's very reasonable. I kind of feel like she should have went to jail for a year.

27:50I'm being honest. like if you're on your phone and you kill somebody like that happened to your family member to just get away with a warning anyway i'm gonna put that aside criminal justice is not our purview here exactly but i think anytime you operate in a city you should have to do one year and let's call it 10 000 successful rides i'm just picking a number folks but it seems reasonable to me that 10 000 trips of an average of five or 10 miles each gets you into 50 to 100 ,000 miles. I have a major concern about the rollout of self-driving. I'm super happy about Uber. When we get back from this quick break, I'm going to tell you my concern that could stop the auto industry from doing full self-driving for years in America and globally.

28:38When we get back from this quick break, your Your business is as strong as you are. And as a founder, you know, listen, maybe you're running on caffeine, four hours of sleep. That's not sustainable. Your startup moves fast and founders are pushing their health to the limit. And it's going to catch up with you. That's just the bottom line. But what? Optimizing your health was as easy as tracking your metrics. Well, that's what Superpower does. I am a customer of Superpower. I found out about it from a friend. Now they're here and partners on This Week in Startups. It's like the ultimate founder health membership.

29:07It's designed to give your body full testing across 100 plus biomarkers, organ health, hormones, inflammation, and more. You're going to get real data. You're going to get a blood draw, clear insights. And then I did this consultation with a trained clinician, and they helped me understand my ratings. There were some things I needed to work on, vitamin D, a couple of other personal things. I started working on them. It has literally changed my life. So right now, Superpower has a 150 ,000-person wait list. But because you're a Twist listener, you can skip the line. Visit superpower.com slash twist.

29:39Unlock Peak Health today because better health equals better founder, which equals a better business. This is incredible news for Uber. Uber is really up on the news. I believe of the 20 people trying to do self-driving, half of them will get it done in the same, let me pick a number, 12 months. Oh, okay. The same 18 months. I'll say 18 months. I'll be a little generous here. I believe 10 companies in the next 18 months will have these products on the road. And in fact, towards the end of that 18 months without a safety driver, with just a remote watch driver in limited capacity, limited geo-fence zones.

30:18And it's going to be awesome. I have a fear. The stock market is obsessed with self-driving. They're obsessed with it. And I know this because when you own these stocks and you see the moves, Tesla announces something, Uber goes down. Uber announces something. Uber goes up. Waymo announces something. Uber goes up. Tesla goes down. It's just, there's becoming, or Zoox and Amazon, there's becoming a little bit of, and BYD now getting in on it because they have self-driving technology. There's becoming, and you know how markets drive this. I think the teams running the self-driving efforts are going to need to come up with a standard for safety.

30:59And it does not, because they're done in local jurisdictions, cities, states, countries have all different jurisdictions and there's also like in other places, regions. We're going to need to come up with an industry standard. Like the MPAA came up with the ratings. This is a PG movie. Oh, Indiana Jones is coming out. We'll come up with PG-13. It's going to be a little scary, but it's not quite an R. But if you have a 10, 9 to 12 year old, Raiders of the Lost Ark might have been okay, but definitely Temple of Doom. With the heart, not good for 10-year-olds, 9-year-olds. It's too scary. So I think we have to come up with an industry standard.

31:38If we don't have an industry standard, now you're going to get some upstart who feels like they're going to go out of business if they don't get the announcement. You're going to have a company whose stock is down or whose stock is surging, and they get pressure from their board or from shareholders or an activist shareholder. Go faster, go faster, go faster. We have to catch up with Waymo. We got to catch up with Tesla. We got to catch up with Uber. We got to catch up with BYD. There needs to be an industry standard. I think the industry should be. Any city you go to, you do one year with the safety driver, everybody agrees to that.

32:08So there's not this pressure where Tesla's been giving mixed signals. Is there going to be a safety driver? Is there not going to be a safety driver? In what timeline? That's putting pressure, obviously, on Waymo to expand faster. Waymo is probably feeling like, oh my God, Uber and Volkswagen, they have a deal. Okay, is Uber going to be the neutral platform? Okay. And then BYD in China is probably saying, you know what, we have to get this on the road in Europe and South America before Elon gets there because he's very capable. One year with a safety driver, perfect record. Then you're allowed to do a 10 by 10 mile area where you had that safety record.

32:44If you want to expand again, one more year. Yeah. 10 ,000 more rides. I think it makes a lot of sense because people, especially people of a certain age, they're there. They'll get into this technology. They'll use it, but they need to like dip their toe and they need to feel comfortable first. And so it's sort of almost a marketing strategy as well. It's like, give people time to adjust to the idea of driverless cars before you just throw them on the streets with them driving everywhere and make everybody feel sort of threatened at first. No, you guys are both so wrong. We should go faster. This is ridiculous.

33:16I've never seen such mealy-mouthed, overly safety conscious. And here's why I think we can go faster. I actually agree with everything. Oh, you're serious. I thought you were joking. Oh, no, I'm dead serious. I actually agree that we should have stringent standards. We should have rules and regulations about this. And for context, the Volkswagen Uber deal is going to have people in the cars and it won't go driverless until 2027 to have time, Jason, to get it right. But I think once you've proven yourself in certain markets and you know how to roll out to new areas, I don't think you should have that many strictures on you because you're not replacing perfect drivers.

33:48You're replacing idiot humans behind the wheels of dangerous weapons. And so to me, as long as it's materially better, I just think we can go faster because we're taking risk out of the system even with that. It is a valid argument that if they are even 10 % safer than humans, you would want to adopt that technology. That is a very logical approach to it. However, on an industry basis, it does not matter. You will be judged so much harsher, a hundred times harsher for a self-driving car than a human. A human could be on their phone playing candy crush, be distracted, kill somebody. And we don't stop people from driving with their phones and phone enforcement has not changed.

34:35I think they should have cars that ride up right alongside you, videotape you if you're using your phone, pull you over, and they should take your phone. Yes. That should be the law. Your phone is confiscated and you get a ticket. Imagine... Listen, we all do it, folks. Somebody calls, you fumble your phone, you got it in it. If it's not in a tray, like if you're holding your phone, I'm talking about if you got it in your hand and you're doing this, like we all see people doing. If you're not using hands-free, you got it in your hands, you should get pulled over, you get your$100 ticket, they take your phone, they put it in the police station and in 24 hours you can go pick it up i'm like oh this feels quasi legal to me like i don't think they're allowed to confiscate your property i think i feel like there's a due process issue i don't think so because a civil asset forfeiture is an enormous problem with the american police force if you have a lot of cash in your car for example they can just take it they can right they could they could see it but it's but that's like you know like okay well listen i don't know the law you're probably right i'm going to go ahead and say you're right this is the new segment we're doing here called jay cow for president yeah right and i already put my first jay cow executive order this will probably be executive order six through ten yeah so i'm just putting it out there we have two executive orders are you gonna sign your executive orders and hold them up for the camera like trump i'm gonna actually do it on the show oh okay great i love that yeah i thought jason was gonna say your phone gets confiscated and and it gets crushed by a machine.

35:58So all the information is safe. But no, I really think that if you're driving and texting, you should lose your license. You're driving a multi-ton vehicle where there are children. I like the idea of a forfeiture. I mean, they could take your car when you speed over a certain limit. Yeah. I have a friend whose Corvette was put on a flatbed because they were going at a brisk speed. Was this friend going over 100, perhaps in a 60 zone? This friend of ours - Yes. was dumb, young, and had run into a little bit of money and needed to learn a lesson that he has carried with him for a long time now.

36:33And it was before he had kids. I want to add to the story though. Waymo did drop this week an announcement that they are now doing 250 ,000 paid rides per week up from 200. There's a lot of noise Jason made about like, oh, Tesla did their thing and now Waymo's trying to step on them. It's Alphabet earnings week. Of course they were going to drop the new number. So I view that as more coincidence than anything. But I want to reiterate that Waymo going to DC, I still think is clutch because I think it does show that they're confident enough to put Congress people in self-driving, which to me implies technology is advancing fast enough that we can go faster on the regulatory side because my Lord.

37:06What else we got on the docket? It's kind of what I'm saying. Yeah, you need to sell the old people on it. You got to show them that it works in front of them. I mean, it works in a controlled zone and controlled weather and, you know, just no accidents, please. And come up with an industry standard. Why not? Why wouldn't you have an industry standard? All right, next up, we're going to the insect route. We're talking about Butterfly Effect, the Chinese company behind the viral Manus AI agent. We talked about this a couple of weeks back. Everyone was buzzing. It turns out Manus, which was a way to essentially automate tasks very easily, was using some anthropic technology.

37:36So some people were like, well, is this really a breakthrough? Is this just great repackaging? Whatever the case, Benchmark is in, Jason. They led a$75 million round into Butterfly Effect with participation from other folks, boosting its valuation up to the half billion dollar mark. Where is the company based? I struggled to track down Butterfly Effect. I was going through a lot of fire lanes. Wasn't it a Chinese company? Yeah. Or they were making this announcement from China? I'm a little, I think there is a rule against investing in China right now. Literally, my question, my first question was, how did they do this?

38:10I mean, I'm seeing Beijing based is what I'm seeing. Beijing based Butterfly Effect. Maybe they're spinning the company out and put it in Singapore or in Hong Kong or something. They could have redomisiled it and they might have an office in China. which would be fine. So, but there was a bunch of divesting occurring because the Biden administration, you know, during the Biden administration, they said, yeah, no more investing in China. And China, you know, that's when that chippiness started. What else do we have on the docket? Perplexity has snagged a big deal with upcoming Motorola Razr phones.

38:41If you are not young enough to recall Razr's. Razr? They still make Razr's? Yes, apparently they brought them back. They're little folding flip phones with a screen on the front. and there's the new Razer, Razer Plus and Razer Ultras, which launched on May 15th, will come with Perplexity pre-installed. Now, the context here is that Google's currently in a lot of trouble for having a blanket, essentially monopolies on people having searched in their apps on their phones, if it's Android. So good for Perplexity. My question for Jason is pretty simple here. Are they doing too many non-core things as a company?

39:14So I feel like Perplexity is always in the news and it feels frantic to me, But I'm curious if I'm just being too conservative about how companies should approach partnerships if they can get major one. So they seem to be very partnership savvy. I notice a lot of times you get a free year to perplexity. So like, I don't know if it's like podcasts or other software companies that are really into bundling and trying to tap into other audiences and quickly get people into the subscription funnel. When you get a year free of something in order to get the year free, I think you have to put your credit card in and you get billed in month 13 kind of thing.

39:51I don't know exactly how they do it, but when you do that, you can then count paid subscribers. So your paid subscribers can go up even though they're on a trial if you have their credit card, but you know, and then you're expecting a certain amount of churn. So you got to be careful with that. I don't know if they're doing that, but I have seen that startups do that kind of aggressive tactic. There is a deal. You get perplexity for a free year if you are signed up to Comcast Xfinity cable and streaming TV platform. So I don't know if you have to put your credit card in to get it for free. You might have to do that.

40:27You have to be an Xfinity customer, so you would have to give them your credit card. So then maybe they're giving Xfinity half of the money in year two if you do it. So probably these kind of commission based deals are very popular with credit cards. If you get somebody a credit card, you can make, I don't know,$50 to$200. If you get them to open a brokerage account and put a deposit in, you can get like$300. That's why a lot of you'll see, like if you have SiriusXM or you have Hulu or YouTube TV, when they have like house ads that they run over and over again, it might be for E-Trade or something because E-Trade pays them a little bit of money, but then gives them a spiff.

41:03So there's an incentive for them to just run the ad in any empty slots they have. but perplexity seems pretty good at these deals. I do think perplexity is going to have a differentiation issue when you start to see people get into the habit of chat GPT, Grok when they're in Twitter, Facebook's AI when you're in those things. It's standing up and getting 100 million, 200, 300 million people to pay for a product. Other people are giving away for free. It's gonna be hard. It's gonna be hard. And there's been plenty of paid search engines that people have tried to take the ads out and give you a paid search engine.

41:39About 5 % to 25 % of consumers, depending on the category, like to pay for stuff. Most people like it ad-based. So it's just going to be challenging, but I give them a lot of credit. So interest in ChatGPT over time versus perplexity. Yeah, I mean, it's kind of unfair. Yeah. ChatGPT is such a juggernaut. I might do a better one would be like, well, everything but ChatGPT. So if you did perplexity, Claude, Grok, and Gemini, AI or Gemini, it might be interesting to do this. As a sort of lay person in the in the AI world, it really feels like ChatGPT, Gemini and Grok are way more visible. And then there's there's a ton of other companies.

42:17And I mean, Claude is like interacting with a lot of applications and like other kinds of. Yeah, that seems to be their sweet spot. Right. But like by far, those three are like the most prominent. And then everybody else, I feel like, is in a whole level. Claude is such a good product, though. I mean, I use Claude all the time. Right, and Claude is interacting with so many other startups and so many other products that I feel like that kind of is the explanation there. So this is interesting. We're looking here at Perplexity Gemini, Grok spiking there, and it looks like Gemini. And Gemini might include other Geminis, like there's a Gemini Trust or something like that.

42:52So that's a hard one to parse if we're talking about the actual Gemini. It's also a star sign, right? Yeah. There might be some baked in interest there if they're using a generic name. You know, I bought a Light Phone 3. I saw the former founder of Pebble wouldn't stop talking about the Light Phone. Have you guys seen the Light Phone before? You know that concept? I'm hearing about it for the first time. Okay. So I got a Light Phone coming, and this is what it looks like. It's an Android phone. I think it's using the Android operating system. It's basically all the things you need, but without the stuff you don't.

43:27So if you need to get directions, you want to play some music, take notes, have a calendar, have an alarm, use your phone. it's got those basic things. It's black and white. You get rid of your screen. I'm going to try this. So when I'm out with my daughters, I have a phone with me, but I don't have like a global streaming YouTube device to watch, you know, Nick's games on and get totally distracted by and no social media on it. And yeah, I can take my notes if I need them. But I think this could be another trend coming. It kind of is like the slate truck we talked about. Less is more. This is a third of the price.

44:00I think I pay for my phone at this point. my iPhone was close to$2 ,000, which I'm embarrassed to even say that because I always just, I'm just like, I'll get me max memory, get me max storage. And now I'm actually like, I'm an idiot. Wait a second. Do I need max storage? I actually don't. I have cloud. So I'm like, get me the lowest storage. I was just an idiot. And then I realized Apple is like literally making their margin from the idiots in their group. And then I'm like, for memory, I'm like, that kind of does matter. So with like laptops, I always try to max out the memory because I feel like you get one extra year.

44:30So you have to basically say, what is the cost of the laptop. And if you have max memory, you can get one more year out of your MacBook if you have max memory. So for an extra$400, if it's a 1500 and you like to keep it for two years or let's say three years, you can get a fourth year from it. But I think on corporate with these MacBook Airs, I'm going to go to keep them for three years, sell them back for 500 bucks, and then keep upgrading them. I feel like year four on a MacBook Air is not fun. What do you guys do? How many years? Yeah, I mean, I ride them until the wheels fall off. Like, that's my philosophy in general.

45:06But yeah, it's... I stopped you from doing that. Once your past three years, you start to really notice the device is getting sluggish as opposed to when you first got it and it was super fast. I keep my smartphone until the battery is failing me and I change out my gaming PCs whenever I want to play something that I can't handle. I have max graphics. So it depends. but probably three years is a good cycle. I just don't get people who still buy the new iPhone every year. Like, do you really need to have your camera back of the case things change? You're not going to notice. I'm not impressed by it.

45:39Well, I mean, the other thing that's in the news is, you know, the New York Times is suing OpenAI for copyright infringement. They got a heck of a case based on my assessment of it. I think it will result in, you know, perhaps a record setting fine for theft of IP. And, you know, these things almost always get settled. I think it could be a billion dollar settlement lawn. And I think that would be fair because I used to subscribe to the New York Times for Wirecutter. Now I ask ChatGPT, which I pay for, or Claude, which I pay for. Yeah. Actually, it doesn't come up on Claude, but it definitely came up on OpenAI.

46:18But I will ask, what did the review site say is the best air purifier? And I'll specifically say, what is Consumer Reports and Wirecutting? I used to subscribe for Consumer Reports. I used to subscribe to Wirecutter. And I'll just ask, what are their best ones? And they wind up giving you the answer. And it's like, well, that's behind a paywall. And then I'm like, well, how did they get the answer? And it was obvious they were scraping it. Now they're not. They should have paid for a license. They should have done it in advance. And now they're saying, oh, we're going to lose to China if we don't respect IP.

46:45But Ziff Davis has now joined the fray, I see. And Ziff Davis has a lot of content, a lot of like real technical content. So what's going on with that? So Dick Ziff Davis says that OpenAI has, quote, intentionally and relentlessly reproduced exact copies and created derivatives of Ziff Davis works, which it claims are infringing on its copyrights and diluting its trademarks. Further claims that OpenAI use Ziff Davis content to train AI models and generate responses through ChatGPT. They're looking for nine figures in compensation. conversation i think it's going to be because open ai is the biggest and they're worth 350 billion dollars and they're making 10 billion and they're making that 10 billion because they have the world's knowledge in there yeah and uh you know it's pretty clear that they were fast and loose in terms of the early days i don't know about today but you know they're trying to be more thoughtful now like in retrospect i think they realized they did the wrong thing which is great that they realized they did the wrong thing but now you got to make amends for it and they should have a licensing program.

47:45You should be able to come to chat GPT and open AI. There should be a form where you say to them, you have my content. This is what my content costs pay me. So if they index this week in startups, I could say, you know what, as long as you link back to it prominently and you link back to the homepage and you link back to the YouTube page and you use no more than 30 seconds of a clip, I'm okay with it. Like if they came to me with that offer, I might be like, okay, yeah, 30 seconds. I'm fine with, or if they said we'll index, you know, 2 ,000 episodes or whatever, and we'll give you$100 per episode per year because it's great content.

48:17We'll give you 200 ,000 a year. If they came to me and said, I'll give you$20 ,000 a year, like literally that would be$10 per episode. I'd be like, okay, fine money, $20 ,000. I'll take it. Yeah. And I mean, so many media companies, you know, companies that produce journalism, they're struggling so much now they desperately need this kind of revenue boost. They probably would take a discounted deal if there were some kind of offer being made. And actually, we already can see how it would work. Like if you use Notebook LM or you use Google's deep research, it shows you where it got every piece of information.

48:48It hyperlinks. There are footnotes that you could follow. Like every AI engine should be doing that. Did you see this story we posted in the chat? Oregon state lawmakers are now considering a bill that would require big tech companies to pay money based on them using AI. Feed that into local newsrooms and journalists. So it would like use some of the profits you're making off of these AI tools to fund professional journalism and provide a financial lifeline to Oregon newsrooms, which is, I think, a great idea. Like that's exactly how, you know, like when Netflix wants to go into a new market like Turkey or France, the local government will sometimes say, OK, but you got to spend 15 percent of what you earn from the Turkish market.

49:33Feed that back into Turkish original content. to just be generous. Everybody should. It's a no-brainer. You know, New York Times is probably going to want to go it alone. But I think Disney, New York Times, ESPN, Ziff Davis, Condé Nast, like some of these are major multi-billion dollar, hundred billion dollar brands. They should just all, tens of billion, they should just all get together and just pile these lawsuits on until OpenAI cries uncle and says, okay, here's the settlement and here's the go forward. Yeah, and these companies have such deep pockets, they could save the next 10 years of American journalism and barely even notice it on their bottom line.

50:10Yeah, they don't care about that. What they care about is precedent, I think. And people in Silicon Valley, I'm not speaking specifically about OpenAI, they generally don't respect artists. They don't respect content creators because content creators have been suckers. And they're like, yeah, we'll build YouTube. We'll build Instagram. Yeah, give us a fun playground. You give us some traffic and free software and we'll put it there. But over time, people start to say, you know what, enough's enough. We want some control over our content. it's our content and you know they they kind of do that there's also i don't know if you've seen it and it's kind of like a i think it's kind of a breaking story i sort of come up this week not for private gain.org a bunch of folks from open ai and these are you know real experts are doing one of these um so they've written this open letter and it's basically to this attorney general or these attorney generals about this nonprofit structure and what everybody bought into.

51:03And, you know, they want to understand these fundamental questions. So they really laid it out nicely. Like number one was understanding open AI's charitable purpose, why nonprofit control protects the mission. And, you know, they're using quotes from Sam Altman and other founders, you know, of open AI to kind of say, Hey, this is what we all bought into. and they're really questioning if open ai is going to solve the problem of profit-driven agi and give it away for free to humanity and it's very interesting because if you look at the signatories these are some like pretty heavy hitters people from open ai but they also did a very smart thing they went to all the law schools to talk about non-profits and non-profits is like a very, I don't know how this concept actually came to be.

51:53And I'm going to try to find the history of it at some point, but I would like to understand why we have this nonprofit structure, why religions, you know, why rich people can pop up a nonprofit, hire their families to work for it, donate money to it, and then use it, you know, for all kinds of purposes. You know, like I'm going to have a nonprofit golf tournament and like all the rules around it. it all feels like maybe it's unnecessary for us to have nonprofits. Like if you're a rich person, you give a donation, would your donation change if you didn't get to the tax break? And should nonprofits of a certain size have to pay taxes?

52:26And it just seems a little bit weird to me, the whole nonprofit structure. But in order to preserve the nonprofit structure, allowing OpenAI or any company to just flip, flip, flip into for-profit and take the original mission and get rid of it, it's as other people have said, well, if you can start with a nonprofit and I, as somebody who, you know, I, let's say I made a million dollars in profit from a stock sale or from income and I get to give this million dollar donation. Well, then I get two for$1, let's say if I'm in a high income tax bracket to donate to a nonprofit, I do that for three years.

53:04If it turns into a real company, I flip it into a for-profit company. And then I was able to fund it at half price. It seems to purely violate the spirit of what we're talking about. Like the whole point of a nonprofit, we're giving you this tax exempt status, we're giving you all of these special benefits because you're not pursuing a private interest, you're pursuing the public good. So if you could just switch at any time, it's like, well, so this whole time you really have been serving your private interest just quietly, you made it look like it was in the public interest. But so, yeah, that that feels sort of ridiculous to me that you You could just change it around at any point.

53:38It's like, no, you established this. This is what you're doing now. It's a new organization if it's now for private games. Yeah. I saw the FT report also that Apple is going to shift all their U.S. iPhones to India as soon as next year. Which is, you know, China's not happy about it, but they had set this up as their escape hatch. And here we are. This is their escape hatch. Yeah. Let's do our Founder Friday. Let's get another pairing done. We got one more matchup left to go this week. We're in the midst of our Elite Eight. Last time, you'll remember MedSimple from Florianopolis. They are our first Final Four contender.

54:14Congratulations to the team from Florianopolis Brazil. Today, we've got Pitchfire from Portsmouth, New Hampshire, versus another international contender, Takhtun from Yerevan, Armenia. So first, we're going to look at Pitchfire. They are a paid prospecting platform where potential buyers can earn money for reviewing and responding to sales pitches from B2B companies. And just as a reminder, we've asked the companies for this round. They're giving us an insight into their two-year plan. 80 % of B2B companies are missing their pipeline number, and everybody needs getting cold prospected. Let's give you Pitchfire's two-year plan on how we're going to fix this with our paid prospecting platform.

54:52Businesses pull buyers in by running paid prospecting campaigns over email, offering people money to respond to their sales pitches. Once a buyer responds and receives payment, they automatically join our marketplace, and we push them to install our inbox plugins. So why would they do this? Because it removes them from that sales rep's sales automation and gives them a shot at potentially making money for their attention. It also helps fund their sales team's paid prospecting efforts. A low-level B2B employee gets prospected about 708 times a year. They would push at least nine new sales reps onto Pitchfire every year.

55:19Pitchfire's path to average$100 ,000 in revenue in two years is all tied to generating responses for paid pitches. As Pitchfire grows, buyer-source pitches will be our main source over the next two years we'll hire eight full-time employees heating to eighty thousand dollars of burn every month in salaries we'll start with growth marketing another full stack rent engineer and a sales rep to push companies to do paid prospecting next we'll hire a life cycle marketing expert and another sales rep beginning of next year we'll have a machine learning and ai engineer on staff full-time as we get more volume we'll need customer success to keep customers on over the next two years our product burn takes about nine thousand dollars a month to operate at peak capacity.

55:56A roadmap will start with adding SMS notifications for people to get pitches, and user reviews to earn trust between buyers and sellers, a UI in our product for paid prospecting campaigns, and a company wallet, an integration for marketing teams HubSpots and SOC2 compliance, a chat interface that welcomes rebuttals and allows upsell opportunities, vendor and buyer matching, and user management to expand teams on the buying side. In Q3 of 2026, we'll deploy an AI auto-reply application that will automatically detect cold emails and send it to Pitchfire. And by the end of next year, we'll have a buying agent answering pitches that come in from sales reps.

56:28If we can get 7 ,937 paid pitches responded to with our 20 % cut, we'll hit$100 ,000 in the last quarter of our two-year plan. What I like about asking them to do a two-year plan, if you can start to see if they actually have a thesis and they understand how to budget money, and that's what we got here. So your confidence as an investor goes up. because if people plant that flag and say, hey, here's what we want. Here's the ask. We're going to raise two and a half million. We're going to spend, we're going to burn a hundred K a month. You can start to have a really granular discussion and look at the assumptions and then start banging on the assumptions and doing 8 ,000 intros.

57:05I don't know if that was a quarter or maybe that's a month. It doesn't seem like a giant number. If you remember, this was an interesting SaaS company. You get paid to evaluate a product. So I'm making Zoom. I am Elon. Hey, we have this new thing. it's better than using Skype. It's better than using Google, whatever they call Google's one, Google Meet. It's better than Google Meet. Would you evaluate it for us and see if it's great? If you do, you get this$40. They evaluate and say, it's not for us. Great. We take you off the list. You got 40 bucks in your account, but you work at Slack and you evaluate it.

57:38And then your Slack team gets whatever they get, like 15 bucks. The other company gets 15 bucks. And you kind of, then you can go take Slack and you can ask people at Zoom and other places. By the way, this is what happens in the startup community. People always ask their friends to buy their products. Certain accelerators are known for like trading. And I don't think the accelerators encourage people to do this. I just think it happens when you're in a cohort. Hey, you buy mine, I'll buy yours. Founders hack things. So we always have to look at like, okay, yeah, tell us your top 10 customers or which five customers did you get this week?

58:12tell us the names, how did you source them? And if they say, oh, we sourced them while in our accelerator, we say, okay, great. Of those five, we kind of, we'll just kind of laugh it off. Oh yeah, we know that trick. You know, that's great. Do any of them actually use it and get value? And then once they tell us if they got value, then we're like in a great place with them. Okay, let's hear our next one. All right, so from Yerevan, Armenia, this is Taktun. They're creating a platform that allows machine manufacturers to develop intelligent devices without having to write all of the complicated code themselves.

58:42Taktun is a no-code automation platform that enables machine manufacturers to make the machines they build intelligent, connected and autonomous. Over the next eight quarters, we're scaling from a working platform and initial pilots to a robust AI-enabled product deployed across the US and EU. Our journey starts with launching VASHA 1.0 and securing our new paying customers. By 2026, we reach autonomous deployment capabilities and begin European expansion. We target 5 million in ARR by early 2027. Revenue accelerates quarter over quarter while expenses remain stable. We cross breakeven in Q4 2026 and stay profitable from that point forward.

59:24Our headcount strategy combines core human hires and full capability AI agents. As of Q1 2025, we have 11 team members. We add new hires quarter by quarter. By Q1 2027, the team includes 24 human employees and 7 AI agents working across engineering, DevOps, support and QAIP. We track what matters. Growth in customers, operating deficiency and cost ratio. metrics improve quarter over quarter as we scale showing strong fundamentals without exposing sensitive financial margins our operating plan is fully aligned with capital efficient strategy we plan to open 3.6 million seed round later this year to fund execution for q1 2027 the capital supports product delivery go to market and series a readiness thank you very much Huh.

1:00:16Okay. What do you guys think? Which one is your favorite and why? I'll start. Taktun gets my win for a number of reasons, but the most interesting one is that they listed AI agents that they wanted to hire in their employee section. That's new to me. Very interesting. Yeah, I noticed that too. I thought that was really interesting. They're like, we're going to have 24 humans working alongside seven AI agents. It reminded me, one of our founder, You Companies, has created a virtual board of AI characters and they consult them every day. but like they gave them different personalities. So like, this is our marketing expert.

1:00:48This is our, you know, business operations expert. I thought that was really interesting too. Of these two, you know, I don't want to make it non-competitive, but Taktun also, I think, is going to get my vote. I just think it's such a smart idea for a platform and they seem to have the, you know, they seem to have already figured out how they're going to grow it and expand it. And I just feel like that to me feels like the future. We're going to have so many companies trying to make smart devices that work with their software. This just seems like a no-brainer to me. Yeah, I think it's going to be a clean sweep.

1:01:19You know, as much as I think Pitchfired did a better job on their pitch, and I thought their planning was better, I do think it's a bigger opportunity. So I think Taktun, congratulations, you're moving on. Taktun in our second Final Four contender. So wow, that's very impressive. Well done to Taktun. All right, everybody, another amazing episode of This Week in Startups. a lot more coming on Monday. We'll see you at 12 p.m. Texas time, 1 p.m. Eastern time, 10 a.m. Left Coast time. If you're interested in Founder University, go to founder.university and learn more. We'll have another cohort starting over the summer.

1:01:54And that's where we have three, 400 teams of two or three co-founders build their MVP and then decide if they want to take a little bit of funding from us or maybe incorporate and take their, you know, business this idea to the next level and go down the venture track. We're doing a lot of great content over there. You get to hang out in 25 sessions with us. 25 sessions over 12 weeks. You're going to learn a ton. Founder.university. He's Lon Harris. I'm Jason Calacanis and he's Alex Wilhelm. We'll see you next time. Bye-bye. Bye-bye. Hey everybody. Welcome back to Twist. This is Alex. Today we have another interview with an awesome Twist 500 company.

1:02:33Now this startup is one that I caught wind of earlier this year when I read that it had gone from 5 ,000 corporate customers to 10 ,000 in less than a year after spending four years to get that first 5 ,000. I looked into the company, didn't fully understand the technology, but thought they were really up to something awesome, threw them on the list, got them booked for an interview, and then the company, Tailscale, announced an enormous Series C right before we jumped onto an interview. So please join me in welcoming Avery Pinneron, the CEO of Tailscale, to the show. Avery, how you doing hey pretty good hey how are you i'm hanging in there i love that you're here and also you guys are a canadian company yeah we are i'm sitting here in montreal right now montreal now is the company based in montreal i thought it was toronto uh it's a remote company so basically all the employees are remote so wow my two co-founders both named david one of them lives in toronto and one of them lives in berkeley california how do you handle having two different davids do you just call one like california david and one like north david we called them by their last name it's uh it's carney and crosshaw now that just got really complicated because the new prime minister of canada his name is carney um yes mark carney um no relation who got david at uh tailscale.com because i feel like that's gonna be the hot domain name inside the business you know what i haven't actually checked that i'm guessing it's carney because he he is sort of the operations focused founder and administrative stuff i was really unlucky i got uh ap at tailscale.com as one of my that turned out to be a big mistake because uh accounts payable is ap oh i get a lot of misdirected accounts payable related emails well that means you're always you have your finger on the pulse of how much money is going in and coming out that's not a bad place to be all right now now listen avery uh tailscale works in the world of secure i'm sorry vpns for secure networks i I want to talk about the technology behind this because I think it's very important for people to understand.

1:04:28But instead of me trying to butcher an explanation of what this is, I think it's best to let you explain and then ask you a couple of questions. So for folks out there who don't know, what does Tailscale do? All right. So super simple answer is Tailscale is technically a VPN. That is the space in the technology world that it occupies. That doesn't really tell the whole story because VPNs have basically stagnated for 30 years before Tailscale showed up. And so it doesn't work like IPSec or whatever you're used to. Tailscale uses this new thing called WireGuard. And we can talk about WireGuard. It's an open source project.

1:05:00I really like it. And Tailscale is a mesh VPN, which means it connects anything to anything in any location and all the locations at the same time. So instead of sending all your traffic through a VPN consultator or a VPN server that you might be used to, you can just make a direct connection between point A and point two things that are running Tailscale. and so as a really simple example of what a lot of people use our free plan for you can download tailscale from the app store on your phone and on your laptop you log into the same often google account or apple account or microsoft account and when you do that those two devices use our servers to find each other no matter how many layers of firewalls they're behind without opening any ports or anything like that and now those two things can talk to each other so you can access a web server that's running on your laptop or you can use a feature that we have called tail drop the list it's sort of like airdrop but it doesn't depend on apple equipment and it doesn't depend on physically being on the same network and you can just send photos from one thing to another uh so if you've ever had the problem of trying to send a photo between an android device and a mac uh this solves that problem now there's many many other things it does tailscale like you know it's a vpn that enterprises use for multiple data centers it's all the same technological problem which is the vpn problem but sort of expanded to the modern world so people are listening to this probably have heard endless ads from like NordVPN and the other kind of major consumer players that love to advertise on YouTube and other social panels.

1:06:22So is VPN in this context, the same kind of like underlying idea in which you can create essentially a, I was gonna say artificial connection between two things, but one that is distinct from the network that you're on currently. So similar idea, just a different application. Yeah, there's sort of two kinds of VPNs in the world. There's the classic original kind of VPN, which is take your stuff and be able to connect to your stuff when you're not physically right next to it right so it's called a virtual private network the idea is you had a private network now you have a virtual one it doesn't depend on the physical location anymore you connect to your private stuff that term has kind of been stolen by these so-called consumer bpns or privacy bpns that actually don't work like that at all they like they let you take your your traffic and then you route it through them where they decrypt it and then send it out to the internet and so it's more like a virtual public network but I guess it's the same acronym.

1:07:12It's the same acronym, yeah. Yeah, so Tailscale is the first kind. Use it for connecting to your stuff. Now, to make things a little more complicated, we have a feature called Exit Nodes that allows you to voluntarily, if you want, write all your traffic through one of your other Tailscale nodes. And we have a partnership with Mulvad, which is the other kind of VPN. So you can subscribe to Mulvad through Tailscale and use Mulvad as your exit node. So you can kind of do both things at once. All right, so I got about 80 % of that. So I think that's probably enough to go back to the start of this.

1:07:40So I was reading a lot of tail scale material and you guys were talking about how the internet kind of hub and spoke model is just super inefficient because data has to be sent too far. For example, you had an example of going all the way across the United States and back, for example, waste of time. Mesh networks allow for closer connectivity points. Can you just explain how that technology works? And then I want to talk about security in a second, but I think mesh is probably a good thing to unpack a little bit. yeah so the really neat thing about tailscale and and the wireguard protocol that it's based on wireguard creates these super lightweight tunnels so where traditionally setting up a vpn tunnel was like difficult and expensive and so you wanted to have as few tunnels as possible and you solve that problem by like having one tunnel between you and the center point and then another tunnel from the center point to the other thing and that's only two tunnels and that's like pretty manageable as opposed to what tailscale does which is like if you've got a thousand nodes in your network you've got a tunnel from each of those thousand nodes to each of those thousand other nodes, which is, you know, comes out to a million tunnels, right?

1:08:40That is a lot of tunnels, but because they're so lightweight, it doesn't matter. But as a result, it means you get a direct connection between the two places. You don't have to go through a center point and it doesn't add latency. And there's no, there's no place to get overloaded. So the more nodes you add to your network, the more capacity of the network goes up. So overall, there's not a throughput bottleneck like you would get in a traditional architect and the way that tailscale kind of does works is that it sits in the middle but it doesn't pass all the data through itself but instead it seems like pass like pass keys and permissions to allow the mesh network to send things quickly without having the hub and spoke issues that we just described right so tailscale we divide the product into into what we call the control plane and the data plane and it's actually kind of clever the control plane is very centralized there's a sas product that you can log into it's got a web console and stuff and it like you said it's basically an exchange point for the public keys that you're are generated on each node that you own so the nodes that you own generate public private key pairs the private key never leaves the node so you get end-to-end encryption but they find each other by basically telling the coordination server in the center it's like hey i'm node x i'm over here i have this ip address and i belong to avery come and and here's my public key if you want to talk to me come find me right and that's basically all the coordination service does everything beyond that is a distributed system right with no single with no bottlenecks or single points of failure so this hybrid centralized distributed system is what makes tail scale so magical i started off with notes about your massive growth in the enterprise space so talk to me about um who are the customers that need tail scale in the corporate world so oh yeah so businesses uh use tail scale and we charge them money because they they're businesses they have money and they get a lot of value out of tail scale uh the way that works is all of these people who use tail scale for free the majority of them are kind of like semi or fully technical kind of people uh technical people have jobs frequently uh and and they find out about tail scale they really get excited about tail scale they solve their own problems at home using tail scale and then they go to work and find out they have the same problems and the way they're solving it sucks right and so they will complain to their co-workers and say like how come tail scale at home that i'm getting for free can solve this problem and this junk that you guys are paying tens of thousands of dollars for does not solve this problem and that's basically how the conversation gets started and then we sort of get you know either a small number of people using it at work on like some corporate credit card or it turns into what we call a top-down sale so it like flips some executive gets word of it and decides to buy it for the company so as a result the way we sell tail scale is almost exclusively through inbound leads generated in exactly this way and i remember like it's so funny on some of these calls they're all get called in and there's like the CTO of a bank I think you know tens of thousands of employees and they're thinking of buying tailscale and we always ask like okay well what how did you hear about tailscale they're like well my friend runs an Airbnb and to control the Apple TV at his Airbnb he installed tailscale on the Apple TV and he was really excited about how it works so he suggested that I should buy this for my bank so here we are so it's always some weird story like that this really gets at a question that I had because it seems like the technology you've built works from the hobbyist equivalent use case all the way up to the highly regulated banking use case.

1:11:58And it's so rare to hear of a technology that works across that kind of scale. Was it hard to build it that way? Or is that just a facet of how Tailscale works that it does function for people in their bedroom and then also people in the boardroom? Well, it's good that you mentioned the word scale because it's also in our company name. uh so the the the the name tail scale is actually started out as a joke it's the opposite of internet scale and so the purpose of tail scale we're like look lots of companies everybody wants to sell to enterprise everybody wants to build something that scales to billions of users like those problems are like solved more or less right might not be solved really elegantly but there's some way to solve your problem if you if you have billions of users and way too much money right the the problems that are not solved are the ones that like we actually have to deal with on a daily basis, the long tail of problems that every software developer or technical person needs to deal with.

1:12:50Like, I want to start a dashboard and serve it to my 10 person team. How do I start a dashboard? And it actually turns out to be a nightmare. Like 90 % of the work you have to do, make a little dashboard for your team has nothing to do with making the dashboard for your team. Everything about like spinning up virtual machines and containers and opening ports and getting a DNS name registered. And what about a certificate? And how do I make the login process work? And how do I lock out the other people? Where should I open the firewall ports? And we made a list. and we're like, this sucks. Why don't we solve that stuff?

1:13:16Right? So tail scale is like for solving the long tail of problems. So we're like, look, we are going to target the people that nobody else targets, these teeny tiny teams and companies at the very bottom of the scale. And some of these tiny teams are in big companies, right? Like, you know, your small team still needs to make a dashboard and your corporate architecture does not make that easier. What has happened in the subsequent years is that tail scale keeps getting pulled up market, which is a good way to go we very rarely try to push up market if that makes sense we have actually been really resistant to taking on big contracts that are going to distract us too much from building like the core product that we know is good for everybody but every now and then the core product that we built is like just one little you know one little feature away from like hey my my 10 000 person company can adopt it now could you just do like this one thing and we're like yeah that seems like a good idea so we build the one thing we'll take that revenue but like a bunch of other companies suddenly can now use tail scale.

1:14:13And so we are now up to the point where we're seeing like million dollar a year contracts with very big companies with really high regulatory requirements. Then we got there just incrementally piece by piece by not chasing them, if that makes sense. We do very little of the actual chasing. And that makes us pretty unique because the nice thing about starting at the bottom is that you can't build a product that is too difficult to roll out, that is too expensive and slow to get started because people are just not going to do it right the five person company doesn't have anybody with enough time to roll out an actual complicated vpn product but they have they have enough time to spend the five minutes to install tail scale and that means this really nice experience can go all the way up to the enterprise where they're absolutely not used to that they're used to products that they buy that are going to take a six month rollout period and a whole team of people to roll it out and they're just sort of like boggled when they try tail scale and it's like hey i'm done already hey i mean great but But I'm a little perplexed between the fact that you're being pulled up market, that you have tons of inbound demand, things are going well.

1:15:16Why raise a$160 million series? Because it sounds like you guys could have just kept growing without more external capital, a rare and awesome place to be in, but your GTM machine doesn't seem to eat$10 million a month. So why raise more capital now? Great question. so tailscale is a very efficient company and tailscale is a very product so those two things together mean that we technically um can we can switch to a path of profitability whenever we want and i try to run the company in such a way that whenever we need to we could just say okay that's it put on the brakes on the right like cost growth um and like get get it under control and we can weather any storm right and that's obviously been super valuable in the last little while with the like ups and downs in the market and so on but i also have another saying which is that the best time to raise money is when you don't need any money uh so you know not needing any money uh has been a really uh good asset to us um so we actually haven't at the time we raised our series c which is we announced a week ago uh and closed a few days before that or sorry we announced it yesterday yeah time flies um i was about to be like wait can i really get my notes that wrong It closed about a week ago.

1:16:30There you go. Yeah. But at the time we had that announcement, we'd actually spent less than half of the Series B round that we raised three years previous. And so the question is like, look, if you're spending money that slowly, why do you need even more money? And the answer is pretty simple. Like, first of all, it's the right time to raise money right now. Secondly, we are getting pulled up markets sort of faster than we expected. And those opportunities getting appearing faster than we expected. So we talked about this, like, you know, we doubled from 5 ,000 to 10 ,000 paying customers, which does not include the free paying customers, by the way, it only includes organizations.

1:17:02In the two months since we announced that we've gone up another 20 % in terms of number of customers, right? So it's just it's, it's, it's going up really fast. And the customers that that's the customers we're getting that we're making money from, there's also this like, even longer tail of customers that we almost get, right? But if only you had this, if only you had this, if only you had this, if only you had this, if only you had this, and it's a whole bunch of little things for the most part right but if we just go after those little things the revenue growth is going to be extremely high and so this comes down to just the financial concept of roi like do i need to spend the money no i can be profitable without spending the money but if i spend this money i can make even more money and everybody wins faster it's a really like you know you do the do the formulas it's like oh this is a really nice equation you guys did mention ai uh essentially demand as part of your seriously and i presume that i've been thinking too small because when i I'm learning more about, you know, high bandwidth memory and how people are using photonics inside data centers to get data moved around faster and so forth.

1:17:59But I presume you guys come into play when I have a data center talking to a data center versus inside of said data center. Tailscale gets used in all of these situations, right? So the internet layer does not provide authentication or encryption by default, right? And people, especially in the AI world, are starting to realize, hey, you know what? Everybody's sending some incredibly ridiculously sensitive stuff to these AIs. Maybe we should encrypt it on the way there. Also, AI companies have this pretty much uniformly. They all have this problem. Their GPUs are in one place, which is the cheapest or only place they can get a GPU because GPUs are an extreme shortage.

1:18:38That place is probably not AWS, Azure, or Google Cloud. right the rest of their stuff is probably best hosted in one of the big clouds and so they immediately have what's called a multi-cloud problem right and before ai the general advice is like just don't have multi-cloud problems uh stick with one cloud provider and just be super loyal don't do it it's horrible right just stay away from this but now everybody's got to get a gpu from somewhere else they find themselves in a world with multi-cloud problems and so there is no product that makes your multi-cloud problems not terrible except tail scale which doesn't care about the physical topology of your network.

1:19:14And so AI companies caught on to this very early. We're actually, I remain a little embarrassed about it. We didn't notice until mid last year that most AI companies in the world were using tailscale. We didn't know we like had won the market. There was no mention of AI or machine learning or GPUs anywhere on our website because we were just like busy with other stuff. Right. And then when we finally decided to investigate, it's like, hey, it seems like there's a lot of AI companies that are like showing up in our support ticket to should we look into this we looked into it it's like oh they all love us a lot and they all go to the same conferences they all recommend tailscale to each other of course there's lots of employees that jump from their own AI company to another they'll leave a big company found a new one they bring tailscale with them because tailscale is the networking thing that just solves the problem so they can get back to their AI if you had to pick a next project tomorrow and you had no choice and you couldn't say no you had to do two things at once what's another thing you would want to bring to the internet that would improve the foundation for digital humanity the most for like the next 15, 20, 50 years?

1:20:15So that's a big question. I don't know the most impactful thing that we could add to the internet. I know what I would add. One of my friends, about 20 years ago, very insightful friend, he told me, Avery, there's only three things in computing. There's compute, there's storage, and then there's networking that connects them to each other right there's many different kinds of compute there's many different kinds of storage there's many different kinds of interconnects but those are the only three kinds of things right and if you look at the way the internet works today right um it works the way it does actually because we failed at the networking part it stopped evolving 30 years ago right therefore we ran out of ip addresses we had to create all these firewalls and things now the only things that you can connect are like your device to the cloud right and then you can connect another device to the cloud and you can get data to transfer like by going to the cloud and back but you just can't connect a device to another device the way the original internet was supposed to work because we ran a device and we don't trust people so we had to make firewalls and then we had to have nets the fact that networking stalled meant that how we do compute and storage had to evolve in this like really inefficient direction where now i have to pay rent to amazon for everything right because there's only they have enough ip addresses basically right so now that networking is fixed right thinking ahead a little bit obviously you know tail skills in everyone's hands but let's imagine a world where it is and it's time for everybody to launch a new product right now that networking is fixed you can start thinking how am i going to do storage differently right peer-to-peer storage is an interesting concept that like dies every time someone starts because how do you connect the peers to each other right and they start connecting peers to each other and you realize it's 99 of the work and you don't have time to work on storage anymore right so i would love a world where you can just like you know i've got tons of free disk space on tons of computers i've scattered everywhere but i can't use it i have to go rent it at 10 times the price from amazon instead right or google or apple or whatever yeah aws is a stand-in for any kind of modern hyperscaled cloud infra yeah and then i've got and then you you know, if you've got storage, well, what about compute?

1:22:23Well, I've got lots of extra compute. I joke sometimes about the supercomputer in my pocket, like my phone is, right? And my phone is actually a dumb terminal in the same sense as an IBM dumb terminal back in the 1970s, right? It doesn't, if you unplug it from the cloud, it doesn't do anything anymore, right? That's a really terrible thing to say about a supercomputer. It's like, oh my God, this thing is insanely powerful. It has evolved like hundreds of thousands of times faster in the last few decades. and I can't do anything with it without plugging it into some cloud and paying rent. Why? Because of compute and storage and network, right?

1:22:58So we can solve that triad, but only once we've really truly fixed the networking problem. Once we fix that, a whole bunch of stuff is suddenly going to sort of come unblocked. And I think that's where the magic can come in. Can I ask an idiot question? Because all that sounds great, but I want to make sure that I'm fully tracking you. So essentially, because we ran out of IP addresses, I can't have six different IP addresses for my six computers that I have here and therefore I can't access their storage super easily from anywhere that I want so I have to go to a centralized bucket which is AWS etc so if we could get around our problem then any spare compute, any spare storage with proper encryption and so forth could be used, levered and kind of brought online for general oh that's awesome and if you want to get a little more complicated so the reason I use storage before compute because both of those things benefit from being relocated as soon as you stop using cloud for storage you're going to run into the call so-called durability problem right it's like look my storage is great i have this really big hard drive in my house but what if it dies right it's no good if all my data disappears that's like that is definitely 30 year old technology stuff right if it's in the cloud all my data is not going to disappear unless i delete it or somebody hacks it but if you have a good distributed storage system you can share storage with like your friends hard drives in their houses or even people you don't know and it's encrypted right then when your hard drive dies it still exists somewhere else and this is why it's so important for peer peer to peer devices to be able to talk to each other so you can like spread this content around so that in case the drive dies your system doesn't die the only people who can do that today are the big clouds because they have all this stuff in one place and they're all doing these tricks with multiple hard drives in the background that you don't see but we could do that with our own stuff at 10 10 times cheaper prices with the same exact hardware if only the networking worked properly all right i'm being yelled at by the production team to to wrap it up uh avery i could talk to you all day long this has been such a treat thank you for teaching me a lot of stuff two things before you go what's the url and what is a role you are having a hard time hiring for all right uh let's see well tailscale.com is the url um i'm sure you'll have it posted somewhere in the uh it'll be in the section down below show notes etc yeah excellent and uh i'd say the role that we are of course we just raised a bunch of money uh we're going to be spending that money uh so we are hiring for engineers in particular uh the hardest kind of engineer to hire for is is what i would call the networking platform engineer people who really truly understand way down at layer three and want to make a difference down there and i know those people exist but they are hard to find and there's they're not making more of them so there's this sort of like limited number of these people out there and I would love if those people would apply for Tailscale because we've got some really fun projects that we're working on Avery, a real treat, thank you so much and for everyone else, it's tailscale.com Alright, we'll see you all next time