5,000+ Tech Workers Laid Off This Week. It's Just The Beginning. | E2286

So here's the problem. You have banks, credit unions, hospitals that want to use AI, but they are not interested in sending their data to any of the public AI providers. They say, how do we use AI but locally within our network so none of our data leaves our infrastructure?

You can make your own decision. For me, if you look at the record of some of the executives at some of the companies, it might make one wonder. And if there's only one way to really know if your information is being shared or not, which is to not share it.

25, 30 days ago, And we have over 1,600 orders of this device.

I wouldn't trust Biden or Kamala or Trump or JD Vance with AI regulation for everything. God kings in Washington, D.C. who are on the grift, who are getting paid by these lobbyists. That system's broken.

The executive branch, to your point, executive power control over what AI models I can use terrifies me to no end. Jason, This Week in Startups is brought to you by IMH Health. Start feeling like your best self every day. Go to im8health.com/twist and use the code TWIST to get a free welcome kit, 5 free travel sachets, and 10% off your order. Evry. For all your incorporation, banking, payroll, benefits, accounting, taxes, or other back office administration needs, visit evry.io. And Sentry. Your team should be focused on shipping features, not chasing down bugs. New users can get $240 in free credits when they go to sentry.io/twist and use the code TWIST.

All right, everybody, welcome back to TWIST for May 8th, 2026. I am at my mom and dad's brownstone. Got to see them for the full week. Got to take in 4 Knicks games. Atlanta, New York, New York, and tonight I'll be in Philly, or when you're listening to this, I will have been in Philly. Acquired courtside for 2 out of 4. It's a really hard thing to do these days. But I am really excited for my Knicks and I'm really excited for today's episode.

Yeah, Knicks are up 2-0 over the Sixers, right?

Yeah. And it takes 4 wins. You play 3 series in order to get to the finals. And so we have to win 12 games to get to the finals, obviously. And we're halfway there. 6 of 12. Halfway. Not bad. Still a long way to go. Mission not accomplished. But we really want to get to the finals and then we'll have to strategize from there because we have some great, incredible folks that we're going to be competing against in the final. Either the Spurs or OKC. But let's get to the show. We have a lot of great guests and we have a lot of news, so we're going to try to whip through this docket for everybody. I know everybody's got a lot of comments, and at the end of the show we'll be doing the plugs. For the first time today, we're going to be doing the AI plugs. So listen to the end of the show for the AI plugs. Let's get started. All right.

First up, we have David Moscatelli. He is the founder of a company called Go Abacus. It's a launched-back startup that wants to deploy on-prem AI for regulated industries. Jason, they have a thing called the Go One. It's a big bunch of GPUs in a box, and I heard that they may have another one coming out. But the idea is how do we get AI in the door of companies that can't use Azure, AWS, OpenAI APIs? That's the plan. So, David, welcome to the show. Good to have you here.

Thank you for having me. I'm so excited.

Now, the name of the startup, the domain is goabacus, but You just go by Abacus, yeah?

Yeah, we have an arch nemesis, Jason, so we had to differentiate by adding the go, which leads to our advantage. We say go healthcare, go banking, go finance, go Abacus.

Got it, love it. And so before we get started here, always remember to start your Plaud, record everything, and Alex has his on his wrist right there. You need to get the Plaud Note Pin S, amazing product. If you go to plaud.ai/twist, and use the code TWIST, you're gonna get 10% off. Alex does it all the time because I'm constantly giving him notes on how to be an amazing co-host. It's true. And yeah, just how to get better and better. Thanks. We applaud, applaud for sponsoring independent media like This Week in Startups. David, we met, I guess, is it 2 years now? You were into Launch Accelerator. Which Launch Accelerator class did you go to?

I think it was LA 33, but I don't know, Jason, 'cause every day is like a year for me. So it's one of those.

Exactly. And, maybe just give us an idea of what you're building in as plain English as you can. You know, we always work on that really simple sentence, and we worked with you really hard, uh, in all of our startups to present these startups so well. And one of the great things about, uh, I think our accelerator and Founder University, Alex, if I, I'm gonna give myself a pat on the back here, is that we really help founders communicate what they're doing and why it's important. So you'll see that when David sort of explains it to us versus other folks who come on the podcast and then I have to re-explain it in plain English. Uh, David, what's the mission here? Who are your customers? How do you make money?

Okay, so here's, here's where I disappoint Jason and the entire team. Um, so we as a company, what I like to say is we're AI infrastructure for regulated industries. Now when I say that, people go, what's infrastructure and what do you mean by regulated industries? So here's the problem. You have banks, credit unions, hospitals that want to use AI, but they're not interested in sending their data to any of the public AI providers— OpenAI, Claude. I think they're fantastic companies, but it's just not their wheelhouse. So they say, how do we use AI but locally within our network so none of our data leaves our infrastructure? And how do we box in the cost? In other words, with those guys, you're paying on usage. You know, if you're a bank or a credit union or an insurance company, that's a no-go for you. You don't like a variable annuity every month, you want a fixed price. Okay, so they want to accomplish those two things. And then that's what we do at GoAvocatus is we provide that on-prem infrastructure through hardware and software so that they consume AI in a safe and reliable way.

And you know, this is something, Alex, we've talked about here over and over again when it comes to running local models, running OpenClaw, And how do we think about all this incredible data I have on startups? We do 2,000, you know, we do about 50 to 100 first calls with founders every week. We have all that data, that data stored in Notion, in other places. We have AI provided by, to us from them. Mm-hmm. But we're always thinking, hmm, uh, should this be training, you know, other people's models? And do we even believe them that it's not? Do they even have control over their models? Are these trustworthy companies? You know, listen, you can make your own decision. For me, if you look at the record of some of the executives at some of the companies, it might make one wonder, you know, and the history of the internet writ large, Alex, is privacy is a game people play. And if there's only one way to really know if your information is being shared or not, which is to not share it.

Yes.

That is the only way to know. So I think this is what corporations really want is to have it on-prem, and now you have a hardware solution, a software solution. Let's get into the nitty-gritty, and then Alex, you can tee up your first question.

That's right. So we have a piece of hardware, it's called the Go One. We came out with that about 25, 30 days ago, and we have over 1,600 orders of this device. So pray for me, everybody on the call. Hardware is a completely different business. You know, and we've had all sorts of people call. So we've had health fitness companies call us. We had one guy who does luxury homes for famous celebrities. He wants to buy one of these to put in the box, to put in the house, 'cause celebrities don't want their data for their tech-enabled home going to any of the cloud providers. So there's been a whole bunch of interest that I could have never have anticipated. But I think, you know, our core value prop is we have a piece of hardware, which is the Go One, Everything you need to consume AI is in that device. That device can support up to 2,000 concurrent users at one time. Okay? We ship that device directly to our clients. They unbox that, and in about 15 minutes, they're up and running with on-prem AI. How do we do that? Well, we preconfigure the software and preinstall it on the device. What's the software? We call it Go1 OS. It's a custom-developed operating system that enables the enterprise to have AI and connect to their data. And then that device connects to all the PC terminals of, of each employee within the organization so they can communicate with that device. We replace that device every year to have the most up-to-date chipset. And in the device, there's redundant systems. So there's 3 power supply units, there's duplicate GPUs, duplicate CPUs. We do that to maintain a certain service metric with the device as we deploy it to our clients. They can be chained together. So if you want more more than 2,000 users, you can chain them together at 4, 6, 8, etc.

So you have a service level agreement, you want to make sure you hit that 99.59s, and you're basically mirroring— duplexing is the term when I was a network engineer we used— but suffice it to say, if a power supply or a card goes out, it's going to still be up and running, and you can, you can daisy chain them so you could have unlimited number of users. And I think that means on a security basis, if it's in on-prem You could say, hey, only these IP addresses, only these specific MAC addresses, these specific hardware, these specific network cards can access it, which is a level of security that you really want if you had, you know, Justin Bieber's home network on it. I'm just making up a celebrity's name. Go ahead, Alex.

So the yearly refresh is very interesting to me. What does it cost to buy a Go 1, and then do I have to repay you once a year? Because I presume that newer chips are going to cost more, it's going to depreciate while I own it, and so forth.

It's a good question, Alex. And since, uh, I have some of our investors on the phone, i.e., Jason, uh, I have to be careful how I answer this question. So I replace them every year, uh, because I want them to have the latest and greatest hardware, and I'm trying to avoid the expense of sending out a technician, Alex. So I have redundant systems and I replace the box every year. I don't have to send out the techni— a technician because on the basis of our stress testing, the chances of failure is incredibly low.

Okay.

That's the main reason.

The second reason, it's the worst nightmare of every founder. You've built a product, everything's working great, then real users start flooding in and suddenly it all breaks. What a disaster. You need to get it back up and running and you gotta do that fast. You're looking like an amateur. That's why you need a partner like Sentry. Applications can break in many different ways, but Sentry sees everything. You'll get all the relevant details like stack traces, commits, releases, and even the developers who push that problem code in the first place. With Sentry, you're not gonna be jumping around between different tools trying to figure out what happened. And Sear, Sentry's AI debugging agent, uses all this data and context to identify the root cause of the problem and suggest a fix. It can even take a look at your code before it ships and warn you if any problems are likely. Try Sear and Sentry for free. If you're a This Week in Startups listener, at sentry.io/twist, use the code TWIST for $240 in Sentry credits. Make sure you use that code, make sure you use that URL, sentry.io/twist, so they know your Uncle J Cal sent you.

The chip technology is changing rapidly, and we want to make sure that our clients have the latest and greatest stuff. So one of the only pushbacks to on-prem is, well, as technology evolves, if I have this static thing, how am I keeping pace with what's coming out? And so we have a bunch of ways that we do that, both software and hardware. But on the hardware side, we want them to have the best experience possible, and so we change it out every We send them a new box, they migrate the old box to the new one, they wipe the old box, they put the old box back in the box it came, and they ship it back to us, and then we decommission it.

What am I— so do I pay one main upfront price and then, you know, a different amount every year? Do I pay the same amount every year? Tell me about the price I'm going to pay as a bank here.

Yeah, it's a really good deal. So you have an upfront fee, which is CapEx, right? So it's a capital expenditure. Then you have a monthly service fee. The box refreshes are included in that initial CapEx expenditure.

Got it.

So every year when we refresh it, there's no additional fee. It's included as part of the service. If anything fails, any of those redundant systems fails, we get a notification of a failure and we immediately ship out a new box. Okay. So that's all included in the price.

Now, these yearly refreshes, is that how often I as a customer who has a Go One at my company can expect to have the models that I can access refreshed? Refreshed, or do those update more quickly than the hardware refreshes?

That's a fantastic question. So we— the Go 1 OS includes our own proprietary models. Now I know, Alex, you're thinking, okay, David, you don't have millions of dollars, how have you done that?

We don't have tens of billions of dollars.

There we go. So we actually, as a company, started in 2018, Alex. Back then, this was pre-LLM, it was NLP question-answer matching. And back then we had the thumbs up and thumbs down button. Now some, some dude decided to call that reinforced human learning. We call it thumbs up, thumbs down. So back then people were really optimistic and they spent a lot of time writing very, you know, editorial about why a question was good and why it was bad. So we had 30 million of those queries, right, as data in financial services. That was our base model training. Now our core model we ship out every 6 months, and our core model in the Go 1 OS is a collection of tiny models, Alex. So we call those SLMs, small language models.

Okay.

From 3 billion to 7 billion. And each of those models specializes in a very specific task. So in banking, the things that folks are trying to do are not— they don't change very often. They're very specific. and they're, and they're very, um, you know, deterministic. Okay. So we have a series of specialized models and we have an orchestrator on top. That orchestrator orchestrates to a model, right? And then that model, uh, performs the task. Okay. So that's our model apparatus.

So what I haven't heard you say is we have a large LLM that you can pick between, like, I, I'm thinking because this is on-prem, it's going to be mostly open source, open weight models.

Do you see KIMI? Yeah, what are we talking?

So we can, but that's not our preference. So you absolutely can install a localized model, but we have our own models that we've trained. Now, just to kind of talk about this, you might, again, back to the money, how do we do it? Well, each of our clients has our on-prem system deployed, right? At night, when our clients are not using their system, this is in their SLA, we batch train on their data, and then we take the weights from that training, not the underlying content, just the weights. and it sends to us every night. So we're decentralizing, we call it fractional reserve training. The more clients we have, the more clients that have the Go One, the more training we have in our network. That training then comes back to us. So we're able to spread the cost of training across all of our clients, um, across each industry. And because our models are much smaller, they're faster and they, they're way less expensive to train. OpenAI and Claude, they don't know if you're gonna write a poem or a sonnet.

Right.

So they have to have massive parameters. Does that make sense?

Yeah.

I ask a really nerdy question about this. So for the customer, the unit is CapEx and they pay you a retainer. When you're doing the training across customer hardware, how do you figure that out in accounting terms? Does that— is that a cost of goods sold or is that like them giving you income? I can't figure out where the value is flowing there. It's tricky.

That's a good question. So on our contracts, we have a monthly price and we say, hey, client, we love you. If you agree to share your weights with us, not the data, just the weights, because if, you know, if I put my computer scientist hat on for a second, an LLM is a CSV file with weights and software to run those weights. Hate to demystify it for everyone. That's what it is. So we just want the weights. So we say, if you, if you're willing to share the weights with us, we'll give you 20% off each month for the life of your contract.

Got it. Okay.

Or we'll discount the price of a Go 1. 10% by 50%, right? And so that almost always does the trick. Not every time. We do have clients that say, nope, not doing that, and that's completely fine. But we have a fair number of clients that opt into that.

Do we get the price of the Go 1? Is this like $100,000, $200,000? It sounds like this should be pretty expensive. It's got 8 of those cards. Those cards are $5K each, I think. So there's— I mean, I just off the top of my head, there's $50K in cards, and then there's got to be $25K in power supplies and memory and chips and whatever. So what's the ballpark here, and is that where the margin is, or the margin in the Yeah, what can people expect, you know, out of the gate to be paying a year? Quarter million dollars, half million dollars? Where does this start at? Because this is not a, you know, roll your own kind of situation. This is a high-level SLA. This is, you know, but it's not a $100 million, $10 million install, right? From, you know, a frontier model company that's got a side hustle. Like I had the— I had the founder of Mistral on the pod and they do like on-prem themselves. But I got to think their on-prem starts at $10 million or $20 million. Before they even get out of bed.

Exactly right. So the Go 1, just as a CapEx item, starts at $250,000. The Go 1 Max, which is coming out, which supports 8,000 concurrent users, that starts at $350,000. Now, most clients choose to amortize that over the life of the contract. And if you think about the math, Jason, $250,000 divided by 2,000 users for unlimited AI.

Mm-hmm.

You could ask a million questions, 10 million questions, 100 million questions. It does not matter. The price doesn't change. Your per user cost is pretty reasonable.

Yeah.

You know, you know, you're talking about it's kind of a steal in terms of price.

Hundreds of dollars, low hundreds of dollars per user, maybe. That's right. Is what a pro account, you know, consumer pro account costs, $250 a year, $350 a year, something in that range. And I think this is, Alex, something we learned when we were deploying OpenClaw agents or doing vibe coding is to get the most out of these products and to get the most out of AI, The caps make you concerned. You know, it's kind of like being a chef where you're like, hey, this is— these are white truffles, this is foie gras, this is Toro. You can't get creative because this stuff is, you know, $100 an ounce, $200 an ounce. You're dealing with caviar, so you can't waste it. Whereas, you know, if you were making meatloaf or you're making some, you know, omelet, you can go all day long. Eggs cost, you know, $0.25 each. And, you know, when you're buying them wholesale, so people can have a lot of creativity there. And, and you can experiment. That's, I think, the magic of what GoAbacus is doing here is GoAbacus has a product that gives you unlimited usage. And the, the next blocker is you're like, oh my God, do I want to have access to my data? That's another psychological blocker. So you think about those two blockers. In order to get value from AI, you can't be watching the register and be scared, oh my God, what is this agent gonna cost? And you can't be scared, oh my God, what if like, you know, GoAbacus is— we have GoAbacus's cap table, we have GoAbacus's due diligence on, you know, in a location. We have two-factor authentication, we've got encrypted, we got all this stuff. We had a— we don't want to have a situation where David calls me one day and says, hey, by the way, somebody did a search in, you know, ChatGPT and our cap table came up because they asked what's the cap table of GoAbacus. I know this sounds fantastical. But remember when, what was the first image generation company that broke out and they were indexing Getty Images and when they started making images of things, the Getty Images logo started coming up. It was the— and then Getty Images sued them because they were like, hey, hate to tell you, but not only did you get caught red-handed, not only did you steal our images, you're now doing trademark infringement because your goddamn slop AI It was Stable Diffusion, I believe. Am I correct, Alex?

I believe that's right.

I've been doing this job for decades now. Not a lot of people have worked with more startups than me. Over 600 investments. But honestly, Every.io might be the exception because Every.io is working with every new startup. It's the all-in-one finance, banking, and HR app. They understand startups and they're now at over 1,000 startups, and that means first-time founders working on their pre-seed rounds, but also veteran teams who've raised a Series A or beyond. They can partner with Everi, and they are. And why are they partnering with them? Just to scale up their business, build out their teams, take care of things that are not core to your product or your service. That means banking, payroll, benefits, taxes, compliance— everything you need to run your company is done in one place by one trusted partner. And they're going to give you 3% cash back on all company purchases. Purchases sent directly to your bank account. So it's time to visit Everi.io and get back to focusing on what really matters for your business. Everi.io is going to help you do that. E-V-E-R-Y.io.

Talk about an embarrassing moment, but I do want to say, Jason, right now we're good. So I asked ChatGPT, what is the cap table of GoAbacus? And it says there does not appear to be a fully public capitalization table for GoAbacus. So you're, uh, actually it may have gotten the wrong company now that I look at it, but anyways, So far, so good on that front. Let's do some math though, Jason, because I think you're making a good point about, yeah, about usage and costs. If Garbicus is correct and people do want this because they do want to have, you know, meter-free AI, the money is huge. 1,600 Go 1s at a minimum of $250 apiece is $400 million. So, David, how long is it going to take you to deliver all those units? Because either you just made the new biggest hardware company in the world or it's going to take a couple of quarters to get those shipped.

This is where, you know, I have to talk to our board and, you know, I be, I become the pseudo board therapist. They call me late at night worried and I talk them off the ledge. Um, so hardware is, is—

It's supposed to be the other way around. I just wanna point out that the founder having to talk the VCs and the board off the ledge. Note to the— I may need to get on this board. I, I may, you know, you always have David. I don't do boards anymore because we have too many portfolio companies. You can always push the button. There's the bat signal, and you can say the JCal bat signal. If you need me to come to a board meeting, I will come with Thor's hammer and Captain America shield and level set. If you need me to come, I will come.

I love that. I might take you up on that.

Press the button anytime.

Um, but yeah, they're, they're worried like, well, David, how are you going to ship this? Well, the good news about enterprise is, you know, when they decide to purchase these items, it's not like Okay, we added it to the cart, we hit buy, we put it in our credit card checkout, and it's like, shit, there's like a 3 or 4 month delay. Alignment meetings. IT wants to talk about where it's going to go. How is it going to ship? Like, can we rack it? How do we do that? You know, we have a, you know, a script file that we send out that like collects their network configuration so we can pre-install it. So what am I trying to say, Alex? We have some lead time. That lead time does two things for us. It, it, it removes a lot of the risk around supply chain. You know, we have time to get the parts that we need. And then the second thing is we have time to build, build the thing. You know, now building the thing is a struggle. You know, I'm, I'm a founder, so I love this stuff. I geek out about it. Yes, I'm, I'm there helping build Go Ones. We've purchased these robotic arms that help us do it.

Do you have a picture of your factory? I didn't know you guys were using robotic arms because if you have one handy, I would love to see this. Jason, we do love a factory shot.

One of the issues with hypergrowth startups is the accounts receivable can exceed the operational capital, which then there is two solutions. You can get a venture debt line or just a debt line in general, factoring, whatever. But the better thing to do— and listen, I'm not on the board. I mean, I think we have a board seat, but I think the better thing for us to do in this situation, if I was going to do a little mini you know, powwow here is we should allow people who pay in full to move to the top of the list, and then people put a deposit down, go next. So there's two options. If you really need this, you put a full deposit down. That's one VIP list. That's the VIP list. And then the second is you put a deposit down, we call you. How do I know this is a great strategy? This is what Elon did with the Signature Series. When I bought Signature 16 of the Roadster, I put $150K deposit down when I bought the Model S serial number 1 and serial number 73. I had to put $150K down for each. So he got $450K of, you know, my money into the company to be operating capital. I took the risk, but I got the reward. And that's really what you should do is just tell folks, listen, you know, if you put a full deposit down. Yeah. So because you're taking a deposit of what, 20%, 50%?

It depends. Some clients surprisingly are paying upfront, which is great. A lot of them are capitalizing it. Monthly.

Oh, right, because they want to depreciate. Yeah, so that's right, there's bonus depreciation, uh, thanks to our incredible, amazing, wonderful, extraordinary— insert a couple more compliments here— President Trump. You always got to do that, uh, just to stay in good standing. You know, he did like this, uh, capital depreciation, which was very good for this kind of thing. Yeah, I don't know where it is now. For what is the business depreciation in 2026? Is it 60% you can just write off immediately? It's not 100%, I don't think.

It's nice, I know that.

Yeah, it was like 100%, I think, in the first year, and then I think it goes down 90, 80, 10 or something.

2026 businesses can utilize 100% bonus depreciation for eligible new or used assets, machinery, equipment, software, and services.

Oh, so it's 100%.

Yeah, that's, uh, great.

Awesome. Great for business.

My taxes didn't get— my personal taxes didn't get such a lovely boost. I'm jealous.

Uh, you do have that actually. I do have— you— well, no, you have a, or you have a, you have a corporation. So what that means is if you bought a MacBook or whatever, you could depreciate the whole cost now as opposed to putting it over 5 years. So that would now, you're, look at you, you're like, huh. So if you want to stimulate the economy, you know, independent of what you think of President Trump or Biden or Kamala, whatever, when you do this depreciation, what you get is a massive rush to you know, uh, to build businesses. So a data center or, uh, you know, putting a Go Abacus or two in, people get a little more frisky. They invest a little bit more. They get off the sidelines because if you were gonna pay your taxes anyway, let's say even with your, you know, one-man, two, two-person operation, you know, if you have to pay $50K in taxes this year and I said, well, you could buy $10K of, you know, laptops for everybody and, and a Mac Studio. And you don't have to pay taxes on that, you'd be like, oh wow, it's 35% off. It's 40% off. Yeah.

Yeah.

It's pretty juicy. And it does get people off the sidelines.

Totally in favor of this. And 1,000%. If I hadn't seen the recent news that the US now has debt greater than GDP, I would be— I'd be totally in favor. I want to ask before we let David go about Chicago, because this company is based in really my original technology hub. I have a lot of affection for Chicago. But I haven't been there in a hot minute. So what's the startup scene like? How active are VCs? Did Chicago finally get some later stage investors? Talk to me.

Okay. So, Alex, you know, I tried to ask my parents tactfully once, like, why do we live here? Like, you could have chosen any place, like, but you chose here. And I never got like a straight answer out of them. But I've never really regretted living here because It's just like, it's kind of a work city. People here work, uh, and it has that Midwestern vibe. Um, and so I just love it in Chicago. I, I know that— I know there are other places in the world that are fantastic too, but Chicago is always home to me. In terms of the tech scene, Alex, this is a corny answer, but it's true. I'm working, so I don't know.

Um, like, I don't love it. That's the right answer.

Yeah, I don't go to those things.

What about hiring? I mean, I, I do think there's been a number of companies that have left. Obviously people are going to certain states for tax reasons. For freedom reasons. So I wonder if there's a— I know there's been some big companies there, from Groupon to others. Um, so there, there is talent there. Do you have an in-office culture or remote?

We have an in-office culture. I was remote for a long time, Jason, but then it's just easier when they're right next to you, when you could say, hey man, let's talk about this. Yes, you know, 100%. So we have two offices that we go between, Merchandise Smart and Fulton Market. So it's in-office culture. In terms of the talent, there's a lot of talent out here. There's a ton of talent that we're tapping into. And, you know, I love San Francisco, but when I've tried to recruit from there, I love you, San Francisco. I love everyone from San Francisco. A lot of the folks that we try and hire, like, they want these crazy high salaries and I think they deserve them. But there's like a certain caliber of work that you have to execute on to get that.

Absolutely. Absolutely. Yeah. And I'm like, they're also a bit mercenary. You know, when you've been through this, like, you know, through 3 or 4 cycles, it's just like 4-year vest and the loyalty to the company is second to the next opportunity. So there's a group of folks who just float between, you know, they started at Google, they went to Meta, they did Uber, now they're at Anthropic, whatever. And, you know, if they leave some shares on the table because, you know, they can work Anthropic off of SpaceX, off of OpenAI and say, hey, you got to make up for what, you know, I'm leaving on the table. You know, they're more than willing to do that. So I think it's really smart to be the big fish in a smaller pond and then have continuity. And that's what I found. People love living in Austin. I think of 23 people, I'm down to like maybe 5, including Alex, who are remote. I like to stay active. You know, I'm getting a little older, but I've started drinking IM8's Daily Ultimate Essentials and it is delicious. I'm starting to notice a real shift. I drink it every morning and I just feel less tired, right? Which allows me to really, you know, catch up and attack the day. IM8's Daily Ultimate Essentials packs the benefits of 16 different supplements into one tasty drink. It's loaded with 92 nutrient-rich ingredients, including vitamins, minerals, adaptogens, along with pre-, pro-, and postbiotics. It's giving me a great energy boost. It calms your digestion and it's an easy habit to stick with. Because it tastes delicious. So start feeling like your best self every day with IM8. Go to im8health.com/twist and use the code TWIST to get a free welcome kit, 5 free travel sachets, and 10% off your order. These statements have not been evaluated by the FDA. This product is not intended to diagnose, treat, cure, or prevent any disease. And what I said to people, I think it was 2 years ago, you probably remember, Alex, I just said, if you're an ultra high performer, EHP, if you're an EHP, ultra high performer, you can stay remote. Anybody else has to return to office. And then over time, people who were not in the top— and I said, people said, what's the ultra high performer? I said, for our organization, it's the top one-third because everybody's a high performer. Ultra high is the top third. Alex would be in the top third. You know, Matt, who's our top sales guy, he's in the top third. Heidi, top, you know, hardest working person. So, you know, I make an exception for them. Everybody else, I was just like, we'll see. And now I got everybody in office. It works better. You can actually build a culture. And then, you know, like, hey, you're like, hey, why does Alex get to work from home? Why does Heidi get to work from home? It's like, look at the performance. All right, listen, we got to drop you off, David. Continued success. I hope we've been helpful to you. You went to Founder University, then you went to the accelerator. So you're one of, you know, the, the true success stories in the, you know, J-Cow portfolio collection. I'm so proud of you. I think we also did a syndicate too. So you, you are literally the trifecta of the vision I had, which is podcasts, Founder University, accelerator, syndicate, which is like, hey, over the life of a company, how could we be helpful? Knowledge and community from the podcast, Founder University, community, knowledge, maybe a small check, accelerator. Okay, now we're definitely in business together. We're introducing you to as many great investors It's got my name on it. People might give you a meeting or two because of that. And then finally, syndicate. Okay, we can scale with you, you know, and keep our pro rata, sometimes even go super pro rata. So just absolutely great to be on that trifecta of a journey with you, David. Very just— I mean, we knew you were going to be successful, but this is a level of success, you know, up there with MicroOne and Thalamus and, you know, TaxGPT. We've got a number of breakout companies right now and just really excited.

No pressure, guys.

No, I just, I mean, here's the good news. GoAbacus is hiring, uh, selectively in, uh, Chicago for killers. David at goabacus.co will get you directly to the CEO. If you're extraordinary, if you think you can add value, you know, hey, this is a great rocket ship to get on now.

Thank you, Jason. Thank you, Alex. I, I have to say, I feel like my life is now complete because Jason said he's proud of me. My next goal is to get Mike Cevitos to say he's proud of me.

Oh, good, good luck.

If I could get that, then I would be complete. I feel like that's gonna be much harder.

I mean, I've been working for him since the early '90s, and, uh, he's, he's still held that. I think he might have said it this year. He might have said, I'm proud of you, uh, because I'm, I'm, I've matured just a little bit. Okay, great job, David. We'll drop you off.

I have to do a little, a little interstitial before we bring up—

please.

Okay, so David said that they have offices in the Merchandise Mart, which does not sound Cool. But I want to just make a little, little note here. This is the Chicago Merchandise Mart. If you're on the audio version, imagine a building that is so large. It is the only building, I think, in the US aside from the Pentagon that has its own zip code. And we were talking about— Lord, right, right.

I mean, it looks like something out of a science fiction movie. Who builds a building like that? It must have a giant courtyard, right? So in order to get all the sunlight in there, or is it just that deep?

Merchandise mark from the top. I'll see what I can find here.

Oh yeah, if you do. You know what, I think it might have been a factory building, therefore they didn't. But in today's regulations, you need to have light coming in. So the depth and the footprint of that building is so deep, they must have air chutes. So when they built those in— no, no, it might. Look at this, a cutout there.

There might be a cutout, but it is deep and thick. And what, what it's for I think in Chicago should be putting every single startup in that building. You want to build community?

That would do it.

Density.

Density is it.

You have the building. Just stuff it full of tech companies. Next up on the show, we have José Caldera. He is the founder of Yanaz, which is a BitTensor-backed company of sorts, Jason. Rebranded from Yanaz Compliance if you're on the Googles. And what does it do? Well, it wants to help people determine if a human is present, if a unique human is present, and how to ensure that when an agent is out in the world, the human that gave it permission to go out and do things for it is legitimate. So José, you're solving personhood and uniqueness on the internet. Welcome to the show.

Nice to be here. I appreciate the time, Jason and Alex. Yeah, so it's a thing that we need all to wrestle with. Moving forward, trying to make sure that every participant that you have, everyone that is in this call, that is actually a human and is actually unique. Some challenges that we never thought we would face, but something that we have to deal on a daily basis now.

Who is the customer here? Who needs to figure out identity and needs to do it, you know, through a service like this? There are, I know, some services, Alex. I don't know which they are, but sometimes when I'm doing a transaction online, it asks me to take a picture of myself I think for age verification now. I'm trying to remember what platform recently did this. I know the dating apps do it, so I didn't experience it there, but I did experience it on— oh, I think I may have experienced it when I was doing CLEAR or TSA Pre. It might have been TSA Pre where they took a picture of you. And then I did an age verification really where it— oh, it might have been TikTok. I went— that was what it was. I went live on TikTok recently and it did a scan of my face and it said, I'm just doing an age thing. And it was like, okay, here. You're old, dude, and it let me start live. So who's using this? What are they using it for?

The notion that you need to figure out whether you're dealing with a human, it's everywhere. So if you think about CAPTCHAs, you know, when CAPTCHA was about, it was really about trying to figure out whether the person that you were interacting with was a live human, right? So you would answer a little bit of a challenge, and then by answering that challenge, then you proved yourself that you were there and you were a human. And of course, across the years, that has become a little bit of a nuance and people are looking for alternatives on how to do this. So who is the user? Everyone. Every organization needs, and moving forward is gonna have to make a distinction in between whether you're dealing with a bot or whether you're dealing with an actual human. And that is different in a way from the types of identity verification systems that you were describing before where you actually have to prove in a way that you are who you say you are, right? And I think there is a distinction there because as we move forward, the use of identity data and identity systems to kind of deal with fraud and so on are going to be very challenged because of the growth of deepfake attacks. Deepfake attacks.

What I would take from that as an investor is the service is new. We don't have too many customers right now, or we don't know who the customers are using it. Do we have like known customers using it?

Our, our, our system?

Yeah.

So we, we are in alpha state for this product and we have about 5 projects that are coming live soon.

Got it.

But one of them are DeFi protocols, peer-to-peer moving money, digital wallets, and that, that kind of product, that kind of industry.

Got it. So crypto really wants this kind of solution and they understand the crypto ecosystem, and you would just do this through an API. So I'm using, I have an app, I'm Robinhood, or, you know, I've got an app, it's a dating app, I would just hit it with an API and then it proves that it's a human, correct?

Well, you would have at some point to, the solution is biometric-based. So at some point you have to do a capture of those biometrics and then a challenge to those biometrics. So there has to, you have to do a challenge and a capture. So that means that you are going to, sort of hand it off, if you will, to an application that is gonna do this verification and the capture of that.

And that's the miners on the other side are coming up with algorithms to do that and that's their role?

Ah, close. Jose, hit him.

Yes. So the miner is, so when you are trying to prove that you are human, right? You need to prove that, well, you're not an AI-generated content. You are there. So what the miners are doing for us is generating the synthetic identity identity data that we use to challenge these models. So in a way, our models for detection, our models to prove that you are there, that you're a real participant, are always being attacked by our miners. And that's their function. Their function is to attack our system every day as hard as they can so that we can build better and better and better detection so we can keep the Arm phrase basically going in that way.

Amazing.

No, I think it's really, really cool. And the agentic use case here is why I'm really excited because I would love to be able to delegate to my agent, the fact that I, Alex, the human, have given it responsibilities and permissions to interact with my data and financial information. So talk to me about how delegation comes into this and how quickly you think you're going to be able to serve customers in the AI space.

Yeah, that is a fantastic case, right? Because agent world is growing very fast, very— and is supposed to exponentially grow within the next year, especially in the e-commerce world. So the idea that you can go and say, these are my agents and I'm authorizing these agents to do this type of activities is something that is going to be fundamental on the other side to prove that Jose actually did this transaction, right? So if you go to a merchant and you go and say, hey, I bought my $1,000 computer or my $3,000 computer and it was an agent that did it, but you can actually tie that agent and give authorization to that, that gives the merchant on the other side the tools required to say, I know for a fact that this agent was authorized to do this transaction on behalf of a human. And that's where the connection into the agent world is going to come. And that is just an example within the commerce ecosystem. But this is true for everything, right? This is true for if you want to sign, say, a contract and you have an agent that is signing that contract on your behalf. You want to make sure that that agent has a chain of custody in terms of the authorization. And one of the things that this technology gives you the ability to is to give that chain of non-repudiation in line, right?

Let's show the video, Alex. We have a video. Yeah. And then we'll give another question.

Yeah, we do have a video. This is from our friends over at IANAS and I'm pulling it up right now. Here it is.

Video's worth 10,000 words and you could sportscast it if you want.

Yeah. So basically what you're looking at is a registration concept here is how somebody registers and how the application is, is capturing your biometrics. And then those biometrics are being processed, cryptographically processed inside your device. And then we generate this privacy-preserving set of biokeys that we store in a decentralized registry. And then when that person comes in either to validate itself or try to register again, we can tell whether they already have been registered or not. And that's what the uniqueness is about, right? That you are a unique person in here. So when you are doing things like, you know, voting, when you're doing things like, try to say, okay, I'm gonna give you rewards to this particular human.

Yeah.

You actually have to make sure that that human is unique, that you don't have 2,000 identities attached to the same human.

This is what I want for my x.com account and what I try to do with subscriptions. I have 2,000 subscribers out of my 1.2 million followers. So sometimes when I don't have time to like look at my replies, I just— I'm like, this is going to be a spicy tweet. I'll just let the $1 a month that goes to charity subscribers post. And by the way, some of them are people who are like MAGA, America First trolls, whatever. They hate me and they pay $1 a month to dunk on me. But I'm like, at least I know there's a hurdle here and it can't be there. Nobody has 2,000 of them because that would be all 2,000 of my subscribers. But I would love to say, hey, anybody who's a subscriber, a blue checkmark, and who, when they post their post, it recognizes they're human. And I think that's what's coming next to X, is they're going to do their own version of this proof of human, because social networks are getting filled with AI slop. My replies, I can tell when something's AI slop, and it's so easy to do because you just have a a virtual machine come up, you log in, it looks— it's moving a mouse around, it's reading tweets randomly, and then it replies and comes up with a spicy reply. And then of course they get followers, yada yada. It's very hard for them to detect. But proof of human, man, that's going to really change everything. And that would just be wonderful. I mean, I don't mind doing that. And in Korea, you have to have a Social Security number, the equivalent, in order to join a social network. Now, that might be too much for Americans. We might still want to have the ability to have an anonymous or semi-pseudo-anonymous thing. But, you know, for consumers to have that level of control, only show humans in my feed, I think a lot of people would click it. If I could say on Instagram, TikTok, LinkedIn, etc., only show humans, only show human-written posts, my God, my God, I would love it. So José, continued success. What subnet number are you?

We're subnet 54.

How long have you had that subnet and Maybe explain to people how they can participate in the subnet if they're technical or non-technical and they're just building an app.

So we launched in June of last year. We saw BitTensor as a great opportunity to create this network of decentralized attackers into our models. And that's what lured us into BitTensor. How do you take an advantage of all these scientists across the world to do this? So the current miners is, as you know, it's a permissionless environment. So everybody could be a miner, everybody could be producing data. And we have the incentives, the protocol has the incentives to bring these miners into the protocol. So with that, then we are open to challenge every time, the incentive algorithm. So Bittensor is a network of incentives. So basically the miners are incentivized, we give them money and they try to break our systems. And that's kind of the cycle of it. And this is that ever, I mean, that's the reason why we can be, you know, hardware agnostic, that we can actually get a leg up of the froster because, well, we are inviting all of them into trying to break our systems before we build the protections in the solution that we provide to the market.

Awesome. And the market cap is but $9, $10 million right now. So if people want to understand why I'm excited, about BitTensor. I consider BitTensor my own little Y Combinator launch accelerator, Techstars. There's 128 of these subnets currently. They have to fight to be one of the subnets. They have these tiny market caps. You can participate on the subnet level, you can participate on the TAO level and buy TAO. And we'll see where this winds up. If Jose doesn't build and deliver, you lose the subnet. If he does deliver, the market cap goes up. It's the Wild West, it's permissionless. Disclaimer, disclaimer, disclaimer. But when I see— and now we've had, I think, 20 different subnets, maybe 10 or 20 different subnets here on the program. Stillcore Capital, my partner Mark Jeffrey's working on that. You know, they're investing in the subnets. I've invested in the TAO on the top. I don't know, Alex, if you bought a little TAO on the side or not yet.

Actually, it's on my— it's on my shortlist once I get A little long story. Yeah. No, no. I have an old legacy Citibank account that I can't access that has like $11K in it.

I'm just gonna YOLO, YOLO, uh, 5 dimes in. Yeah. So anyway, I, I think it's like an interesting project. It could go to zero, it could go to the moon, probably winds up somewhere between those two. Jose, thanks for coming on the program.

Thank you, Jose.

Thank you so much for having me.

No, what they really want is rapid-fire Founder University Japan applications are now open for cohort 2.

Oh, well, that's going to be June 7th, uh, when we close the applications. It starts July 21st. I know that because I'll be in Tokyo in July, which is going to be a blast. And I'm really excited because I'm going to bring the fam. You can apply at tokyo.launch.co. The program graduates in October in San Francisco, so the top companies from that program come to America and get to meet a bunch of people here. It's a great program. It's our second one. We do it in partnership with our friends at JETRO. And what else can I tell you? You know, you— if you're— I think in terms of what qualifies you to be in it, you can go to the website. They basically want people who are in Japan or citizens of Japan or based in Japan. Go get the details at the website, tokyo.launch.co. We'll see you there.

It's going to be fantastic. How did Cohort 1 go?

I mean, it was incredible. We had 60 great companies. The top 10 came to America. And, you know, they're incredibly innovative. And something magical happened in Japan over the last decade. It used to be your parents pushed you really hard to go get a safe job. There's a term called a salaryman, a little gender bias there, but a salaryman as a concept means you go work for a company for 30, 40 years, you get your retirement, you're loyal to the company, company's loyal to you. That's every parent's dream. And kids who said, I wanna go work at a startup, the parents would be like, oh my God, that doesn't sound safe. Today, because they have, um, so many jobs available in those, you know, large corporations, young people have said, why not do a startup? Why not join a startup? And the big corporation job will always be there.

Right.

So if I wanna flip to become a salaryman or woman, I have that. As my safety school, but my primary— I really— the, the going for the brass ring is now doing a startup. So that culturally is amazing, which is why I'm investing my time there. If it was 10 or 20 years ago when, you know, it was kind of hard to get people in Japan to think that way, think startup first. The entire planet is going startup first. Why? Layoffs. Tech layoffs. Which I think is a good segue.

Yep. We'll start there. So this week we saw layoffs from Cloudflare. We saw layoffs from Coinbase, and we got some updated news from Block, the fintech company that went through massive layoffs earlier this year. A couple of quick notes here, Jason. Cloudflare is cutting, I think, something like 11% of its staff, a large number. And they said that their AI use has gone up by 600% in the last 3 months. They are trying to re-architect their company for the, quote, "agentic AI era." Coinbase, similar idea. They want smaller teams. They want managers that are ICs as well. They want to have fewer layers. And we're hearing it again and again and again. Every tech company wants to get smaller and faster. Jason, I'm kind of blown away, though, by how many people are getting fired.

Yeah, Coinbase is 14%. And that just happened. That's 700 people. Block was 4,000 people in February. That was 400 people. 40%. Crypto.com, 12%.

Uh, Cloudflare is 20%, not 11%. Sorry, I got my numbers mixed up there.

No problem. Uh, and so these are not insignificant layoffs. And now certain people are gonna wanna say these companies were bloated. Certainly that was true.

Mm-hmm.

And we saw that starting in the post-COVID era, they had run up too many employees and everybody started, uh, the layoffs then. That would be '21, '22, '23 layoffs. Now, when we look at '24, '25, '26 layoffs, a lot of that is people who weren't brave enough or didn't want to take the medicine back then for being bloated. But now it's clearly AI. It's clearly AI-related. Now, are some people using the COVID of AI to make themselves feel better about it?

100%.

Okay. So multiple things can be true. I am not a partisan. I am not trying to defend the Trump administration. I don't have a horse in the race. I'm a moderate independent. You know, you, Alex, you're a little bit left. Some of my other friends are a little bit right. I'm straight down the middle here. I can tell you exactly what's going on. It used to be that deploying people was how you won.

Mm-hmm.

You could use capital and you could use people. Talent and capital were two vectors for beating people. That's now switched. What is the playbook for founders today at scale? It's not the Google playbook and the Facebook playbook of we're just gonna hire everybody smart. Yep. Figure it out later. And it wasn't, we're gonna raise as much money as possible and figure it out later, à la Uber. Now we're moving into a new era where it's infrastructure and AI as a weapon. The people deploying AI the fastest are getting the gains the fastest, which increases their margin, which means they're more profitable, which means their stocks go up, which means they have higher value RSUs to give to the most elite employees. And that's the virtuous cycle. Or you use that capital to build infrastructure that gives you a definitive advantage, like we're seeing the hyperscalers do, whether it's Amazon, whether it's Facebook, whether it's Google, whether it's Elon with Colossus 1 and 2. And now having Cursor and Anthropic as their customers. So let's let that sit for a minute. And what you'll understand is the people who are making AI, the people participating in AI, understand the application of AI 100, 1,000 times better than legacy companies. Therefore, they are the canaries in the coal mine. And Block restated their earnings forecast. Correct, Alex? Explain that piece of this, because I believe that that is the key piece.

As you can say, we are raising our full-year guidance to reflect our strong execution and increased expectations for the remainder of the year. Why does this matter in the context of what Jason just said? Well, the company went through massive layoffs earlier this year, partially predicated around AI, and it said in the same document that As of early April, 100% of Block employees are using AI tools to do their work. Production code changes per engineer are above 2.5x. So the company cut, it refactored how it ran, and now it is seeing results that are both better internally, how it operates with engineers and so forth, and in terms of financial outcomes. So it's kind of a very quick validation, Jason, of the things they were saying in March. And I think a validation of your point about how quickly things are changing. The thing that I want to narrow down on though is if you're a startup, you're already building a relatively small team, right? If you're a Series A company, you don't have 1,000 employees to cut. So what does, in your view, Jason, the perfect or ideal startup setup look like in terms of people, investment, and AI at maybe the seed and Series A level? I want to just make this operational.

Well, if we get to the operational basis, it's Use AI, be AI first, or don't work here. Uh, and so I right now have a company where I would say I'm at 20%. I probably have 4 team members who are AI first, and I am gonna give the edict this year, by the end of the year, AI first, or this isn't the place for you.

Can we, can you define AI first here? Because I, I think we use that almost any bit too—

Any task.

Yeah. Very, very simple. Any task you're doing, you have to ask AI first, how could I do this task better? And then any task that you do over and over again, you have to work with AI to let AI do that task, or at least attempt to.

Yeah.

So frequently, yeah, you see me doing this with the team because you are seeing me behind the scenes and I'll have somebody say, yeah, okay, I don't know how to do that. And I'm like, boom, want to shoot myself. And I'm like, okay, here's what you do. Ask AI, how do I make, you know, social media clips for this podcast more effective? Then the next level of thinking is that's like you learning with the AI. The next level of thinking is here are our last 10 YouTube clips. Please vet them. Create a scoring system for that. Please run that scoring system against this this, you know, Google Sheet or this RSS feed and then send a report every day and take your SkillsMD file and run every Sunday a deep learning of every way that people have talked about that's different than last week on how to make better titles and better thumbnails and just increase the, you know, engagement with your content. And then tell me every week how that skill changed. So you're like Managing the agent and you're managing yourself with AI. I still have people on my team. The majority of my team is not doing this. And the excuse is, well, I'm busy, I'm doing other things and oh, I didn't get to it. If you don't have the time to get to it, you don't have a job. That's what startups are doing now. Yeah, that's what Block has done. Block specifically said 100% of employees are using this. And the real headline is not like, you know, us talking about this. It's the brass tacks. So that's what I was trying to get you to, which is Q1 2026 earnings per share. Block earned $0.85 per share versus $0.68, which was expected, and $0.56 a year ago. That's roughly 26%.

When you say beat guidance or when they raised guidance?

Whatever it is. Oh, earnings are going up. Yes. So just to get to brass tacks here.

Yes.

Their forecast— Block raised its 2026 adjusted EPS guidance from $3.85 a share to about $3.85 a share, which is 62% higher than 2025. Separately, analysts' estimates show earnings expected to grow from roughly $2.22 per share to $3.33, about 50% next year. So what's happening is the— all of this is now dropping to the bottom line. And once one set of founders have now set the bogey, they've said we're going to increase earnings dramatically. You're starting to see the earnings guidance go up and the costs stay the same. That is the key here. And it is a prisoner's dilemma for anybody who wants to be loyal to their lollygagging employees or even their average employees. I know I sound like a, a manic capitalist here. I'm just telling you how people think in Silicon Valley, how people think in corporate America. And how people think on Wall Street. If there is a gain to be had, you have no choice but to take it. That's capitalism. Because if you don't take it, your competitor takes it. Their earnings go up. They can attract the best talent. You can't. They get more investment because people believe in the management team more. And then people stop believing in you. This is the prisoner's dilemma. This is the savageness of it. So how does one counter this if you're an employee who's been laid off? Go ahead, Alex.

Go super deep on AI, build stuff yourself, and show your future employer that you can kick ass and don't need help.

I mean, not to make this personal, but you worked at TechCrunch. Uh, TechCrunch got bought by private equity. It got sold 3 or 4 times in like 18 months. I don't know how many people have owned this thing.

It was brutal.

It's like a hot potato. I'm guessing you're doing plus or minus the same as you were, but you're independent now, correct?

In terms of total work?

Total compensation, your total revenue versus being, you know, having one, one revenue stream with like TechCrunch versus now having multi-revenue streams.

By the end of this year, I will be. Right now I'm still building up towards it. But yeah, I'm going to get there. Yeah.

So let's take a pause there and say, hey, you're seeing this in journalism over and over and over again. The top folks start a Substack, they do a podcast, they collaborate on podcasts like we're doing. And they basically become independent startup founders. The only way you will solve this problem, whether you're a graduate or you've been laid off, is to study your previous employer. Look for what was too small. If you've been laid off by Block, by other people, very simple playbook. Find 2 or 3 people who are also laid off, maybe even 5 or 6, and say, hey, let's have a meeting. Let's get revenge on the person who laid us off. Get a chip on your shoulder. Do a revenge startup. Look at your previous employer and say, what opportunities were they screwing up? What opportunities weren't they taking? What projects would they never get to? What did we learn from running that company that was smart and what was dumb?

Yep.

Brainstorm, brainstorm, brainstorm. Try, you know, making 3 different products. Excuse me. You can leave that into the short. Try making 3 different products and put them in the order in which you think they'll succeed. If you try the first one and it gets traction, keep going. If it doesn't, kill it. Do the second one. If that gets traction, then double down and double down until something breaks out and can cover your 6 people or 3 people co-founders' salaries, and then you're independent of them. And that's where the economy is going. There will be a Cambrian, and it's happening already. There's a Cambrian implo— there's a Cambrian explosion of startups right now and independent contractors. It is the future. The future is independent contractors, independent solo startups, startups of 2 or 3 making low hundreds of thousands to low millions per employee. And you know what? It's like the Wild West. This is like America at its greatest. It's like going west. Go west, young man, and go prospect. Get your 40 acres and a mule, get your, get your gold mine going. You know, try to, you know, sell shovels, do whatever you got to do and get your claim. Go stake your claim, folks. This is the opportunity. You can look at it like a victim or you can look at it as an opportunity. This is the greatest opportunity to stake your claim.

Another thing that I saw, Jason, very recently was that Stripe, which has an Atlas program, which I use kind of as a proxy for people who want to start companies. I used Stripe Atlas to build my AOC. This chart shows how many people are incorporating using Stripe Atlas. Again, single data point, one company's information. Caveat, caveat, caveat. But we are seeing from previously a steady ramp up to a massive step function change in the first quarter of this year. Jason, if this holds across the broader US economy, you're dead on.

That's actually— put the chart up again. It started actually in '25. You started to see a 10-20% increase each quarter in 2025. So if we look at Q1 2025 to Q1 2024, that looks like 10%. And then the next quarter, Q2, it looks like a massive, you know, growth. Now, some of this could be the marketing, as you said, with your disclaimers of Stripe Atlas. They could have put more marketing into it or whatever. They could have lowered their fee. You know, they could be beating their competitors, who knows. But this is, I think, what you'd see across the board from anybody in corporations. It's an undeniable trend. And so Yeah, go get it, folks. All right. Lightning round. Last story.

Anthropic, $50 billion at a $900 billion pre-money valuation. The FT reports that its revenue is about to reach a $45 billion annualized run rate. Expensive, cheap. How would you rate that round?

If they continue adding, you know, a billion or $2 billion or $3 billion every month, you know, we could be sitting here, uh, in 6 months with another $10 $20 billion in revenue. Maybe they end the year at $50 or $60. So, you know, this is going to get them into eventually the Mag 7, you know, territory. Right now they're a fraction of that compared to Amazon, Google, you know, and the people who are doing, you know, $50 billion a quarter, $100 billion a quarter. So, you know, there are some pretty big companies out there, but I think it's fait accompli that they will, if if there is an addition to the Mag 7, they will be it. And then I would say obviously SpaceX will be behind them, and SpaceX plus Tesla will rocket to the number 1 position or number 2 position in the Mag 7, which I think is, you know, fait accompli as well. And so congratulations to them. Uh, I, I'm not buying, uh, shares at that level. I think all of these are going to trade cheaper in a year or two after. So if you were going to buy shares in these companies, Right now I think you're paying well above market and there's risk factors. I think after they IPO, you know, like many great companies, they'll be trading below their IPO price. Uh, you'll be able to get in on them later.

So yeah. Also, here's the thing. Do you know who owns a lot of Anthropic shares? Every single major tech company that you already own in your index funds.

I mean, Amazon owns 20%, right? Right.

And you know what I own a lot of in my index funds? Amazon. I, I feel like people talk about like, oh, we almost had an OpenAI if you weren't an early investor. Not really. I've owned Microsoft shares through my index funds. Since the first time they put a dollar into that company.

You have exposure if you own those, you have exposure if you own Nvidia because they're a big customer. There's a lot of proxies here.

Yeah, absolutely.

Last story.

What's our last story?

All right.

So Whoop, everyone's favorite fitness wearable, is bringing clinicians into their application via video chat. Now, part of this is going to be an add-on cost if you want to talk to these people live. But Jason, I'm curious, what do you think about these doctors on demand?

Yeah, this is a major sea change. Aura, Function, Superpower, and Whoop are all at the forefront of consumer-led healthcare. And while everybody is hand-wringing about the, you know, universal healthcare, the technology industry is coming up in the back of the race and they're rebuilding the stack. ro.co/twist to get your GLPs. Tell them your Uncle J Cal sent you. All of these are leading to consumer-led healthcare. And Whoop now lets you do blood panels in it. They have an AI in it. And now, you know, I want a doctor. I— and listen, I, I can afford to have, uh, you know, like a, uh, you know, high-end concierge doctor. I've had them 2 or 3 times in the past, and those cost generally $20,000 to $50,000 a year. There's some super ones that might be $100,000 a year for a family. But this is going to give people the ability. I'm going to guess it's $300 a consultation, $200 consultation. I don't think it will get covered by insurance. Maybe it does, maybe it doesn't. Who knows? But it doesn't matter because if your copay is $5,000 or whatever your deductible is, this stuff is so cheap comparatively to the cost of health care that I think people are going to start rolling their own. And I can't wait because, listen, if I'm— I want my doctor to start with my Whoop data. I want them to start with my blood panel from Function or Superpower. I want them to start with my Eight Sleep data, but they won't. They generally won't. And then how do you get them the data? You know, inside these products you can take screenshots. I want a Whoop first doctor. I want an Eight Sleep first doctor. I want a Peptide first doctor, a blood test first doctor. Superpower has a concierge that's kind of like a consultant, but not a doctor. Maybe they added one. This is going to change everything. And I've been thinking about states' rights a whole bunch as we've watched this country go, you know, you know, so divided over the years. I think the state that provides universal health care— I think we should get rid of health care from the federal government. The federal government should give some amount of money to each state to deploy as they wish, which I think is kind of how it works already.

Block grants and so forth.

Yeah. I think we should break this up state by state. And if a state wants to attract people, let's say you're Idaho and, you know, you got a big push to get people to go there and you're trying to, you know, get people to, to move there. You could create universal health care or you could say, hey, everybody who lives here gets a blood panel, gets a whoop, gets a GLP, gets $1,000 in, you know, these kind of services as part of being a citizen here and paying your taxes. We need to have 50 different competitive states going at it, just like we have 50 different states fighting it out for data centers, for, you know, different headquarters of companies, tax breaks, etc. The federalization of this country and executive power is killing this country. The, the absolute ridiculous amount of debt we're going into is killing this country. The only way to fight it is to fight against federalism and to let it be 50 different experiments going on. I'm pretty sure you agree with me on this, Alex.

Well, I just want to understand it because you're talking about the federal government spending money to give it to the states, let them run their own experiments. So to me, it still feels a little federal.

Yeah. Or you deprecate that over some period of time and you let them collect it themselves. So your federal tax goes down, the federal government gets smaller. They, you know, like the Department of Education on a federal level, It's not working. You know, the health care at a federal level, it's not working. It's too big. It's too unruly. And then what you have is the states gaming the federal government and corruption at all levels, from federal down to state cities. We're seeing it everywhere. It's not just a blue state thing. It's everywhere. In Texas, we had a bunch of, you know, I know the Republicans are doing a really good job of exposing California, Minnesota, but we have a bunch of this fraud in Texas as well. Florida too. Florida has it. I mean, it might not be as acute. Putting it all aside, there's too much fraud. There's too much gaming of the system. If it's a distributed system, it's gonna be easier to control cuz you have local control. The local politicians are gonna fight harder and they're gonna be closer to their constituents who are gonna knock 'em out of their position in terms of local governance if taxes keep going up. Up and they'll be fighting to bring them down. Too much federalism. The federal government has to get smaller, and then we have to give the regions the ability, the states the ability to fight and innovate, whether it's on housing, healthcare, or education. Thanks for coming to my TED Talk.

You heard it here first. Jason Calacanis comes out and endorses state-level AI regulation.

I mean, yes. I mean, thank you for pointing that out. I do think I know there are things, and, you know, David Sacks and I have a big difference here, I wouldn't trust Biden or Kamala or Trump or J.D. Vance with AI regulation for everything. We have poker regulation by state. Is it a bummer that you can't go to one poker site and everybody's on? Yes. But each state getting to decide gives them some, you know, gives us some ability to test it. Cannabis regulation, gambling, gay marriage. Listen, I know that we all have strong feelings about these, but Being able to pick the state you wanna live in across 50 different ones to the one that matches your particular, uh, view of how the world should be. That's beautiful. That's great. That's called choice. Yeah. And so if we broke apart the AI regulation, um, yeah, I don't think we need to regulate models on a city-by-city or state-by-state basis, but for self-driving, Sure. Why not? Why not have states have the right to decide what they want to do for employment issues, licensing of taxis, whatever? I know it's unfair. Listen, I'm a big shareholder in a lot of these companies doing ridesharing and full self-driving. But if, you know, Massachusetts and New York and California and Texas have different views of this, I went through this already with Uber. Texas, you know, I'm sorry, Nevada took a long time to get Uber. It's true. People fought it in different states. Okay, we have a lot of differences. Better to have that than to have some god kings in Washington, D.C. who are on the grift, who are getting paid by these lobbyists. That system's broken. We know that system's broken. So why not let the states have at it? I just— the older I get and the more I see the corruption, the more I see the executive power getting abused, the more I want more independence and more state sovereignty.

I wrote about this the other day and I completely agree with you.

Cautious optimism.

Yeah, well, I just— there's a lot of talk right now about AI regulation. I didn't bring it up just to be slightly annoying, but more to the point that the White House was floating some ideas about kind of having pre-release rights to AI models and then people freaked out and they're kind of, they're walking that back. But even that hint of giving, not Congress, because I think that would be slightly different, the executive branch, to your point about executive power, control over what AI models I can use, Terrifies me to no end, Jason. No interest in that. Pass.

Thank you. Absolutely. Yeah, we just— not where we want to live. Not where we want to go. Not where we want to go.

Long live capitalism. Long live democracy and federalism because that gives power to the states.

All right. And we'll see you next time Monday. That's twist. Bye-bye.