VC experts on why Physical AI funding is heating up | E2333

Physical AI companies now have raised almost $50 billion in venture funding. Wait for it. In the first half of 2026, up 80% from 2025. 6 months of this year outraised 2022, 2023, and 2024 combined.

Jensen very famously predicted that every industrial company will become a robotics company. 10 years ago, the smartest engineer I met, they wanted to work on ads. Today, that engineer, she wants to build rocket engines in Moses Lake, Washington.

They just told us, don't do hardware. It'll, they'll run outta money. There's no margin. The Chinese will just copy it. 450 deals and $45 billion in physical AI in the first half of '26. Back in '23, it was not even 5.

This Week in Startups is brought to you by Odoo, the all-in-one business platform. Your first app is free. Get started today at odo.com/twist.

Lightfield.

Name one person who's ever enjoyed updating a CRM. Exactly. Lightfield's AI agent does it for you. It even prospects and books your meetings. Used by thousands of startups. Free at lightfield.app. 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 This Week in Startups. It's our VC roundtable. We call it This Week in VC. Uh, I'm getting very creative with the names around here. We got This Week in Startups, uh, in year 17, This Week in Venture Capital, This Week in AI. New offerings, trying to tighten those up with roundtables with great guests who are actually doing those jobs. Something happened in media. I started my career in zines and magazines, and then I did Weblogs Inc., and then I did podcasting. And along that journey, it was 100% journalists. Then in blogging, it became like half journalists were blogging and half actually domain expertise. And now, uh, in this latest third era of media, The audience, uh, there's no dig to journalists. They don't wanna hear from journalists. They wanna hear from people in the field actually doing the job, and they wanna hear from them directly. So, hey, uh, that's where we are today, and we are super lucky to have 2 great practitioners of venture capital. Paige Doherty is back on the program. She's the founding partner of Behind Genius Ventures, BGV.

Yeah.

Started her fund at 22 years old. Yeah, she's 26 now. She's getting very old. Um, you are, uh, Gen Z. You are a Gen Z investor. Uh, you're still on your first one or you're on your second, right?

Uh, we're currently investing out of our third fund.

Third fund. Look at you. Yeah. By the way, making it to fourth fund means like, that's when it gets real. You know why?

Yeah. 3, 4 more years. It's going to be real.

You know why it gets very real?

Why?

You have a track record. Oh boy, now you've got to explain the first 3 funds. This one did great. This one was during peak ZIRP, uh, and this new one is very promising. And here's our latest offering. But you've got a lot of track record. You're in year 6 or 7 of investing?

What were you? Uh, 5. Yeah.

Year 5.

Wow.

So moving, moving along nicely. Uh, any particular investments that the audience might know, or That you would want to share with them that are particularly doing well?

Yeah, I would say Knox Metals is one of the companies in our portfolio that's growing the fastest. They are a next-day metal servicing platform. The founder is Zane Hengsberger. He's a great Twitter follow.

Knox Metals?

Yes.

Knox Metals. Spell that for me.

Oh, there it is.

Yeah, that's it.

That's it. Knox Metals. NOx Metals, metal at the speed of software. So explain to us how that works.

Oh, I don't have my aluminum block on the desk. I usually do. So basically NOx is an intermediary between the steel mills and then downstream manufacturers in aerospace, defense, um, and other— yeah, and manufacturing companies. And it's a $200 billion market, um, and There's less than 40 software engineers currently employed in the industry. So they are modernizing this, yeah, incredible sector.

So yeah, if I need a brick of some super strong aluminum, I can go grab it and have a brick and put it on my desk.

Yes, you can.

Incredible. So this is like real world. This is atoms. And isn't that interesting? Atoms now taking over for bits in our industry. Also with us today, Caitlin Holloway. She is, a founding partner at 776. Now, this is interesting. I have a friend, Alexis Ohanian, who's come on this show no less than 6 or 7 times. His fund is also called 776.

Well, fancy that. Alexis and I, uh, happen to be partners there.

We, uh, there it is.

We started this firm 6 years ago together, but this is our 12th year working together. This is our 3rd rodeo. Yeah.

I love it. I love it. Uh, and 776 stands for something very specific. Maybe you could share that with the audience.

Of course. So, 776 BCE is actually the year of the very first Olympics. And Alexis, and when we were trying to think of how we would name the firm, naming a firm is as hard as naming a child. You belabor it literally and figuratively, and then it happens, and then suddenly it just always has been, and it's not a big deal. But the short version of the story is Alexis's oldest daughter's name is Olympia, and the first Olympics were held in Olympia, Greece. And so it's a nod to his family without being too on the nose in the event that he had a second child. And he did.

And he did. Yes. And, uh, his, uh, his, his new late-stage growth fund will be inspired by the second daughter. We're both, uh, girl dads. We both spoke at something recently. Where was it? Oh, PayPal. PayPal had their big annual conference, and I, I got to interview Alexis. Um, just really great human being. Um, a girl dad. And, uh, inspired by the Greeks. I'm not saying inspired by this Greek, but yeah, inspired by the Greeks. I like it. Uh, I can get behind it. And so, uh, interesting notable investments from the firm 776 that you'd like to share or that people might know or you're particularly excited about?

Sure. So, uh, we are generalists, uh, by trade and by practice. Um, and so you'll see everything in our portfolio ranging from, uh, things like, um, uh, Beast Industries. So MrBeast's companies, including Feastables. That was a check out of Fund 1. Fund 1 also housed Stoke Space, reusable rockets, and Angel City, the NWSL's very first expansion team out in LA. So that's, that was kind of the early, earlier vintages. And then if you fast forward to today, you know, I think we're going pretty hard at several of those things still. So you talk about atoms over bits. And, uh, that happens to be my, my area of interest and expertise these days, all things deep tech. Um, we, uh, recently had a very lovely, um, uh, markup in a company called StarCloud Orbital, uh, Data Centers.

Yes.

Uh, so I was very, very lucky to, to come in and, and, um, co-lead that seed, um, in the earlier days. And then, uh, they, they just got a, a beautiful new round, uh, markup at $2.3 billion, which—

Yes, on the docket.

Yes.

Yeah.

So we've got Star Cloud. We've got a great number of companies that are really doing things, taking things offline. Like I said, I really care deeply about deep tech and space tech. My joke is that now everyone's rushing to space tech. But the reason I'm qualified is because I worked at Pixar once upon a time on the film WALL-E for many, many years.

That would make you overqualified to work in space. WALL-E. I was just talking about how there was a— I guess Claude or Anthropic was— there was this rumor they were talking about them being the last company. And I believe the premise of WALL-E was there was like a final company that was a—

Buy n' Large, baby.

Buy n' Large, which was like supposed to be like a send-up, I guess, to Walmart and overconsumption. What a great movie that was, man. I gotta watch that again. And hey, before we get started here, Always a good idea for me to turn on my Plod. See this, this little pin I got here? I wear this and then I record when I'm doing stuff, and then it feeds into my AI. I get a nice summary. I can get it at the end of the day. I can get it, you know, after a meeting. A nice little signal here, a little light telling people that I am recording. I have it on my watch, uh, or I can put it here on my lapel. And if you, uh, basically talk for a living— conversations, interviews, meetings, calls— Or you're just taking notes. Hey, to-do list, remind me of this, remind me of that. I have an idea for a guest on the podcast. I got an idea for a topic. I can just talk and then go for a walk on the ranch. Boom, I take my Plaud with me. Go to plaud.ai/twist, plaud.ai/twist, and use the code TWIST for 10% off. Thanks to Plaud for supporting independent media like This Week in Startups and This Week in venture capital. All right, let's get to our first story. First story, Pentagon has rolled out its own ChatGPT and Grok for the military. This is interesting. Yesterday, Department of War— I prefer when we call it the Defense Department. War is a little too aggressive for me. I don't think we should be seeking out war, but okay. Finished an 8-month rollout of special— of a special version of Grok and ChatGPT designed for military personnel. New tools include ChatGPT MIL, which handles paperwork, logistics, and policy work. Grok for Government, Shandell Supply Chain Analysis. Neither vendor trains models on Pentagon or DOW, Department of War data. Claude is missing entirely from the platform. You remember there was a big beef between Anthropic and Emil Michaels and the Department of War because they wouldn't reduce their rules against using Claude's technology to do certain things around war. Um, so Anthropic is left out of the party, but Paige, uh, we're starting to see the military, uh, industrial complex, defense tech, and AI come together in a major way. Uh, your thoughts on the space, especially from an investment category? Here we are, the Pentagon is talking about deploying AI for their service members, and they're being quite vocal about it, and they're engaged deeply with the AI industry here in America. It's the worst The 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 pushed 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 Cyr, 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 Seer 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.

Yeah, absolutely. Uh, this hasn't been an area of our deep investment, but I will say I spent 3 years at Northrop Grumman when I was in college, and I do think one of the, one of the opportunities is deploying this great technology within a wall because there are certain constraints that you need to Yeah, basically taken into consideration when deploying civilian tools, uh, within the defense and prime areas.

So I'd say you gotta definitely, uh, Caitlin, keep this sovereign, contained, uh, and be careful with it. What are your thoughts on just broadly deep tech and AI, ATOMS, and this new military approach? Hey, we're going to be AI first, we're going to be drone first, we're going to lean into this technology, because we spent the last 20 or 30 years with the technology industry essentially not engaged, not engaged at all with the military.

Totally. Well, I mean, I remember back when I was a young, uh, angel investor trying to, to, you know, earn my stripes and, uh, cut my teeth in this, in this wild world of venture. And, uh, you know, it was, it was a known fact. It was like, if you were selling into the government, like, run. this is not an early-stage, you know, bet that you should be making. And so it used to be all about the time of the contract, right? Like selling enterprise software to the government, for example, was like just absolutely like a death trap. And so we were told, you know, don't touch it with a 10-foot pole. And so I think what's really changed for the better is that now it's not about getting the contract, right? It's— we have new challenges with that. I think now you have to deliver against that contract when you're a 30-person company instead of, you know, an agency who is used to working with a 100-plus person vendor. So yeah, I think that the bottleneck used to be winning the contract, but now it's really just surviving one, which is interesting. And so as these companies have signed these contracts within the federal government, I think they're very optimistic about delivering against them. But I think that the challenge is they kind of surprise themselves by getting in. And now they're like, okay, whoa, this is a lot of capital now that's coming in, you know, and oftentimes it's either non-dilutive or revenue. And so they're trying to sort out very quickly what their seat is, and then also like, is it gonna last, right? And so I think that's, that's the next challenge that a lot of these founders and these portfolio companies really have to struggle through here is, is can they last and will these contracts be renewed? Is this revenue we can count on?

Yeah, and the interesting thing about this approach is they're being very vocal about it, and they are incorporating multiple members. People may not remember, but at this portal, which is for the 3 million folks who work in the military— that's what this portal's for— they already had Gemini. They had some sort of a deal with Gemini. Now they're adding 2 more, ChatGPT and Grok, for government. ChatGPT is for administration, document management, logistics, policy work. On unclassified data, chats, files, projects, custom GPTs, etc. Grok for government via Starshield is from obviously Grok and SpaceX AI. Faster ops analysis, supply chain and acquisition work, quote unquote playbooks and adaptive reasoning models. Obviously, that might be for maybe some more forward-looking. And then all of this technology is being incorporated into weapons. And that's where it's going to get super interesting. I think we could have much less collateral damage, and you could also have, uh, Paige, a lot of other issues when you trust the computer to do certain things.

Yeah, yeah, absolutely.

Period. Yes, absolutely. Period. I like how you're definitive. Yes, J-Cal, be careful. I, I think that's really where this is going to get interesting. I was talking to the head of Boston Robotics when I was in Paris for this conference. It was a robotics conference. And she said— she was the interim CEO— under no circumstances should we trust any robotics out of China, should not be allowed in the US. It's an important national security issue. And then I kind of pressed her. I was like, do you think the Chinese are weaponizing robots? And Caitlin, she said to me, yes, of course. And then I said, but you're not weaponizing robots? She said, of course not. And I said, wait a second. This is incongruous to me. Our biggest adversary is putting guns literally onto that dog-style robot, you know, the 4-legged robots. They're literally putting guns onto it. It's pretty obvious that these are— and obviously drones are robots, and they're being sent in with payloads, sometimes human-controlled, but I think often AI, or increasingly AI. The US should be thinking about this possibility. They must be, uh, doing some basic research on creating autonomous soldiers. Yeah, Caitlin?

I would find it very hard to imagine that they are not.

Yeah, it's going to be a very interesting moment in time when instead of a war being fought with the drones, fixed-wing or, you know, quadcopters, I think the next war we're going to just see 100 robots show up on a beach somewhere to try to secure some area. This is going to be Yeah, different. I'll just say super different. Yeah.

I think, I think there have been a lot of movies about this, um, and I would say 99.9% of them don't end well, uh, for humanity, um, at large. Um, so there's the, the techno-optimist perspective, which is I have to believe and I want to trust that we have very smart people, uh, that are, are you know, not just thoughtful about the products that they're building and that they have the capacity to build it, right? You can build, if you can do it, you should, but should you? And so I think that the conversation around like, just because you can, do you really take that into your own consideration? And so like, where does morality play into this? Where does the thoughtfulness and empathy and compassion come in? What are, you know, I think that there's probably a much bigger conversation to be had around, you know, does humanity strike an accord around this? You know, we have the Artemis Accords for space where we're talking about, you know, we We need to start thinking about humanity as one versus these individual dots on a map. And there are no borders when it comes to space. And if we really sincerely believe we're going to become interstellar beings, we have to start thinking about this from perspective of survival as a collective. And there's a really, really, really big gap between that logic and that line of thinking and that philosophy and what's actually happening today on the ground in real life. You know, it are, are backflipping, you know, robot dogs shooting laser beams and fireballs. Like the in-between, we just have to learn the lesson the hard way. I hope not.

Hey everybody, I'm here with Heath Paris. He is the founder of Lightfield. They are an AI-native CRM and it builds and it updates itself in real time based on whatever's in your email, your calendar, your Slack, your meeting notes, you know, all the most important stuff that keeps getting lost.

Uh, great to be here.

Keith, let me ask you one question about Lightfield. I have a list of customers, but I also want to know, based on those customers, hey, am I doing a good job? Am I not doing a good job? And hey, if somebody churned or they might churn, how can I not lose them? Because the best customer that you can get next is not losing one of your existing customers. Tell me about saves. Yeah, so we've built an elaborate skill and automation playbook into Lightfield using some folks we've hired from Gong and Snowflake and— even Salesforce, and you can use any of those playbooks to do an account health check, account health score, go look at their product usage, their customer tickets, the way they respond to emails, the way that they spoke to you on your last call, and it'll try to help you triangulate, hey, do I need to save this? And if I need to save this, what do I do? All right, everybody go check out lifebuild.app. It's perfect for you if you hate your CRM system. And I know you do. Everybody hates their CRM system, right, Keith? I haven't met anyone that has it. Yeah, exactly. You know, the UN is actually working on this. It's called LAWS, Lethal Autonomous Weapon Systems. So there are going to be broad treaties around this, and I guess we're going to try to get people to some consensus around, um, yeah, just maybe we can all agree to not create killer robots. I mean, it's a really interesting moral question. Um, and, uh, one of the things, uh, just do a little pause for the cause here as we shift. When we're doing research for our venture capital firm and for the show, we use Harmonic. Here's Harmonic. And producer Jacob asked it, hey— and Harmonic's the database of every deal ever done, uh, and all these venture deals— what's going on with this military Pentagon rollout? Gives a nice summary of it, tells you all the companies involved in it, how much they've read, then gives you an analysis. Hey, here's the foundation model providers, and you can get links to every single one of these companies, a really rich detailed profile. And we asked for examples of who's getting into defense, and it gives us all these breakout companies, Series A, Series B, etc. And we can then book meetings with them, and you can take all of this. Data, match it with your calendar, phone book, and match it with your LinkedIn and figure out how you know people at these various companies. Uh, you too should take a look at Harmonic. Harmonic, great product that we use here to do deep research both on our investment side and also with, uh, the show. The EU just decided that ChatGPT is legally a search engine. On Monday, the EU formally designated ChatGPT as a search engine. This is part of the Digital Services Act, which most people I think are not aware of. But ChatGPT is officially a VLOSE, a very large online search engine. This is a designation they give to Google, Bing, and it requires you be audited and that you are giving access to your data to the government in the EU so they can audit it. If you don't pass annual audits, publish risk assessments, and allow regulators to access your data, you can get fines. The fines are a little more than a speeding ticket. If you don't do the audit, you can have up to 6% of your global revenue fined. That's a big number. That would be, gosh, like billions of dollars for OpenAI and Anthropic. What are your thoughts on the EU regulations in the face of AI, they are going to be a much tougher filter. Yeah, Caitlin?

Yeah, I think so. I, you know, it's funny because there's definitely a lot of talk about Europe right now. And there's, I don't know if you've landed on that part of, uh, TikTok or Instagram Reels yet where, uh, the, the meme of VCs, US VCs just hating on Europe, like super hard and European founders in particular because of all of this regulation. Yeah.

And they often do that, by the way, when they're at their, um, residences in Tuscany or in Saint-Rémy in the south of France. The VCs are quite vocal when they're on vacation in the south of France.

Oh, the irony in all of it. But I think, you know, for, from my perspective as a seed investor, um, this doesn't, this doesn't really impact our portfolio and our founders just yet. Um, it's definitely something to be watched and I, and I do, I watch it very carefully, but I think for, you know, a seed company, the, the, this designation for ChatGPT, um, is almost zero because nobody at 15 people big is getting designated anything, right? And so, like I said, I, I watched the precedent for sure. I think, you know, it, the EU is, is trying to figure out how it feels about a, a lot of things, but so is the rest of the world. And I think America in particular is also trying to figure out how it feels about a lot of these things, right? And so, you know, I think watching what the EU does because they're typically out far ahead of the pack in terms of regulation. And I don't mean that positively or negatively, it's just a fact. But I think that definitions like this really travel and then they evolve from whoever, you know, threw the first spear in the sand. And so I think, you know, the part for me right now that does reach my founders really isn't the law or the regulation, but it will show up in a few years in like a questionnaire, a compliance questionnaire around like an enterprise sales cycle or something like that. And so again, it's not this exactly right now thing for my particular portfolio, but I'll take a contrarian take on the EU in general, or actually Europe as a larger opportunity. Like I said, I think GPs love hating on Europe right now, but I think if you can really squint and see past that upset about rules for half a second and I should let you know, like, I was an HR lady. I was an, I was a senior exec running, uh, people and culture teams for a long time. So like, I actually look at rules and regulation and probably through a much different lens than most. But I think that it's a real opportunity and I don't, you know, I don't wanna give away the secret sauce here, but like, I really think it's an opportunity because what's happening in Europe is really incredible.

Mm-hmm.

I think Switzerland in particular, like Zurich is producing some of the best physics and quantum work. that the world and humanity has ever seen. And I think it's just super dramatically undercapitalized because of these memes, right? Because of all this narrative that we've created around regulation. We're so scared of rules that I think we're really missing a lot of upside and opportunity. And so, you know, I know Switzerland isn't a part of the EU, but, you know, I really think that half this conversation doesn't apply to them. But I really do think that, you know, if I had $100, I would much rather back a brilliant quant team in Zurich at the same, you know, at like a normal rational price than like the 40th Agent rapper, you know, ripping here in Selma.

Yeah.

And Reddit and Roblox also got added to some of this regulation page. Um, lest people be super cynical about this, some of the stuff is common sense regulations that also parallel what's happening in America. So if regulation in Europe— chance to dunk to Caitlin, your point— but Harm to minors, uh, fundamental rights like privacy, um, and free expression, the electoral process and making sure that that's safe, illegal content being disseminated. Obviously, sometimes people share things they're not supposed to that are very harmful. I won't mention those specific things because I don't want this video to get dinged by YouTube. But, um, what are your thoughts on startups in Europe? Do you get many pitching you? Do you look to the country as a place to find founders? Or is the typical scenario when you do meet with a European founder that they've moved to the US because they want to move a little bit faster with a little less regulation and a lot more capital?

Yeah, it's a great question. In terms of where we invest, we mainly focus on the United States and Canada, but we have backed a few entrepreneurs who have come over from Europe to build businesses. I think an organization that's doing an amazing job of bringing folks over from the EU is Entrepreneur First. They opened up a huge hub in San Francisco and the talent of the entrepreneurs that has been coming out of there has been really strong and exciting to see. I think to go back to your point on the regulation in the EU, some of this, some of the things that they're regulating, like you said, are common sense. I think that It will be, it will present an opportunity for startups who are willing to go through that level of compliance and oversight to capture market share. This is what we've seen in one of our companies, specifically a company called Intermezzo. This was actually one of the opportunities that they saw within AI because it is so, software is so heavily regulated in the EU. They're like, okay, let's The phrase they use is like, eat the frog. Like, let's do the hardest thing first. And if we can get our AI engine to work in this really complicated regulatory environment, then it'll be like amazing when we transition it over to a less heavily regulated space. So I do think there is an opportunity for companies who are willing to eat the frog like that.

Uh, you do, um, create some roadblocks and some expense if you have to hire 10 lawyers and spend $2 million managing all this regulation, then the upstarts that are trying to compete there that only raised $10 million, now 20% of their capital, uh, has to go to obviously, uh, some lawyers, etc. I do think they have one thing right, um, which is they designate this kind of, uh, oversight to things that are very big so that they don't make it oppressive for small ones. So it is incredibly Funny to make fun of the name. Like, it's a very large search engine. Wow. Very creative. Yes. You know, they, they could have named it like at-scale online services or something that would be a little less easy to mock.

But—

But I kind of like it. It's like on the nose, you know? You're like, yeah, that makes sense. It's not like sheathed in some words where you're like, what does that mean? And you get down to the bottom of it and you're like, oh, a very large online search engine.

In the AGI future. Maybe you'll just be able to ask your personal agent any question and get immediate, perfect answers. But today you still need simple, time-saving tools to send invoices, run your website, and keep track of all the data you need to understand your business. That's why here at Twist, we recommend Odoo, O-D-O-O, the all-in-one management software that's already being used by 16 million users across more than 170,000 companies. Odoo is bringing everything you need into one platform. That means your CRM, sales, accounting, manufacturing, website, inventory, and point of sale all right where you need them. And these tools are all in constant communication with one another. So no more logging every change across 3 different apps and spreadsheets. Make a sale, an invoice gets created while your inventory gets the same update. So easy, so simple. So if you're still cobbling together your back office across 5 different apps or more, get started today at odo.com/twist and your first app is free. That's odo.com/twist. One of our guests is pretty excited about this next news story. StarCloud has raised $250 million at a $2.3 billion valuation for the crazy, insane vision of AI data centers Uh, StarCloud puts GPUs on satellites, runs them on solar power, and sends the signal back down to Earth. Looks like a— the round was led by somebody called Manhattan West. Unaware of that firm, but Nvidia and Cisco are both coming in as investors. StarCloud put— StarCloud put the first H100 chip into orbit last year, and Nvidia is now using StarCloud flight data to design its first chip built specifically for space. It more than doubles StarCloud's valuation from 5 months ago. Caitlin, this is your baby. Quite nice to get a markup of 2x in 5 months. We're living in interesting times. But, uh, how do you think about when you first invested? What was the ballpark valuation? What was the round?

Yeah, so, um, first credit to Morgan Beller, uh, who was at at NFX at the time, who really got in very early. So StarCloud was a YC company. And truth be told, when I met Philip, the founder, it was not under a traditional, like, pitch setting. He and I were accidentally seated next to one another at dinner. Both of us were, like, wildly and inappropriately overdressed, and were also probably very socially disinterested, or incapable, maybe is the better word.

Yeah.

And so I kind of looked at him, him, you know, as they were passing around appetizers, and I was like, oh man, I have social anxiety disorder and I really don't want to be here. And he was like, yeah, same. So we started talking about everything, um, literally everything, uh, from life, um, to his brother getting married, uh, that following weekend, uh, and some of the challenges there, to, um, extraterrestrial life and Carl Sagan and philosophy. And so we talked all night long, and kind of by the time dessert was delivered, I looked at him and I said, now what do you do again? And he was like, oh, it's this, you know, wild idea that we can have, you know, off-planet data centers, orbital data centers. And I was like, data centers in space, interesting. And he goes, what do you do again? And I was like, I'm a space tech investor.

Yeah.

And he was like, oh my God, I should have given you a much better pitch. And the thing, the thing I love so much about, about StarCloud is, is not just the, the obvious success that they're having right now and the lovely markup, which is always great. I really, I like, I truly fell in love with Philip and the whole StarCloud team. His, his co-founders are also phenomenal. But I, I, what I was in love with was there was an incredible audacity to what it was they were doing and, and a fearlessness that came with that audacity. But having that, that dinner with Philip and then we proceeded to spend the, the, the weekend together at this event and—

Oh, just tell us the event already. Were you, were you, was this the, the lobby? Was this TED? Where were you? Okay, you can say.

We were, we were, um, it was actually Founders Forum, uh, out in the Cotswolds. Uh, so we were out in like the countryside of, uh—

Yes, I know it well. I've been invited for 15 years and just haven't been able to make it. Yeah, but, uh, so it's, it's a pretty well-loved event. Yeah.

It was, and I, I honestly, it was my first time attending. It was Philip's first time attending, and, and both of us knew nobody, and, and there wasn't really, you know, set programming around like, this is the, the activities. We both just were invited as speakers, and so as speakers, we were sitting next to one another, and It developed in such a lovely way.

It's like 500 people or something? It's supposed to be somewhat intimate, 500 speakers in this, what is it, like in the late spring or summer?

I don't remember. I think that they have a few now throughout the world, but that primary one in Europe is usually in the summer, kind of just before all of the other summer conferences. So, shout out to them. I will forever credit them for this.

So wait, is this gonna be your best investment ever and you just happened to randomly run into the person person at a conference. Is that what you're saying?

Well, I hope this is the low watermark, Jason. I really, I hope they start—

Well, what, I mean, you're up 10x. Give us an idea of where you're at right now. You're up 10x right now?

Yeah. So, so we got in at, um, $1.50. Okay. Wow. Um, and so it's now at $2.3. Um, and they're, they're doing really well. Uh, and I, I will tell you, like, this was my very, very first non, like, highly non-consensus bet, um, because I really was betting on, on the people. I was betting on the founders aggressively. And I really had to go to bat even internally to talk about, you know, suddenly I had to become a thermal engineer and try to understand how space dissipates heat and the fact that it doesn't and like how, why it's better to do it.

What was Alexis's reaction? He's a large, he's, I believe he's the largest LP in the fund. So that's his daughter's money.

Exactly. And that's how he describes it every day.

Yes. So, okay.

Yeah.

We're gonna put my kids' money one more time, Caitlin. in space data centers because there's not enough room on Earth. And this is 2 or 3 years ago, I'm assuming, or 2 years ago.

Yeah.

What— how did you convince him? What was his initial reaction? Was he just like, are you crazy to put data centers in space when that's likely very expensive? Take us inside the, you know, the firm when it's non-consensus. Another way to say people don't agree and it's a crazy outlier idea. Tell us what that's like.

Luckily for me, Alexis is, is always a very good sparring partner. And, you know, in his career, he had to make several non-consensus bets within his own organizations. I think, you know, a good example of that is Coinbase. And that, you know, that was long before there was a crypto thesis and, you know, it was a primary driver of some pretty significant returns at, at that prior firm. And so luckily he knows the feeling. And we actually have a term internally that we use called skin rippy, which is really gross. But like, you want to rip your skin off. You are so— like, you could not be more passionate about this. You want to literally rip your skin off and then run around in a muscular skeletal suit, apparently. And so—

There's a visual for us, Paige.

There is a visual. I'm investing in Uber and—

It's now a horror film.

Correct.

It's not a slasher film.

Biotech and slasher.

You want to go full Hannibal Lecter on this investment. I get it, Caitlin.

Yes, exactly.

You're passionate about it like a serial killer.

Exactly. Deeply, deeply obsessed.

Right.

And so because of that, you know, we use the term iron sharpens iron all the time here. And, you know, having the conversation when I came, you know, it was that night I put it in Slack and I was like, guys, I just— I have seen the future. I have been invited behind the curtain. Philip and the team are going to do something phenomenal. And I really, like I said, he really made me work for it because that's how you get better. That's how you get smarter. That's how you get sharper. So I was able to make a good enough case. I visited their offices up in Seattle where we have a few other companies like Stoke Space and even touring the facility, which was post-check writing. I was like, whew, like, thank goodness, because I really had to go all in, right? And after after being able to convince the court that it was the right decision to make, I did not expect to see the public and kind of big names in the space kind of do this really 180 on it. You know, I mean, at the time of investment, there were people, there were very, very smart people that had actually created websites to like take it down, like to say like StarCloud is the dumbest thing, you know, orbital data centers are the dumbest thing ever known.

Yeah.

This is so stupid. And then literally within 6 months of the wire, It was like Elon and Bezos and Altman and everyone came out and it was like, this is it, this is how we're gonna solve this. And so I didn't expect to get that feeling of, okay, so I wasn't totally crazy so fast. Usually it takes a longer time when you're, especially when you're making a non-consensus bet like that. But I feel very grateful to be a part of it.

Yeah, and thank you for taking us inside the curtain there. I think it's like really interesting to hear how partnerships get these non-consensus bets over the line. And Paige, sometimes there are data points that lead to opportunities. And one of the data points is, hey, the cost of putting a kilogram in space has gotten unbelievably cheap compared to historical costs. Here we're looking at almost $10K back in, let's say, $14,000 to put a kilogram in space. And now, uh, you know, it's, it's come down this year down to $4K. And, you know, projection is in 2030 and then 2040 going down to $1,500 and maybe even $273. This is, uh, obviously because of SpaceX.

Oh, I was amazed by how large the market actually was when you dug into the details. So I think that was a big lesson is when there are questions around like, what is the opportunity if this company is wildly successful in the next 10 or 20 years? And focus on that rather than what other folks may be saying about the opportunity.

And other folks who are not talking to the founder and who are not doing the primary research and are not having the, the blade sharpening blade, uh, research. Caitlin, have you ever looked at the a trend and tried to find a startup based on a trend, i.e., hey, everybody's gonna have GPS in their pocket. What does this enable? Or kilograms to space going down. What does that enable? Have you done that or can you give us an example of doing that?

Yeah, no, this is actually, that's one of my favorite ways to sharpen my own thinking on something is if you can develop a thesis, especially as a generalist, I think that we have a huge opportunity to really, you know, the world is our playground, right? And so, To be able to develop a thesis independently based on things that you've read and thought and feel, or just generally want to see in the world, then you can reverse engineer that back into, hey, I'm, you know, I, you know, Paige, you said it, Twitter is still a very, or X, excuse me, is a very, very fertile place for a lot of folks building, you know, without political commentary. Like it is a really great place to source deals and have dialogue. But But what we've done in the past, and something that I've done, there's a few theses that I've worked backwards from. One is infrastructure for the inevitable. And so if I believe, for example, launch cost is solved, if I believe, for example, that there's going to be this entirely new race within the stars and for humanity, I can say, okay, great. Or AI, for example. If we believe that AI is inevitable, how does the world work? Well, I know that I don't want to be downstream of a lab, for example. I know for a fact I don't want to be downstream of a lab. I want to be investing where the labs are reliant on me, right? I need to be their supplier for something. And so I think a lot about supply chain. I think a lot about being upstream so that there's reliance, not just from a revenue standpoint, but also a learning standpoint, because that is how you get your best data is talking to customers. So there is this infrastructure for the inevitable where I've been able to reverse engineer into companies like HGen who are solving, you know, heavy industry. It's onsite hydrogen production for all heavy industry. They're building down in Southern California and just have a phenomenal opportunity to take out natural gas in a way that I think is just absolutely unheard of. And Paige, we should connect on Knox and HGen and if there's any amazing synergies there. But Uh, that, that's a good example of saying like, oh, I, I do believe that the industry and, you know, this whole trend around the reindustrialization of America and American dynamism is important. And so what, what are some of the flaws and bottlenecks within our existing systems?

I've been thinking about this specifically with unmetered tokens. I don't know if you've, either of you have experienced unmetered tokens in your life, but somebody gave me like a, um, an API key to, um, they gave me an API key to a Bittensor subnet that had put like GLM-51 from z.ai. And I was like, okay, I'm not being metered, great. Instead of doing this daily, this cron job to do research on something, I was like, give it to me every hour. And it starts giving it to me every hour. And then I'm out in Hawaii on my vacation with the family And I'm like getting these like incredible nuggets of wisdom. And then I was like, okay, and every 4 hours I want you to look at your last 4 and then, you know, drops to me and tell me what you could have done better and then rewrite your skill every day to be better at finding this type of information. And I was like, oh my Lord. When you have unmetered, then this becomes like a free human. Like, I'm getting a free human that doesn't sleep that I can put on a task until I tell them to stop doing the task.

Whoa.

And that is going to be possible, I believe, when everybody has a $20,000 desktop, a $10,000 to $20,000 desktop. I don't know if you saw the Perplexity Computer announcement page, but they now with Perplexity Computer are going to automatically load, um, on your Apple Silicon a and I'm gonna predict right here, Perplexity's getting bought. The first thing the new CEO of Apple's gonna do is buy Perplexity because Perplexity now allows you to load, it allows you to load, Paige, an open source model in their harness. And when you do a job, it'll go to the local model first. And if it gets you the response it thinks is good enough, fine. And then it will go to the cloud and do— Through your open router, another cheap open source one, and then fall back to a, uh, finally fall back to maybe a Frontier model. Those— that concept of everybody buying a $10,000 to $20,000 desktop, to me, Paige, seems the future if you could become unmetered. So thoughts on either working backwards to a thesis— you heard a couple of different ways to do that— working backwards from a thesis— mine is unmetered tokens, Um, do you have one of those that you've done recently, or just your thoughts on my, my idea of Apple buying Perplexity and having their own search engine and having their own harness that they could put into iPhones and Mac Studios and MacBooks with 128 gigs of RAM?

I mean, that sounds pretty incredible. We should, uh, call the new CEO of Apple and let him, let him know. Um, I guess one of the theses we developed in the early part of 2024 was around multimodal AI. So I'd gone to a conference and I'd seen one of my LPs who helped build the wearables business at Qualcomm, um, basically discussing the incredible opportunity in application layer AI and specifically in multimodal AI, which if you're not familiar is when you, instead of having text input to text output with an AI model, you might have videos, machine-level data, all of these like different complex inputs and outputs. And so we started thinking about that in 2024, and I would say my specialty has really become applied AI, especially multimodal. And actually that piece helped— I would say like helped sharpen my thinking when we encountered incredible companies. So one example of this is a company called Maneva building To your point on the hardnesses, I believe edge AI is really the future, especially if there's any sensitive data compute constraints, is having those local models run on the edge on very specific tasks. And so that's basically what Maneva does on the factory floor in terms of quality assurance, safety, especially in high-volume factories. And so when I had written that piece, the founder, Ray, as we were going through diligence, read a few of the pieces I had written about Applied AI and was just like very impressed by my vision, um, and let me write a $500K check, which was the largest investment I ever made. And they ended up raising— yeah, they ended up raising a Series A from USVP less than a year later and have just continued to expand dramatically within factories, which has been very exciting.

So continuing to pursue opportunities in that thesis Yeah, and let's take a pause here and talk about another issue that forks off of, uh, StarCloud. Caitlin, it's about to become a fund returner. If it's at 15x now, typically 20 to 30x— there's usually 20, 30 bets in a fund. I'm assuming yours is probably more like 30. You know, you're, you're one turn away from this returning the fund, and that means When you're at 15 to 50x, you got to start thinking about the reasonable thing to do for your LPs. Maybe Alexis is super long, is to trim that position in a hot market, in a breakout company, putting StarCloud on the side here for a second so you're not tipping your cards that you're selling your shares in it because that's rife with— it would be rife with like, oh my God, you don't believe in the founder kind of thing.

Totally. We're also an RIA, so like, there's a lot I—

Yeah, of course. So I'm just broadly speaking, what's the philosophy at 776 for when do you pare positions? And do you tell founders ahead of time, hey, this is our position now as a Series A fund?

Right. It's a good question. And it's actually a really topical one for us because we're only 6 years old. And so we've raised, you know, a significant amount of capital. I think we're at $1.3 billion AUM. We're on, you know, we're investing out of our 4th fund. and a bevy of SPVs. And so we're getting to the point, to your point, of maturity, right? Where we're starting to see some of these companies do really well. You know, we have one company out of Fund 1 that was acquired by X, and so goes xAI, so goes SpaceX, right? And so having that conversation has been really important to us, and having it with our LPAC and our LP community has been really important as well.

Yeah.

You mentioned earlier, Jason, that Alexis is our largest LP. And so he obviously has his own thoughts and feelings about this. And there's a deep, deep responsibility and a fiduciary duty to making those decisions. And so right now, because we're young, we're doing it case by case. And so, you know, whether that's what happens when the SpaceX lockup, you know, occurs to, you know, even secondaries, which has suddenly become a thesis 'Cause every, you know, GP has to consider.

If IPOs are getting pushed out to year 12, 14, 20, or never in the case of Stripe, it seems like, well, then secondary has to be a thing. Our approach has been we tell folks as an early stage, like pre-seed fund, we have an accelerator. We're kind of like in the Y Combinator space. We're investing, we're the first money in. We have an obligation to pare our position at $500 million, at $1 billion, at $2 billion, just along the way. And we're pari passu, obviously. So just upfront, you should expect us, if you're going to be selling or another investor selling, we will be doing that alongside you. Has absolutely nothing to do with how we feel about it. If it was me just doing seed investing, I probably would run it forever. And I, right, I still have a very large position in Robinhood, Uber, Nubank, DoorDash. I, I still own those shares from when they were private companies in the seed round. I still have those shares, same shares, haven't sold. So if it was me and my shares, I've never sold a Robinhood share, as an example. Um, great. But I have LPs, and I literally had to have this like hard discussion with the founder. And they're like, you see, you don't believe in us. I'm like, no, I just have LPs and I have to stay in business. And if I just have markups that are paper page, as I talked to you early in the show, like Fund 4, which is what we're wrapping up now, and I'm doing Fund 5 next year. When you get to Fund 4, you have to have DPI. It's just, there's no option, right? Caitlin, you're in the same position, I think.

Yeah.

You have to start distributing. Uh, capital.

That's it.

Paige, how do you think about it now as you're in year 6?

Well, we're in year 5, so I would say we're a couple years away from those discussions. I do see them forthcoming. We've been pretty proactive talking with our LPAC about liquidity opportunities. Um, we actually did a great episode with Megan Reynolds from Altimeter and Jason last year, and I learned a ton. We wrote like a whole guide on secondaries because It was very interesting to see it rise from this like swept under the rug taboo topic to now generating, I think it was like the majority of exits.

Majority of exits is, yeah, majority of distributions are coming from secondaries, correct?

Yeah. So it's a, it's definitely something that we're actively thinking about and as Caitlin mentioned, like on a case-by-case basis.

Yeah, it's definitely the talk of the town right now. Caitlin, final question on this incredible investment of yours. And I talked to Philip about it on This Week in Venture— This Week in AI, which he came on. He's a great guest. You have a pretty significant competitor, probably the one competitor no entrepreneur ever wants to go up against. No, I'm not saying Travis, but that's one you don't want to go up against.

Yeah.

Elon Musk. So how do you look at the investment? And when your LPs say, wait, is this the same business as the one Elon's been talking about now? And our way of getting to space is with the same guy we're competing against, so we have to pay him to get to space, and then he's going to be meeting us up there with his data centers. How do you think about having that level of competitor and being partners with them to get to space? This is a challenging thing.

Yeah, I, you know, it's funny, people ask this a lot and they ask it of several of our space tech companies, including Stoke. But I really think, again, you have to— the conviction that I have around what is next for humanity and more importantly, how going off planet can actually help us retain and maintain life on Earth in a sustainable way where humanity can truly thrive. is like, I fundamentally believe that nobody has to catch SpaceX for another company to win. So, you know, right now the entire American space economy runs on one company's launch manifest, right? Like, that is, that is a known fact, and that's what you're asking about. And I think that that's, you know, you can look at it from one perspective, which is like, it's the single point of failure for StarCloud, for Stoke, but it's also the single point of failure for the Pentagon, right? It's a single point of failure for everybody. And so You know, when I think about, about Stoke, like they're building the first rocket where both stages actually come back successfully and fly again. And the reality is, is our country, humanity needs more than one airline, right? And, and doing it differently and having things that make you different and the things that you are spectacularly good at, um, that, that helps. But it's also, it's, we're to the point I think, again, thinking about humanity as one and as a whole, it's okay to have more. And in fact, it's better to have more. And so when I look across the entire space and specifically launch sector, I want everyone to win regardless of where my money is. And I really sincerely think that this is a case of rising tides buoy all ships. And in the meantime, you see a ton of collaboration. Elon and Philip at StarCloud, They, they sincerely are incredible partners. Um, you know, the mini lasers that are going up on, on the, the satellites for StarCloud, they were hand delivered, right? And, and named very cutely. I, I was there for the unboxing of one of the, the mini lasers named Cheddar. Like, there, there is a real true camaraderie, just like building a startup. I, I, I view space tech as like the world's startup. And so when you're building a really hard thing under incredible stress with insane potential, I think across this industry what you're seeing is that camaraderie and that linking of arms between people because they have such clarity of long-term vision that frankly the rest of the world can't see. And so yes, if Starship is a cargo ship and then Stoke's Nova is the delivery van, right? But like, and frankly, I think most of the space economy needs a delivery van. And so I just think, I really think that if we can think about it not as there's a winner and a loser, but it's really, you know, ultimately humanity will win because of this collaboration. I think that that's the better approach to take, especially as an early-stage investor thinking about it.

Well, and to be clear, Elon, I think, is acutely aware of this issue that he doesn't have yet a competing airline. He's like, there's United Airlines, and if You'd like to get to space, you could also take United Airlines.

Yes.

And we also could recommend United. But that will not always be that way. There will be other options for people, and it may take a little bit of time, and he might have the best price to get to space. But he has never denied any customer going to space, specifically not a competitor. So, you know, if you look at other people putting, uh, data and Starlink competitors in space, my understanding is Many of them have used, uh, you know, SpaceX's rockets to get to space. And then, uh, the same thing with StarCloud. StarCloud put their first H100 up with, uh, SpaceX, and they have multiple launches in 2027, uh, I, I see here in my notes. And so I don't think it's an issue. I think you're exactly right that this will be something, um, that will, um, just make its way through. All right.

Physical AI.

Let's talk about it. Physical AI companies now have raised almost $50 billion in venture funding— wait for it— in the first half of 2026, up 80% from 2025. 6 months of this year outraised 2022, 2023, and 2024 combined. Aerospace and defense are the top 2 categories, obviously.

Yeah.

And this is not cheap, and we're all early stage. So I guess the question I have for the panel here, and maybe we'll start with you, Paige, and then go to Caitlin, how does one get these companies early and then get the conviction that your tiny investment and tiny ownership stake is going to actually result in some meaningful ownership in the future? I think these companies have a lot of dilution. Yes, Paige? Because they have to raise such enormous amounts of money. Obviously, we're in a hot market, valuations are pretty hot, but still, dilution's dilution. So how do you think about that issue of, hey, I'm going to be the seed, pre-seed investor in this, or I'm going to squeak into a competitive Series A and get a $250K check in. But what is that worth after they massively raise? You know, billions of dollars, potentially tens of billions.

Yeah, it's a good question. So I would say the approach that we've taken in this space is that we focus mainly on verticalized applications of physical AI. And I think there you see less of the issues around massive dilution. Um, you see companies getting to revenue faster, to commercialization faster. So I'd say that's mainly where we've focused our investing efforts at Behind Genius. And I think there are incredible opportunities in more general purpose physical AI. And that's not to say that we wouldn't invest in one of those companies in the future, but I would say the, the verticalized opportunity is very, very interesting.

Do you require pro rata in order to, Paige, participate in deals, or will you participate in a deal without pro rata? If you do have pro rata, but you have a small fund, uh, and an early stage fund, what do you do with that pro rata? Have you figured out how to leverage it? Maybe with partnerships, maybe with SPVs like we do at our firm. Yeah.

Yeah. Um, so we mainly focus on making initial investments. We'll make 25 to 30 investments per vintage, roughly the same size at pre-seed and seed. And we'll very rarely double down in companies. We've done it in a handful of companies where we wanted to increase our ownership.

And you do that from the fund? You'll double down from the same fund that we're in? Obviously much higher valuation, much less ownership, but you think you have a winner.

Yeah, I would say usually it's like pre-seed to seed. We have asymmetric information advantage and the valuation is not so astronomical that that check isn't meaningful. Um, and then from a pro rata perspective, pro rata is interesting because sometimes it's written into the docs and then sometimes it's earned as a as an output of building a relationship with the founder. So we've been able to drive $10 million in additional investor capital from mainly from our LP base, uh, directly onto cap tables. So that has been my focus.

Do you do that with an SPV and take carry, or have you just done that as like a favor? Like, hey, you should talk to this company LP.

I mean, I would, I would say it's like very much a win-win for, for us, for our LPs, for the companies that we back. And I think it's one of the joys of building a collaborative size fund where we're not leading rounds and we can afford to bring some of our favorite people into.

Well, and it's the reason why some people do, uh, back early stage funds is to get that information. But Caitlin, please explain to Paige that she deserves carry on those pro rata. Please explain as an, as an elder millennial and as a Gen Xer, we're going to give you a different framing here. Go ahead, Caitlin.

First, I want to, I want to say that I think that the, the, the appetite for co-investment has definitely changed and being able to capitalize that in any capacity page is, is phenomenal. And so like the fact that you're, to your point, it's, it's all about your relationship with the founder. Um, and you know, even if it is written into the docs, your ability to get that pro rata, super pro rata, I mean, demonstrates your access big time. And so sometimes, you get what you get. You don't have a fit. And as someone who's built something from nothing myself, you do that. And then to Jason's absolutely sublime point, like, please do not undersell yourself. And your partners will understand, especially when they are the right partners. And, you know, being able to ask for something, you know, based on market value and what's going on in the ecosystem, you should never undersell yourself just because Uh, you think you need to be constantly building.

Yes. And so you, uh, your, your Auntie Caitlin and Unc Jason, or your big brother and big sister as the case may be, are saying, let's, let's, let's spin up an FPV here and you can just take 10% and 1%. Just a little money. Just a little.

Just a little.

You can, uh, you know what, the next time it happens, Paige, talk to my team. I'll put you in touch with Kathy and Heidi and John. And Jackie, and they'll just show you like exactly how to do it. Easy peasy lemon squeezy, and boom! Just get a little cat. It's a little tasty poo. Just so you're you know you're in the game.

Little sweet treat. Like a little sweet treat.

It's like what? It's a little. I mean, I can't wet. Paige can't wet her beak here. I mean, Paige, you guys don't wet her beak. Just to wrap up this one segment, you guys are great, by the way, together. Let's look at this chart one more time. Just one more time on this chart. This is unbelievable. I mean, we were sitting here 10 years ago and I passed on Zipline because I was like, Keller, I love you, I love this technology, but you're delivering blood in Africa on a slingshotted fixed-wing drone. I don't know if this is a donation or this is an investment. I don't think I can get my LPs to do this because hardware is so hard. And now, first half of 2026, it was like, yeah, hardware is the greatest thing ever. It's a moat. It's fantastic. If anybody can build any app and vibe code it in the weekend, I mean, Caitlin, just looking at that, what does that tell you? We're in a bubble or that entrepreneurs are taking on harder problems? A little bit of both? What's your read on a chart that's so absolutely stunning with 450 deals and $45 billion in physical AI in the first half of '26? Back in '23, it was not even 5.

I would— I'm going to give you a yes and, Jason. Yes, please. Thank you.

A little improv here. Yes and.

Are you ready?

I'm ready. I'll catch the ball.

Thank you.

You know, we think about this and talk about this a lot at 776 too. You know, Jensen very famously predicted, excuse me, that every industrial company will become a robotics company at some point. And, you know, I think that's very advantageous for Jensen to say, but I also think it's a bit prophetic, right? You know, I think, so remember, I, you know, prior to getting into venture, I spent over a decade as an executive leading people teams. So I actually, when I'm assessing, is this a bubble? Is this not? Is this something we should be investigating? I really do it through the lens of talent and understanding where the best humans on earth want to apply their potential. And so when I look through our portfolio, you know, I think 10 years ago, the smartest engineer I met, they wanted to work on ads, right? Not because they wanted to work on ads, but because that's where the money was. Yes.

Ad tech. It was ad tech.

Woo! Hey, I worked at Reddit for a long time. That was our bread and butter. And, you know, I, but I think today, and this is the difference that I'm, I'm feeling, is today that engineer, she wants to build rocket engines in Moses Lake, Washington, right?

Why?

Because she wants to feel connected to humanity, to a badass mission, to something that is meaningful. And, and if you can do something that is meaningful, and make money, now you've got something really, really interesting. I think, you know, I've lived in San Francisco for nearly 25 years, and this idea that, you know, San Francisco is back and everyone's coming and everyone's signing leases again, it's a beautiful thing, you know, from a city perspective. But really the conversations that are happening in this town and in the ecosystem at large are really about like, how do I do something meaningful with my life? And capitalize on this moment in time. And so I think there's this convergence that's occurring in general. And I think, you know, there's the hardware taken. Yes, I hear you. I also, you know, at the same time they were saying, don't sell into governments, don't invest in companies that are selling into governments. They also were saying hardware is hard.

You know, those two things— I mean, those two things were what was pounded into your head, Paige. They just told us, don't do hardware. They'll run out of money. There's no margin. The Chinese will just copy it. And then Whatever you do, do not sell into education, the government, healthcare, anything the government touches. You will die on the vine. It's crazy.

Listen to your Unc and Auntie. That's what my boys— I have Gen Alpha boys, and they call me Auntie when I'm—

Auntie. People start calling me Unc. I said, what does that mean? What does that even mean? They're like, okay, Boomer. I'm like, I'm Gen X.

My dad's a Boomer. What? Unc can actually be a compliment. Just so you know.

I think it is. Like, when you say— yeah, I think it is.

Wait, what? Guys, I don't think unk is a compliment.

No, it is.

It can be.

Jen Alpha says it very much can be.

So see, Jen is like—

it might be like a loveable— like, you know, this is like a— oh, it's so sweet.

Is it? Is there even a chance that saying unk is a compliment, Paige? She says no. Caitlin?

But you know what, Jason? This is another learning lesson for Paige. I promise I'm not talking down to you at all. This is a funny conversation. No, no, not at all. What I've discovered is Gen Z is actually learning how to not be the youngest generation, and you actually don't know what's happening culturally as much as you used to. And it happens. This is why millennials hated Gen Xers. This is why Gen Zs hate millennials. Like, there's this whole thing. And so you are going through the transition. You are becoming— you're in the chrysalis. You are going to become this butterfly of suddenly being an elder in the room because Because my boys, Gen Alpha, they will come in every day and tell you that Unc is a term of endearment. And it's like, you are the cool uncle who comes in and you know what's up, but you have zero Fs to give. So Unc's back in a positive way. Okay.

There's a big debate going on amongst my producers in the Slack room for the live taping.

Wait, I would love to.

Yeah. I'm just going to say right now that One young producer saying, never heard UNK to be good. And then another producer was like, hey, man, UNK is like Obi-Wan. Everybody wants to be Obi-Wan. Yeah.

Yeah.

So why— there's wisdom in it. And also like a great side of—

Boomer is bad.

—inside a tunnel. Boomer is definitely an insult. Boomer is bad. Yes. I'm going with UNK is Obi-Wan. I'll go for UNK is Obi-Wan. That appeals to this Gen Xer.

See?

I'm going to— okay.

I'll post a poll.

One thing that's super appealing to enterprises is AI sovereignty. This has become the topic of the summer. And Anthropic has been forced to change their data retention policy because they've been getting major pushback from customers who fear that they are going to feed the last company in the world, as rumors were saying they were describing themselves internally at Anthropic, a claim Dario denies. But if you give your data to Anthropic, the consensus in the entrepreneurial community is that they will do to you what they did to Cursor, which is launch Claude Code, or do what they might have done to Figma, which is launch Claude Design, or do what they did to legal or biotech and launch a competing product. So if you are a customer giving them tens of millions of dollars a year in your token use to build a really great vertical product But most people in venture and startups believe they will eventually take your data, take the lessons you've learned, and use it for reinforcement learning and eventually create the vertical app based on 20 different people giving them information about venture capital or 2,000 people giving them, and eventually they'll create a venture capital firm and put us all out of business. Anthropic says, now, They're reversing course. Data's gonna live, uh, there'll be a 30-day retention window, and data will sit on customers' cloud environments instead of Anthropic's. What's your thought here, Caitlin? Have you watched startups that you've invested in or that you're familiar with look at the frontier model companies differently today than they did 3 years ago in terms of, am I training and funding my eventual demise?

Yeah, I mean, I think that that's the question, not just with data, but also with skill and talent, right? People are experiencing this on, I think, every side of their livelihood. And so these are really existential questions, and they're really important questions. And I think I'm supposed to say at some point that, like, we're also investors in Anthropic, and so I need to tread lightly, but I need to disclose here. But I think that the best thing we can do is evolve, right? To learn, evolve, have the right critical conversations. And I, you know, I think about my own experience here, like this laptop that I'm on, you know, every single thing on my machine belongs to somebody else, right? I've got LP commitments, founder cap tables, financials from companies that haven't announced a round, board minutes, like all of these things that are very, very precious and important, not just to me, but especially to, to my portfolio companies. And, you know, I, I pride myself like Paige on building real, authentic, and deep relationships with my founders. And they trust me with that information. Um, and, and they, they trust that I'm not gonna skin it or upload it or put it into something where somebody else can, can, you know, take them out and, you know, with a sniper shot. I, I think that this is an important conversation. I think it's an important evolution. And so, you know, when like a lab tells that my data sits on their servers for 30 days, that's not really like a procurement question, right? That's me deciding if I can keep my promise and my commitment and maintain my integrity with a founder. And so I think generally what's happening is, I say this trepidatiously, is like the right shape. I think it's important that we continue to have open dialogue and continue to invite people into the room that can share this perspective so that we can get to the best Yes, please. But I, you know, I think the better signal here is that like enterprise buyers really now have enough leverage to sincerely change like a Frontier Labs policy. I think 18 months ago that was completely out of the question. So I don't know. I'm, I'm still learning about it and I'm still trying to understand how I feel about it is the honest answer.

Paige, your thoughts?

I think it's been something I've been thinking about a lot is, um, Paki McCormick had this amazing essay. Maybe a couple years ago of like everything is technology. And I think if you look across the past 5, 10 years, technology has just become integrated in so many more things that are, yeah, like our day-to-day lives. And as technologists, as venture capitalists, we have a responsibility to think about how to manage that responsibility. And I think that Anthropic taking that customer feedback and addressing their policy As it relates to that. Yeah, it's amazing. Um, essay by Packy. Uh, is it a step in the right direction to really just listen to the customer feedback that's happening?

You know, there was a famous, um, blog post from Fred Wilson, Union Square Ventures, formerly Flatiron Ventures, uh, Be Your Own Bitch. And, uh, not Growth Is a Bitch. Maybe he did— he take it down because people didn't like the term bitch? Maybe. I gotta look Yeah, we gotta find this one because I can't find it right now. I'm doing a quick search for it. And the point of the be your own bitch in the social media era, the social media companies would look at the data of the API companies and they would open an API. So Twitter, Facebook, Facebook very notably, Google, they would open up an API, they let you use it. Then they would look, what are the top uses of our API? And they would study those uses.

Mm-hmm.

And it could be everything as simple as Facebook Connect. You let people connect to Facebook, then you add the social graph, and they'd say, wow, this thing is growing very quickly. I wonder if we should make that part of Facebook. And they did. And so at one point, Fred Wilson wrote a blog post on this and he made a very provocative statement. Hey, be your own bitch. Don't be their bitch. Be your own bitch. Build your own platform, build your own API. Get out of that. This is the same parallel, but with higher stakes because it's not just about the social graph. It's not just about a feature race. It's about the actual intelligence, the actual trade secrets. And I think this is what ElevenLabs, Lovable, Figma are all starting to realize. And actually, I think, uh, Legora— is that the legal, uh, one? Yeah, Legora. I believe Legora announced that they now have their own language model for their kingdom and that they've made small vertical language models For fiefdoms. So Sherman Sterling and Wilson Sonsini and Auric can all have different language models in on-prem, I believe, or on their own clouds.

Yeah, I was going to say we're going back to on-prem.

Back to on-prem. And one of our big investments, Advocates, we'll throw that on the screen, is like literally providing on-prem LLMs, vertical LLMs. And they just raised at an incredible valuation. They went through our accelerator. That's going to be a big hit for us. Or is a big hit.

I agree.

Yeah. It's great when something works out in this industry.

It's so lovely.

It's so lovely. It's like, wow, it worked. We incubated something and it's successful.

It's amazing.

It's a very weird thing. You know, sometimes I feel like an absolute superhero. I'm like, I can fly. I can touch things and turn to gold. I have got the lantern. I'm Green Lantern. I can do anything. Then all of a sudden, like, I take the ring, I put the ring on, and I'm like, Boom, I can fly. And it's like, yeah, your ring's drained, nothing's working. The last 10 startups all imploded, nothing works. You'll never hit another winner. And then it's like, oh, there's another winner. It's possible. Thank the Lord.

But power law is a beautiful thing.

It's a— I mean, it really is like a crazy thing. Um, but I'm very excited about this possibility of people taking the open source stuff and then um, building their own verticalized language models. And I think ElevenLabs, Lovable, are starting to realize, hey, backing up the Brinks truck and giving tens of millions of dollars, if not hundreds of millions of dollars, to Frontier Labs, and then giving those tokens to our users in a wrapper to provide a service is really dangerous. Like, we're living on the edge here, and any moment they could rug pull us. Um, all due to people building Frontier Labs, but they're going to be under a lot of pressure to make money. And when they get under pressure, this whole like, I'm going to be courteous and I'm not going to create an ElevenLabs competitor, I'm not going to create, you know, a lovable competitor— it's like, that's out the window when your stock is in the, uh, yeah, shitter. I said it.

The woods.

This is literally what I mean. When you get taken to the woodshed— maybe that's a little less graphic. Yeah, when you're, when you're taken to the woodshed and your stock's like on the floor, which could happen if they don't have profits.

Yeah.

All right, listen, another amazing episode of this week in startup/venture capital roundtable. Paige Doherty, Caitlin, thank you for coming on the program. If founders are watching this saying, I'm going to build the next great company and I would like to have Paige, Caitlin, and J.Cal lead my seed round, my Series A, How do they reach you? How do they pitch you, Paige? How can founders find you?

Yeah, you can reach me on my Twitter, which is Paige Finn with 3 Ns, or at our website, behindgenius.com. Behind Genius.

There it is. Such a good name.

There it is. Thank you. Behind Genius. I like it. That kind of, it's like a little bit humble, a little bit of a humble brag, but Mostly humble, but a little bit of a humble brag because like we're behind the geniuses. So, but it's not us, but it's good. It's a really good name. Thank you, Caitlin. Also a great name, 776. How can people reach you?

Thanks. Yeah, so similar to Paige, you can find me either directly on the website here. My email is the letter K at 776.com. You got to spell it out. We're still searching for the domain that I can't yet afford with the numbers. So spell it out, 776.com, or on Twitter, Caitlin_Cruz, which is my middle name. People want to know. And Cruz is the last there. So yeah, would love to hear from you. And Jason, thanks so much for having us on the show.

This was really fun. And Advocates changed their name, so I got to give them a proper shout out here. They're now GOAT. Go.ai. There you go. Nice. This is incredible. Look at this.

It's like the Chemical Brothers song.

Go One. In one box, you plug it in, there's the Ethernet cable going in, and AI happens on-prem in 15 minutes. On-prem. On-prem. On-prem. You own it. Look at that gunmetal steel. Look at that. Looks like an Apple product. This is a waitlist. Maybe J-Cal helps you get past the waitlist. Go to go.ai. Y'all know how to reach me. We'll see you next time. Bye-bye.