Continued from here
Okay so I’ve been trying to write “Part II: AI” for a bit and I just keep crinkling the draft up and throwing it in the waste basket. Problem is two fold 1) I feel underqualified. I’m not a techbro by nature, not a long-term investor by nature. And then 2) something new happens literally every day. DeepSeek–can you believe that was a year ago!?–was supposed to be this big landmark event that crushed the AI bubble but now it’s a big nothing burger. Then there was OpenClaw. Then Opus 4.6 came out, amazing. Then GPT 5.5 came out, even better (or worse, I’m not sure). Then Mythos. Anthropic itself over $1 trillion now. Memory has been huge for 8 months. Then the semiconductors went absolutely berserk mode in the last 2 weeks. It’s become a job in itself just to keep up with all of it.
I have to write something at this point.
So let’s just talk about every single investment in my portfolio, one by one. I have 25 of them but I’m going to be easy on myself and start with the first 14. I’ve been sort of cagey about unvealing my entire portfolio because I’m a little afraid of being exposed 1of course–most of my audience is traders who don’t know a lick about fundamentals so most of you won’t do that, it’s just an irrational fearfor not knowing what I’m doing. People will be asking me: Pete, why’d you invest in ____?? And all I can say is: Idk man, I just like it.
Let me be clear and just get in front of it–I don’t know what I’m doing. I honestly feel in my bones that there’s this extremely special period of investing in the next 2-5 years that I have to take advantage of. That’s the biggest part of this. Not a ton of deep research. No on-site research at all. This isn’t about recreating a Buffett-esque 50 year track record, I could just retire after this “special period” ends. If I’m just roughly correct about that, I should make some money.
If I had investors–and thank god I do not–I would struggle to explain everything I do. It’s partially because my views have changed a lot since I started but mostly because a lot of my “strategy” is just based on gut feelings.
This is all for posterity. Maybe 5 years from now, I’ll look like a genius. Maybe I’ll look incredibly stupid as the bubble bursts and I’m stuck holding the bag. Right now I don’t feel all that smart because you could’ve bought a number of AI-driven names on the public market and made X times your money with liquidity and none of the SPV risks I’m taking, as of the time of this writing. Freakin’ Intel, really????
Anyway, here goes nothing. The first one you all know already.
1. SpaceX
Date of investment: July 2024
Entry valuation: $210 billion on a 4/7.5. SPV2(for reference when looking at SPV slash line, the first number is usually the upfront charge, 4% and the second number is the carry at 7.5%, which is the percentage of profits the manager takes from the investors)
Bet size: 2 units
IPO status: June 2026, S-1 filing is confirmed.
SpaceX was originally just a launch company that sends rocketships into outer space. Their big mission is to eventually go to Mars. Their current launch project is called Starship—which is essentially a giant fully reusable rocket system designed to dramatically lower the cost of getting mass and humans into orbit. But along the way, SpaceX became many other things: a defense contractor for the government, a global telecommunications provider through Starlink, an AI lab through xAI/Groq, a massive compute cluster with Colossus 3that’s going to be rented out to Anthropicso they’re a neocloud too?, and even a social media platform, formerly known as Twitter. They supposedly want to build datacenters in space, which is a thing humankind hasn’t done yet. They also have a $60 billion option to acquire Anysphere, the maker of coding platform Cursor. So yeah, a lot going on there.
It’s slated to be the first trillion dollar IPO ever–currently rumored to be at $2 trillion, which would be a nice 6-7x win for me. It’s probably a tad overvalued at this point at 100x revenue but 25-50 years from now, they might just be like the Weyland Corp in Alien that dominates everything related to space economy. Truly unknowable upside IMO.
2. Gecko Robotics
Date of investment: August 2024
Entry valuation: $650 million on a 12/10 SPV.
Bet size: 2 units
IPO status: 1-2 years, depends on their traction
Gecko Robotics is a company founded in 2013 by college friends Jake Loosararian and Troy Demmer. Their original idea was to build inspection robots that could climb the walls of large infrastructure and defense assets—bridges, factories, power plants, naval ships—and collect data on these systems for maintenance and preventative repair purposes. For years, it resembled more of an ambitious science project than a fully scaled company. That started to change around 2021–2022, when Gecko raised its Series C and began recruiting a number of ex-Palantir alumni and then adopting a business model centered around forward deployed engineers. Gecko then evolved into a broader AI and infrastructure data platform company through its software system called Cantilever. Its Series C extension—roughly when I got involved—brought in major backers like Founders Fund and US Innovative Technology Fund. In 2025, Gecko raised its Series D at a $1.25 billion valuation, finally achieving unicorn status after more than a decade in existence.
They’re doing something important–some may say “mission critical”4another buzzword I’ve been reading a lot lately–and they don’t have any obvious competitor. I’m thinking I can make 5-10x on this.
3. Databricks
Date of investment: January 2025
Entry valuation: $61 billion on a 7/0 SPV.
Bet size: 3 units
IPO status: perpetually 12-18 months away5they’ve been ready for a couple years but they keep pushing it off with Series I, J, K, etc
Databricks is a company founded in 2013 by a group of researchers out of UC Berkeley who created Apache Spark–an open-source software engine that lets users process and analyze massive amounts of data across many computers extremely quickly. They became a SaaS company that would sell a data platform that they called “data lakehouses”. It gets a bit technical from there, so I’ll stop.
Then the 2023 AI boom hit, and suddenly Databricks found itself sitting directly in the middle of one of the most important bottlenecks6oh boy does this term gets used a lot in tech speak in technology: organizing, storing, cleaning, and training data at scale. Now Databricks is positioned not just as a data software company, but as an AI infrastructure layer for the enterprise world.
It currently trades at 3x of its direct competitor Snowflake (SNOW). Does it get the “AI infrastructure layer” premium or does it get docked for being a SaaS company during the on-going so-called SaaSpocalypse? Time will tell but I’m obviously biased for the former. At current secondary priced 7as of this date on May 12th, it’s priced at about $140 billion, I’m up about 2x. I think it can pop to $300 billion on the IPO pop assuming strong conditions, which would be a 5x.
4. Lambda Labs
Date of investment: February 2025 and September 2025
Entry valuation: $2.5 billion on a 12/10 SPV, $7.2 billion on a 0/10 SPV
Bet size: 1.5 units on the first SPV, 3 units on the second (4.5 units total)
IPO status: they’re saying second half of 2026
Lambda Labs started in 2012 as a small GPU hardware company founded by brothers Stephen and Michael Balaban. In the early years, they were mostly known for building high-performance deep learning computers and workstations for AI researchers at a time when machine learning was still relatively niche. Small developers really like Lambda.
In today’s current AI boom, Lambda transformed from a hardware provider to what is now known as essentially a “neocloud”8as opposed to the hyperscalers which provide all purpose general compute–a new cloud provider focused specifically on renting GPU compute for AI/ML development. They discontinued the old hardware business last year and now the new mission is to build AI factories to power intelligence for everyone–“one person, one GPU” is their new slogan. They’re aiming to be a vertically integrated neocloud by owning the entire physical datacenter, the GPUs, and the developmental software layer. The broader thesis is that AI compute is becoming a foundational utility like electricity and Lambda wants to give everyone access.
They’ve recently crossed $1 billion in revenue and they have a more diversified customer base than their embattled neocloud breathren CoreWeave, which I think will play well into their IPO story. I’m guessing $20 billion IPO, for posterity.
5. FigureAI
Date of investment: March 2025
Entry valuation: $39 billion on a 0/10 SPV.
Bet size: 0.5 units
IPO status: too hard to say
Figure AI is a humanoid robotics company founded in 2022 by Brett Adcock. Figure’s purpose is to build general-purpose humanoid robots capable of performing physical labor in the real world. They’re one of the few vertically integrated players out there because they aren’t just building the physical robot itself—they’ve also developed their own intelligence layer called Helix, an in-house AI system designed to handle the robot. Brett believes the real bottleneck is the software stack coordinating vision, movement, memory, and decision-making in real time. By owning both the body and the brain, Figure is attempting to build an all-in-one platform rather than relying on third party suppliers. Soon they will be deploying the Figure 3.0 in BMW factories to help build cars.
In late 2024, I got obsessed with robotics and wanted to chase down Figure supply while it was still trading below $10 billion in the secondary market. I spent about 4 months trying but before I could secure anything, Figure raised their Series C at a whopping $39 billion in March 2025. As much as I liked the company, it was too much to swallow a 10x valuation bump on a pre-revenue company, so I only made a small bet where I just own them to own them as a part of a larger diversified portfolio of robotics.
6. Neuralink
Date of investment: May 2025
Entry valuation: $9 billion on a 5/20 SPV.
Bet size: 1 unit
IPO status: very far away I’d think
Neuralink is Elon Musk’s brain-computer interface startup trying to build a direct link between the human brain and computers through surgically implanted chips. The company’s first mission is medical—helping paralyzed patients control computers, phones, and eventually robotic limbs using only their thoughts—but the long-term sci-fi vision is human/AI symbiosis. WaitbutWhy has a much more comprehensive post on what they’re trying to do, so I’d suggest reading that.
I wasn’t sure about this one. While there are some neat edge-cases for this product, it might take many years for the mainstream to accept having invasive surgery to have an advanced chip in their brain. This was more of a risk-reward play that I talked myself into and it’s been blazing on the secondary markets9currently trading at$40 billion, a 4x to its series D valuation despite few meaningful updates. One day, all the Elon companies might fold into one super congolomerate and you can use your Neuralink chip to create and play your own video game on xAI while your Tesla drives itself. Something like that would be worth a lot.
7. Anduril
Date of investment: May 2025 and September 2025
Entry valuation: $41 billion on a 1/10 SPV, 60 billion at a 12/15 SPV
Bet size: .8 units in the first SPV, 2 units in the second (2.8 total)
IPO status: sometime before the next presidential election, so 2027-2028
Already wrote about Anduril in Part I. Love this company and I honestly might just hold for life rather than sell into an IPO. I just bought the book The Anduril Thesis, which I am currently half way through. I actually learn more about my companies after investing in them, which is probably backwards to how one should do things but in the world of highly in-demand and quickly closing SPV deals, you kind of have to invest first and ask questions later.
8. Rain
Date of investment: July 2025
Entry valuation: $340 million on a 12/10 SPV
Bet size: 1 unit
IPO status: 1-2 years
Rain is an earned-wage access (EWA)10not to be confused with Rain the stablecoin company fintech that lets employees tap into wages they’ve already earned before payday, aiming to replace overdraft fees and predatory payday loans with real-time payroll liquidity. You click a button, you get your money, you pay the app a low fixed fee, app gets their money on payday. Founded in 2019, Rain integrates directly into employer payroll systems and has rapidly scaled to millions of workers across industries, distributing billions in earned wages to date for workers at large companies such as McDonald’s and Marriot. Rain has the broader ambition to become a financial wellness app for all users.
So this one doesn’t really fit the dominant themes that I’m investing into. EWA is sort of this new theme in fintech that I suspect will be more covered in the next few years. Did you know that 37% of Americans cannot cover a 400 dollar emergency expense without borrowing or selling something? Yeah, that’s not good. 11Source
9. SkildAI
Date of investment: July 2025
Entry valuation: $9 billion on a 12/10 SPV
Bet size: 3 units
IPO status: no idea
Founded in 2023 by Carnegie Mellon robotics researchers Deepak Pathak and Abhinav Gupta, Skild is a robotics AI startup building what it calls a “general-purpose brain” for robots — a single foundation model designed to work across many different robot bodies and environments rather than training separate software stacks for each machine. You don’t need a human to program some specific task for it, it will just be able to train itself.
The upside here is a WindowsOS product for robots. Once robots become more of a thing in society, I expect an explosion of smaller-use case robotics companies to pop up and just license the Skild Brain rather than training their own model. Robot to untangle Christmas lights in your garage? Skild. Robot to throw tennis balls at reasonable distances to safe locations for your golden retriever? Skild. Robot to perfectly rotate gas station hot dogs at 1am? Skild. Go check out their videos.
10. Erebor
Date of investment: August 2025
Entry valuation: $2 billion on a 10/30 SPV (!!!!!) with 2% annual mgmt fee, capped at 10 years12I just threw up in my mouth typing these fees out
Bet size: 0.5 units
IPO status: lightyears away as they just exited stealth mode
Erebor is a tech-focused digital bank founded by Palmer Luckey and Joe Lonsdale. Created in the aftermath of Silicon Valley Bank’s collapse, Erebor aims to become a modern banking layer for the “innovation economy,” serving startups in frontier technology sectors that traditional banks often avoid. Supposedly this idea was inspired from when Palmer tried to get traditional funding for Anduril and was turned down. Silly banks, don’t you want exponential returns?
I honestly feel horrified to pay these insane fees but there was no other way to access this highly exclusive deal. I also wanted to A) establish a new relationship with this particular venture syndicate and B) have priority access to follow-up rounds on Erebor as it gains traction. Gotta invest in those Tolkien-titled companies. They just started but already have over $1 billion in deposits after six weeks. I think there’s 10x upside, although it will be dampened by the ridiculously high fee drag.
11. Canva
Date of investment: August 2025
Entry valuation: $42 billion on a 12/10 SPV
Bet size: 3 units
IPO status: next 12-18 months
Founded in 2013, Canva is a design software platform that lets anyone create presentations, social posts, videos, documents, websites, and marketing materials through easy drag-and-drop templates. They have over 220 million users and their estimated 2025 revenue was about $4 billion.
At the time I invested, I liked it but now I think they’re vulnerable. If they did an IPO today, I think it could take hit because of the SaaSpocalypse. There’s a current narrative in public markets about how all AI Labs will eat software’s lunch. Adobe and Figma’s stock are both getting crushed. Canva’s trying to do things that give them an AI forward narrative but I don’t know if it will be enough. Maybe they’ll wait it out. There’s still a good business here but a 10x revenue multiple could pull in a bit. This is the only investment of these 13 companies listed where I’m not sitting on a markup.
12. Deep Fission
Date of investment: September 2025
Entry valuation: $3.00 on a September Private Placement (implied fully-diluted valuation at the time was around $180 million)
Bet size: 1.5 units
IPO status: next few weeks, as their S-1 just became effective
Deep Fission is a nuclear energy startup pursuing an unconventional approach to small modular reactors by placing nuclear reactors a mile underground inside deep boreholes. Founded by nuclear geniuses father-and-daughter duo Richard Muller and Elizabeth Muller, the company argues that going underground could dramatically reduce construction costs and duration, simplify containment, and improve safety. They’re one of 11 companies that’s part of the DOE Pilot Program, which is designed to accelerate approval, funding, and deployment in nuclear SMR technology. Broadly speaking, we do need nuclear power to fuel the expansion of the AI and compute boom. It’s safer than you think.
So this is a bit of a speculative play, as this company has less name recognition and VC support than all the others. I got in via private placement at $3–it’s the first one I’ve ever done. They did another private placement at $15 in February, which means I’m sitting on 5x markup on paper. They should list on the OTC any day now, with the plan to apply for Nasdaq immediately. I think even at $15 (less than $1 billion market cap), they have a favorable valuation compared to existing SMRs on the public market13(as of writing this, OKLO is a $12 billion company. are they really better?).
Am I qualified to research in a nuclear company and believe in their success? Hell no. Am I qualified enough to say “right place, right time, nice price, why not?” Yes, I am! We’ll see what happens.
13. Crusoe
Date of investment: September 2025
Entry valuation: $10 billion on a 1/10 SPV.
Bet size: 5.5 units
IPO status: one more round is rumored, then the next 12-18 months after that
Crusoe is yet another AI factory company. Originally founded in 2018 by Chase Lochmiller and Cully Cavness as an energy-focused crypto mining company, they pivoted aggressively into AI infrastructure in 2023. Crusoe is often described as an ‘energy-first’ neocloud because of its focus on securing power supply, including the novel approach of capturing stranded natural gas flare from oil fields. The company combines power sourcing, modular data center construction, cloud infrastructure, and GPU compute into a vertically integrated stack. Crusoe is the lead builder behind the massive Stargate AI infrastructure project in Abilene, Texas.
If Lambda is the developer-first neocloud, then Crusoe is the infrastructure-first neocloud. They are the best at securing power and building massive-scale projects. If the rumors are right and they raise at 30-40 billion pre-IPO, then I think the actual IPO round could be in the $50-60 billion range. This is the largest investment in my portfolio. Fun fact, NFL running back Saquon Barkley is also an investor.
14. Anthropic
Date of investment: November 2025
Entry valuation: $240 billion on a 12/18 SPV
Bet size: 1 unit
IPO status: October 2026 maybe?
Anthropic is an AI research lab founded in 2021 by former OpenAI researchers, Dario Amodei and Daniela Amodei. They created Claude–you familiar with Claude? Anthropic is the AI company at the center of the current investing zeitgeist—the one everyone talks about as a potential winner-take-all eat-everyone’s-lunch AI giant. The one AI giant to rule them all. They just passed rival OpenAI in secondary market cap valuation, which seemed unthinkable a year ago.
At the time I invested in this, I was hemming and hawing and thinking–oh man, how much upside is really left? Gosh, it’s so hard to stomach all these steep fees. Gosh, I hate second layer SPVs. Okay, okay, fine, fine, I guess I have to do at least a 1 unit investment. Uh, looks like I didn’t invest enough because the secondary market currently has it trading at 1.2 trillion. This might be the most in-demand company I have ever seen–might want to be careful though.
That’s it for now.
I’m going to have to expand on this in the actual Part II AI article but as you can see, with 9 total units invested in Lambda and Crusoe, I am very, very bullish on AI infrastructure. I think it was around August of last year when I had my a-ha moment that AI was going to become a far bigger deal than almost anyone expected—that the boom would grow much larger, and last much longer, before it eventually bursts. This is the new oil we’re talking about. Events since that crystalization moment have only further confirmed my thesis.
Next post will be about the next 11 companies. These first 14 comprise almost all of the late-stage pre-IPO stuff and the next 11 are friskier plays more in the realm of true venture capital.14Oh and if you want to know the company I bought with no revenue and no disclosed product worth $5 billion, it’s a company called Hark.
