YC W25 Portco Writeups

I wrote a bit about YC in March, but only touched briefly on the actual investments. I do these recaps for LPs every batch and it always gets me excited for the next one. Below you’ll find notes on my thinking for each of the 7 investments. Our average check size was $215k and the average valuation was $27M. This was about 20% more expensive than past batches, but it was also one of the most impressive groups of founders we’ve seen. I’m extremely excited about the founders we’re able to support from this last batch and I think you’ll see why in the writeups below.

When you make investments before demo day (6 out of the 7 were) there's always a little voice in the back of your head that worries if you made the right choices and wonders if seeing all the demos live or finding out what's "hot" will make you feel like you missed out on something. I'm happy to report that this was not the case! We felt very confident in our choices before demo day and even more confident afterwards. 

Without further ado - here's a look at our W25 portfolio with links to their YC pages and some notes on our decisions for each:

Karoo - Canadian neobank focused on lending

This Toronto based team is actually from England... More specifically, the founders all met while working at Lendable - a London based fintech company where they helped grow the company to a $4.5B valuation. Lendable has built their success by offering consumer loans in a digital first format at fair rates while structuring themselves to remain asset light (both from a debt perspective and by famously not raising much VC). The Karoo team felt there was an opportunity to run this same playbook in a different market and that ultimately led them to Canada - which has one of the most concentrated/profitable banking sectors in the developed world (90% of market share dominated by the "big 6"). Between the team's experience, their roadmap clarity, and the ability for small teams to do more with less via AI - we're extremely excited to see them ramp up operations in the years to come.

Pax Markets - Crypto exchange on a chip: faster and smaller than a datacenter

If you haven't read Flash Boys by Michael Lewis I would encourage you to pick it up. It's a quick read and a crazy story. If you're feeling a little bit lazier - there's also a movie about the same story called The Hummingbird Project that's also a fun watch. In short - it outlines the not-so-humble beginnings of the High Frequency Trading market and the general idea that if you're physically closer to the order book you can consistently front run trades. Pax is taking this to an extreme by actually putting a complete crypto exchange on a single chip... Such proximity unlocks unprecedented value to high frequency customers and PAX shares this value by offering zero-fee (similar to what Robinhood was able to accomplish in public markets) with cash-back to every other market participant on every trade. The founding team has deep experience in the HFT market as well as the unique technical ability to make it happen.

Red Barn Robotics - A Roomba for weeds on a farm

My wife comes from a third generation farming family (organic wheatgrass and alfalfa mostly nowadays) so whenever I see new tech that's relevant to agriculture I'm always interested. I personally believe this is a massively under-developed / under-invested area. To put it in perspective - of the 5186 companies that have gone through YC - only 28 of them carry the "Agriculture" tag in the YC directory while 2488 have a "B2B" tag. Meanwhile, the global agriculture market size was valued at approximately $11 trillion, compared to the global B2B software market at around $420 billion. Despite agriculture being over 26 times larger by market value, it receives a fraction of the startup and venture capital attention. The Red Barn Robotics team is bringing their top-tier technical talent (one founder worked on special hardware projects at Apple he literally can't talk about and another white-boarded the design and assembled the first 600 Starlinks) to solve a simple, but pervasive problem for farms all over the world: Weeds. Traditionally, farms will hire seasonal workers to weed hundreds of acres by hand. Their robot is faster and cheaper than human weeders - ultimately saving farmers money and increasing yield. This is a perfect example of founders using new technology (advancements in computer vision as one example) to solve a seemingly simple, but pervasive problem. They are already live on farms in the pacific northwest and I'm excited to see them continue to roll out their solution.

Inversion Semiconductor - Manufacturing the most powerful chips, 15x faster

I'd be lying if I said I'm a semiconductor expert, but odds are - neither are you. Despite having invented the industry the manufacturing process has largely (in some cases entirely) moved out of the US. Inversion Semi is taking a step towards bringing some of that process back by building a next-gen lithography machine. If you're not familiar with the process already, lithography uses light to pattern circuit features on silicon and advanced Lithography has been monopolized by a Dutch company named ASML that famously builds "the most complicated machine ever produced by humans" (to reference Chip War which is worth a read). Inversion Semi is scaling transistors to their physical limits by shrinking particle accelerators 1000x, to create a high power light source. The team has a unique background to make this happen and the founders are part of only a handful of people in the world that are familiar with beam-driven plasma wakefield accelerators (relevant nature article here if you're into that sort of thing) which have only recently become stable enough to be used in this way. When I asked friends who are deeper in the semi manufacturing space they responded with "only a team in silicon valley would be audacious enough to even try this" and that's exactly the type of team we're excited to back.

Asteroid - Browser Agents for Complex Operations

AI agents everywhere. You have probably seen something on Twitter or Linkedin about AI agents by now and all the promises of them replacing human workers. It can be hard to wade through the hype and understand what/who is actually driving value in this space. It's also hard to make investments with 5 to 10 year time-horizons in such a rapidly evolving space. We got excited about Asteroid because of their background in the space (researching/building in AI before it was cool) as well as their clear focus on tasks that require human oversight in regulated industries. This is an interesting niche for two reasons: 1. These regulated industries are often home to some of the most mundane/repetitive processes that browser agents are well suited for even in their current state. Workers clicking through multiple apps to input information and move outputs to other apps before producing a deliverable. 2. Our bet (and Asteroids) is that these industries will not see value accrual "trend to zero" as quickly given the artificial supply/demand market constraints of the human licensing process. Where other agents may seem high value today, many will be easily undercut on price as competitors emerge and the "price of intelligence" continues to go down. We're excited about what's possible with agents, but great new tech doesn't always accrue value and we feel Asteroid is taking a thoughtful approach that we're proud to back.

Pig - An API for automating Windows apps with AI

You know when your SF friend says that AI is going to automate all the manual data entry type tasks that bog down massive industries like finance, healthcare, manufacturing, and government? I think it can (and should), but it won't... not unless someone figures out how to allow agents to interact with software that was built back when you downloaded applications to your desktop using a CD. Eric - the solo multi-time founder for Pig - ran into this problem personally while trying to implement agent loops for these types of businesses and it's why he decided to build the solution. Similar to our thesis on Asteroid - we believe (like many) that these kinds of legacy businesses have the most to gain from automation, but they are also typically the hardest to implement solutions for. This isn't because the decision makers are old and don't like change - the common excuse most silicon valley founders/investors use when talking about the difficulty with selling software to these types of business - it's because the owners/decision makers are practical and it's often hard to justify the upfront effort/cost required to implement these types of upgrades and automations. By decreasing this adoption cost Pig will be able to unlock massive amounts of value for customers and the agent builders who aim to serve them. We believe in the promise of agents, but that promise will go unfulfilled for the majority of the economy without solutions like Pig.

Uncommon Therapeutics - A Disease Focused Biotech, Building Multiple Billion Dollar Drugs

When we talk to founders, a big question I'm always asking myself is "Was this founder born to do this?" This is similar to the classic "Why you?" and "Why now?" questions, but it drives deeper in an effort to push past why a founder could do it or why they should do it towards why they have to do it. All VCs (at least the good ones) want to work with life's work founders. These are people who are singularly pursuing what they truly believe they were born to do. It may sound a bit mystical, but in venture-land this is the highest calling of both founders and the investors who look to support them. Oftentimes - unfortunately - the true life's work founders story of how they came to build what they're building isn't a very happy one, but it's always meaningful in an extremely deep way. We believe that Noah (co-founder and CEO of Uncommon) is one of these founders and I'll let you read a bit about his story in his own words (can find their full writeup on the YC launch section of their directory page linked above): 

"Imagine while you are building your high flying tech company, your 18 month old daughter starts to lose the ability to walk, speak, use her hands, and will soon develop a cascade of maladies (seizures, breath holding and hyperventilation, scoliosis and on) – this is Rett Syndrome. Over the months that followed Noah dedicated himself to this problem, reading over 300 papers in the first 9 months alone, speaking with every expert until he became one. We identified multiple repurposed drugs, not commonly used for Rett, that while not curative could immediately push back against the skill loss and successfully give her a higher quality of life. Today, Ellie is 5 and in the top 1% of people with Rett. While still limited, she is happy, walking around, not in pain, without seizures, and is learning to read and communicate with her eye gaze device."

Noah had to do this for his daughter and we believe he can use the same skills at the intersection of AI and biotech to bring these results to patients at an unprecedented pace. To quote from their launch post again:

"At Uncommon we believe now is the best time ever to build a disease focused biotech. So while Biotech may not get the first 1 person billion dollar company, in this AI-driven open source system, it may deliver the first 2 person one."

We share this belief and we're excited to support their effort.

I always enjoy doing these recaps because it helps me articulate and build my own conviction outside the rush of actual deployment/demo day. I’ll be back out in SF for the Sprig Demo Day on June 11th and I’m excited to do it all over again.

Talk soon,
Adam