# 10 Highlights from USV's 2025 Annual Meeting

**Published by:** [Union Square Ventures](https://blog.usv.com/)
**Published on:** 2025-11-10
**Categories:** ai, energy, tokens, programmable
**URL:** https://blog.usv.com/2025-annual-meeting

## Content

We like to say that USV is a continuous conversation. It’s through conversations that new ideas and observations emerge, take root as theses, and continue to grow through public engagement. Over the last year, we’ve been sharing ideas and sparks as they surface through the @usvlibrarian and the OH @ USV series. We believe that the best way to develop our thinking is to get our ideas out to the public. AI helps us capture and share more of those ideas across all our conversations. In that spirit, we ran a transcript of our 2025 Annual Meeting through a custom prompt that extracts highlights while preserving the texture of the conversation. Here are 10 highlights from USV’s 2025 Annual Meeting, edited for clarity:1. “What does a new consumer payments model look like?”▼Fred opened the meeting by sharing a story that illuminates how USV develops and executes on a thesis:In 2016, we were talking about our Stripe position, and Albert said something along the lines of, 'I'm really nervous about the long-term sustainability of the credit card system and the interchange fee business model.' So we started thinking about what companies building a new model for merchants and consumers might look like. We called this thesis area moving away from credit cards, or eliminating the interchange fee business model. It took a while for us to find companies that fit into this thesis, but we've had it front and center in our minds for close to a decade now. The first two companies we'll lead off with are great examples of how we've executed on this thesis, Super and Blackbird.2. “Our mission is to use data and AI to make payments free for businesses”▼Samir Desai, founder of Super Payments, framed how the biggest internet markets have trended towards a free model and monetized through data:If you look at the internet, large markets like search, social, video, have all gone down a free model, and those businesses have made money through the data that is there. I was increasingly seeing the same effect in fintech, with businesses like Robinhood doing free stock trading, and Credit Karma with free credit reports… the mission of our business is to use data and AI to make payments free for businesses.3. “Restaurants are still largely misunderstood and undervalued as entertainment assets”▼Ben Leventhal, founder of Blackbird, shared his belief that restaurants are premium entertainment and should have its own native loyalty and payments layer:We believe restaurants are still largely misunderstood and undervalued as entertainment assets. If you look at how restaurant companies are being valued at the high end of the market now, like Nobu and Carbone, you're seeing pretty sky-high valuations around a billion dollars… there is something about restaurants that make them fundamentally premium entertainment assets. Restaurant payments are problematic because they cost restaurants a ton of money, 300 to 400 basis points, and they aren’t getting much for it. Restaurants themselves have no real tools when it comes to customer acquisition or retention… the average independent restaurant is competing with huge behemoths for customers, and they can’t win with the tools they have.4. “AI is poised to be the solution to its own energy problem”▼Carl Hoiland, founder of Zanskar, shared how they're using AI to discover and develop geothermal resources at a speed, cost, and scale that wasn’t possible before:We were founded with a belief that AI was poised to be the solution to its own energy problem. Geothermal is the perfect energy source, it has a tiny footprint, no carbon emissions, and if you know where to drill, it’s already the lowest cost source of energy on the planet. We train AI models to understand geothermal systems by simulating all physically possible subsurface realities, and we're now out discovering and developing geothermal resources at a speed and a cost and scale that just simply wasn't possible before.5. “Innovation happens when you combine a new technology with societal shifts”▼Andy referenced Steven Johnson’s concept of the “adjacent possible” and how AI and collective intelligence can unlock opportunities that were previously impossible:There’s this concept that you’ve probably heard us mention before over the years called the "adjacent possible," which was introduced by our friend, the writer and entrepreneur, Steven Johnson. The idea is that innovation happens when you have new tools, new technologies, new societal shifts that allow for new combinations of things that were impossible before. If you think about healthcare, there's always been a choice. You can have instant access, you can have clinical accuracy, or you can have affordability, and you only get to pick one of those. Two or three of them simultaneously could not be done at scale. Matt from Doctronic asked, ‘What if AI agents could engage in collective intelligence — multiple specialists working together, debating diagnoses, and delivering instant and inexpensive care — to solve the historical trade off?'6. “We believe that AI will be able to legally practice medicine”▼Matt Pavelle, founder of Doctronic, explained how their approach combines AI accessibility with human medical expertise:The way Doctronic works is that any user can come to the website and talk with the AI for free, and by default, it's totally anonymous with no PII. At the end of any conversation, we give a full assessment with four diagnoses and four treatment plans, and anyone can take the information We’ve had 17 million patient interactions this year, and our doctors and our AI are aligned on treatment plans 99.2% of the time. One of the cool things that happens with our AI is that it gets smarter every time somebody talks to it. So the more people talk to it, the more it learns. Our doctors are in the loop providing feedback on every single conversation.7. “We are in bubble territory… but there’s a chance that it continues for quite some time”▼Albert shared one case why the current AI bubble might be different from every previous bubble in history:We know that every technology bubble has burst. It goes as far back as the South Sea bubble where Isaac Newton famously lost his shirt, to the railway bubble, the roaring 20s, the dot-com bubble inflating and bursting, and then multiple crypto bubbles. But there is some chance that this particular bubble is actually different. The notion that AI can be a self-improving technology makes this fundamentally different from any bubble that has come before. We've never had technology that accelerates itself as you build it. Where does all of this leave us? I think all this leaves us in a place where we think there's a chance it bursts, but we also think there's a chance that it continues for quite some time.8. “The market will move towards a fundamentals based view of crypto protocols”▼Nick observed how crypto's shift to fundamentals-based valuations could smooth boom-bust cycles by anchoring valuations to real revenues instead of speculation:Our general feeling is that the market will move increasingly towards a fundamentals based view of crypto protocols. This is the clearest in DeFi protocols, things around lending and swapping, where there are real on-chain revenues that are directly correlated to volumes and transactions. At the moment, there's still somewhat of a substantial premium in the market, and that may continue for some time. But the medium/long-term trend will be more towards fundamentals-based pricing, which could lead to a substantial re-rating of some networks that don't have strong fundamentals in place, or that don’t have economics connected to the token. The other thing that this could lead to is a lessening of the boom-bust cycle, because things start to have fundamental, sustainable, lasting value.9. “The best way to transform the structure of a market is to make it programmable”▼Rebecca shared a learning about how companies can most effectively change market structures by making industries programmable:Our last horizontal thesis was about transforming the structure of markets from the outside in, what we think of as the ‘edge’. What I mean by that is to approach markets in ways that avoid gatekeepers and transform how they operate structurally, rather than working to change them from the inside. What clicked for us is that the best way to do this is by making them programmable. By making each of these industries programmable, you have more levers and better tools to transform how they operate in a structural way.10. “Tokenized equity onchain could be a huge new asset class”▼Fred shared a perspective on how tokenized equity in private companies can be a new asset class in crypto:Think about what could happen next with companies that remain private, OpenAI, SpaceX, the list goes on. We are starting to see people buy those shares in secondaries, put them in custody, tokenize them, and put them onchain. Maybe what we will see is some companies never go public in the traditional way, but have massively liquid, tokenized equity that's trading onchain. And I think that could be a huge new asset class. And that would look a lot like stablecoins. It would trade on these Layer 1s. It would be available in wallets and be available to be lent in DeFi protocols and so on and so forth. Pretty exciting what you can see if you peek around the corner.We've put together a video recap of the four portfolio founders who presented:

## Publication Information

- [Union Square Ventures](https://blog.usv.com/): Publication homepage
- [All Posts](https://blog.usv.com/): More posts from this publication
- [RSS Feed](https://api.paragraph.com/blogs/rss/@usv): Subscribe to updates
- [Twitter](https://twitter.com/usv): Follow on Twitter