加密货币的价值捕获难题，以及 Robinhood 为何自建区块链

All right, everybody, welcome back. Happy Friday. We’ve got the man, the myth, the pessimist, the bear, Santiago. Hi, cabrón. Listen, dude, I thought it was going to be a very bullish episode last time because you weren’t here, and it just turned into Rob and me arguing the whole time about VVV. So I got quite the controversy, right? I saw a lot of people posting, and then I think Sahil kind of went out and—

People were very mad.

Yeah, people thought Blockworks was attacking VVV, and that really was not the goal. I want to start the episode by saying I have a ridiculous amount of respect for Eric, VVV, and the Venice team. It was more of a commentary on the state of the industry in this dual model.

Anyway, how are you doing? We’ll talk about that in a second. How are you doing coming back from the Seleni conference? I think it’s one of the best private conferences out there, other than, of course, the Blockworks ones. Really good crowd and very good discussion.

I was on a panel and, obviously, a debate. The prompt was, “Are generalizable L1 tokens investable?” It was such an easy debate to have, right? It was me—

Who did you debate?

I debated Brian Pellegrino from LayerZero, a phenomenal builder, and Sahir from Plasma—

Sahir?

I believe that’s his name. If it’s not, I’m sorry, but I think it’s Sahir. Yan Liberman from Delphi was on my side. It was a good discussion. We won the debate, in case you’re all wondering.

That was a 2-on-2. Interesting.

Yeah, 2-on-2.

Nice.

No, it was really good. I basically got to share a lot of what I’ve been thinking yet again, with a little bit of a twist, and you had the benefit of seeing it. It was a good discussion.

Yeah. Markets are so bad you’re having to hold your mic. We have no stand today, or what’s going on with this?

Yes, markets are tough. What can I say? No, this is me bringing passion. I’m trying a new form of, like, Wolf of Wall Street reading this one.

No, I changed it.

The stand doesn’t work.

1. Inversion’s Crypto Thesis Reset

Good. Good. Okay. I don’t actually know how much you want to share here, but I got your investor update for Inversion. We can cut this out or skip through it quickly if you don’t want to talk about it, but I thought it was one of the more thoughtful investor updates I’ve ever received.

Most investor updates—and even the way Blockworks sends ours—are often, “Here’s our revenue, team size, bank account, financials. Here are the wins and losses, and here are the goals for the next quarter.” I think you’ve done that in the past, but this one was several pages long, and it outlined not only your thesis today, but what you’ve learned since starting Inversion.

You had this goal of acquiring a company and embedding it with crypto, and I don’t want to put words in your mouth, but it seems like you’re updating the strategy. We can skip through this, we can even cut this, but if you feel comfortable, I’d love to dig into your learnings.

First of all, thank you. It’s a testament to the team. Obviously, what goes into producing a shareholder letter is a collective effort.

My team has been amazing. I’ve got Mr. Perplexity, I’ve got Mrs. Claude, and I’ve got—

Fable Five. I’m happy to show it. I was actually thinking, and I asked you, “Hey, is it possible that we’ll share a redacted version?” I do think it’s useful.

I’ll start from the very end. The very end was just, “If our investors have seen hard evidence to the contrary in operating results—not pilots—we would rather be corrected now than later.” We’ve been quiet for a while, but under the hood we’ve been doing an incredible amount of work evaluating the question: Can we buy businesses where we can implement crypto and have a great return on investment commensurate with the risk?

I want to emphasize the last part, because none of this is saying that crypto isn’t useful. None of this is saying that the technology isn’t real. On the contrary, I wouldn’t have started Inversion, and I wouldn’t be here, if I didn’t believe that. There’s a lot of evidence to support that. You can look at your own dashboards, and I don’t need to regurgitate any of that.

The nuance is: By implementing crypto in these businesses, does it actually produce an above-market return? I’ll hold myself and the rest of the team accountable to that, because rule number 1 of investing is, “Don’t lose money.” Rule number 2, as Charlie and Warren would say, is, “Don’t forget about rule number 1.”

I started Inversion under this hypothesis. I really wanted to understand how real it is and how we’re going to see it. I took—

Wait, maybe restate the hypothesis. What was the hypothesis?

The hypothesis was: Can you buy a business and make it dramatically more efficient with crypto? We’ve looked at 20-plus sectors and companies.

Tell me about the journey, because I know you guys were looking at the MVNOs. You were looking at—

2. Where Crypto ROI Breaks

A host of companies. Yeah.

Totally. Yeah.

People might have heard me say here and on other podcasts that the MVNO thesis was predicated on the idea that you could bundle telco and fintech together.

I’m a student of history. The thesis is stablecoins. There’s huge demand for stablecoins outside the U.S., especially in places where there isn’t enough financial infrastructure and services. Stablecoins have really worked, and Tether is a true testament to that.

Then you look at companies like M-Pesa that bundle telco and fintech into basically an everything app. I said, “The best way to build the largest fintech in a place like Mexico or Latin America is through telco,” because it’s a core service.

What I’ve started to notice is that once we started looking deeper and deeper into MVNOs—which are very competitive businesses on their own—you’re renting infrastructure from large carriers. You don’t own the infrastructure. It’s basically about whether you can acquire customers.

The problem is that these businesses have terrible unit economics. There’s a lot of churn, and there’s no meaningful cash flow. It’s really hard to underwrite this bundling when you don’t have cash flow to invest in the business and make the transformation. Again, there’s a lot of risk, so that strategy fell apart on its own. We’ve deprioritized it.

We then looked at credit and this idea of, “Hey, could you borrow on-chain? Rates are low.” There are a number of businesses outside the U.S. that just don’t have access to capital markets. They don’t have the same availability of capital.

We uncovered so many of them. I talked to a former Ribbit partner who launched his own fund, and he’s investing in places like Venezuela, Colombia, and Asia. Our conclusion there was that credit is challenging.

There is a capital market on-chain that mostly works for crypto-backed loans. Any time you start doing anything other than that, you’re faced with a couple of challenges. One is smart-contract risk. Another risk is that your borrowing is variable. With Maker/Sky, we looked at what they’re doing with variable rates as an interesting experiment.

You’re exposed to utilization rates and borrow curves that are difficult to underwrite. The true borrow rate for what I would call B2B business lending is closer to 8% to 12% on-chain today, not the 2% to 5% that you see in DeFi. That is—

That’s not competitive with TradFi.

Yeah. Again, if you want to buy a good business, a good business in the real world has access to capital at a cheaper rate.

The question is, if I’m buying a business that is borrowing at 20%, that’s likely not a good business.

There are a lot of risks embedded in that. The common denominator—and I’ll just say it now, and you can interrupt me whenever—is that the mistake I made was holding constant the filter of which business I wanted to buy where you could release a lot of value with crypto.

You end up looking at businesses that aren’t great businesses. They’re broken in their own way. They’re melting ice cubes. They’re out of favor.

You’re falling—you’re catching a falling knife, in the private-equity version of the term.

You and I have talked extensively here about remittance companies, companies like Western Union. Obviously, go look at the chart and the financials. It’s a business that’s challenged, right? You have more technology-forward companies like Remitly and Wise that have really compressed the unit economics on certain corridors.

They're also facing heat from companies like Félix Pago and stablecoins. I instructed the team to look at pretty much any remittance company that we could buy after hearing Western Union boldly claim that they could really cut prefunding. We went out and talked to a number of remittance companies, and again, remittance companies are challenged: the multiples have come down, and the business itself is, again, a melting ice cube. What we found there that was actually more revealing is that, in order to fix a remittance company, it's not a technology problem. It's a stakeholder-incentive problem.

Going back to my initial observation, the technology works. Stablecoins are very real; you can definitely implement stablecoins to move money in real time. But if you're dealing with stakeholders that don't want to implement the technology, it's a hard sell, and it's not one that I want to undertake.

Yeah.

I'd rather—we did a lot of work there, and we came out of that bearish. I'd rather be in wait-and-see mode. If Western Union can pull it off, then we'll reengage, no problem.

3. The Distribution Quality Trap

Yeah. What's your take on distribution? I think a couple of times on the podcast you've said, “I want to buy distribution at the best price possible. If I can get cheap distribution, I can do a lot with that.” Tell me your takeaways from distribution.

Another mistake, I think, is that not all distribution is created equal. Said differently, you want to buy a business that has pricing power and brand, and is not operating in a supercompetitive industry. That's what makes a great business. Those are real moats. A downstream effect of that is distribution, because distribution is: do you have an installed base? Do you have connectivity with a customer?

If a company has distribution but doesn't have pricing power, brand loyalty, or profitability, distribution doesn't matter at that point. I think you can get into this trap of buying businesses that have a huge installed base, a huge user base, like MVNOs, but it's not high-quality distribution.

High quality meaning the users are sticky, or the users are valuable—kind of, dare I say, wealthy users who you can upsell more and more products?

A little. It has to be both, in my opinion. For instance, in the telco case, you could definitely have access to millions and millions of customers. The question is: are they actually going to be willing and receptive, and how costly is it going to be for you to upsell? Upselling is not a slam dunk in and of itself.

I'll give you the example of MVNOs. I very quickly understood, “Wait a minute. We're going to buy an MVNO in a place like Mexico that doesn't cash flow, and you have Nubank, which is probably the best-run fintech in the world, absent Revolut, that has invested over $1 billion in Mexico and has only converted around 9 million users. If they can't do it for $1 billion and I don't have cash flow to do it, that's a hard sell.”

Yeah.

So, I guess distribution—I would reframe that today as: I love listening to the Invest Like the Best podcast. He talks about big ideas and billion-dollar P&Ls. We talk about that. I think, simplistically, Inversion now—and part of the investor update was—“I want to buy a good business, a quality business.”

What makes a quality business is that it cash flows. If a business has been able to produce cash flow over different market environments, that tells you everything you need to know about a business, right? If there's only one filter, it's that: does a business cash flow? And the second piece is, I think, a business that can be transformed with technology easily is not a good business.

Interesting. I would actually be curious to get your opinion because, in media—newspapers, for instance—they were definitely transformed with technology.

But the newspaper companies weren't the ones who won, by the way. Media was transformed, but it went to TikTok and it went to Facebook.

Yeah, yeah.

Yeah, yeah, 100%.

The other sector I want to talk about, which is really popular among the crypto crowd, is FX. We actually ran an RFP for a very large multinational industrial company. They said, “Look, we have operations all over the world. We want to do treasury management; we want to be moving money much faster.” We said, “Okay.” We went out to a bunch of providers in crypto and said, “Quote us your best rate. What can you do here?”

What we learned is quite the contrary: most of these crypto aggregators were quoting above the mid-market rate—20 to 50 basis points. That's a lot, especially for exotic corridors like Colombia, Egypt, and Israel. It's a liquidity problem, not a technology problem. You need to have liquidity against these pairs to be able to price competitively. Someone's going to solve that, and it's going to be great, but that's not the type of risk that we're willing to undertake at this point.

4. Inversion Broadens Its Mandate

Yeah, yeah. So that gets into what I think is the important question: what do you do with Inversion now? You raised this money, and you're sitting on some funds. Are you going to wind it down? Do you invest in other things? Do you broaden the mandate? Do you get more specific? There's also another question here, which I think is probably tied to this: is there a play on AI that you will look at? What's the update?

The update is: broaden the mandate, and first and foremost, go buy good businesses. You want to buy a good business whereby implementing technology—and that technology could be AI, it could be crypto, it could be other things—best practices will make that business a better business. It's as simple as that.

Again, as a student of history, I look at the 2 best examples of what Inversion is trying to copy: Berkshire Hathaway and Constellation Software. Everyone knows Berkshire. They got started very small, and they were a super-small team as a holding company. Warren and Charlie just went out and bought really good businesses—good operators, just give them money, good brands, good-quality businesses.

Mark Leonard started Constellation after leaving venture capital. He left and built this holding company by buying software businesses. You probably know the story really well. Most people would be surprised to know that Mark Leonard started with 25 million Canadian dollars back in 1995, and he built a $500 billion-plus business. He would do $2 million to $4 million acquisitions, and he did many, many of them. He was just a capital allocator.

He built what I think is a really good proprietary data set of what makes a good software business. He got to a point where he had such a good understanding of what a good software business was that he could very quickly buy—or not—or make a decision on capital allocation, because he had perfect information on what a good software business is: margins, magic number, growth, sector, vertical, horizontal—everything, right?

So we're broadening the mandate, and we're going to deploy balance-sheet capital. That's my desire. That was in a shareholder letter. Of course, we have a shareholder meeting. But this whole process has been an incredible validation of a couple of theories that I had in my mind.

One of them is that the holding-company model is better than the fund model. If we were a fund, you have this identity, right? You have a mandate to invest in crypto. At the Seleni conference, I was talking to a bunch of fund managers, and they're like, “Guys, I wouldn't want to be in your seat because you've sort of promised your LPs you can invest in crypto.”

I think all of them privately will tell you that they've moved on. You see it, right? They want to invest in other things, like robotics and AI. Credit to Paradigm, which just came out this week—they got a lot of heat for it. In their prior fund, they said they were going to pivot to AI, and everyone was laughing at them.

I think they're smart. That's what you should be doing. The game of investing is ultimately truth-seeking, and sometimes you're going to figure out stuff that doesn't conform with your hypothesis, and it is your responsibility to adapt. I think it'd be irresponsible not to explore implementing AI and other stuff. It would be irresponsible to buy shitty businesses just to prove to the world that you can buy a business.

And it would be irresponsible to buy a business to prove that you could implement crypto. Of course, you can implement stablecoins in a business overnight. Any customer of Stripe has already done it. The question is whether it will release enough value to outperform the broader market.

I was having a discussion with Brian after the podcast, and he's an investor—he's great. I said, “Bro, Brian, if you really wanted to be long this thesis of businesses implementing crypto, honestly, go buy Stripe secondary and go buy Robinhood equity, because I do think that the larger players are likely in a better position to reap a lot of the benefit from this.” You're already seeing that with Robinhood Chain. We should talk about that—

And Stripe.

You know, I think they're in a better position. I want to emphasize that it's not to say crypto won't work. Of course, you can buy a business, shave some costs, and improve that business with crypto. That's not the message here. The message is that it may not be sufficient to clear a good return on investment.

Yeah.

5. Crypto’s Value Capture Problem

Maybe let me ask it differently, because I think the thing that people would probably be wondering right now—and the thing that I'm wondering—is, are you bearish on crypto right now? That was also part of the letter, because I get asked this question a lot.

You hear me talk in these debates, and here's what I wrote in the letter and firmly believe. First off, I think, as a group, Inversion has done probably the most amount of work going deep into trying to understand how crypto fits into ordinary, traditional businesses. I think the technology is real, the interest from businesses is real, and the willingness to implement it is real. Stablecoins are real, and tokenization is real.

But the value accrual is still not resolved. Open-source networks have a habit of pushing value away from companies and onto consumers. The technology cost curve—you know this well—is incredibly disinflationary, and a lot of that just ends up in consumer surplus. I don't want to be caught on the wrong side of that trade, and I still don't know where value is going to accrue.

I still don't know if Tempo is going to be a profit center for Stripe or if it's just going to help Stripe, the actual business. I still don't know what Robinhood Chain is going to do to Robinhood. I certainly know that Robinhood equity probably benefits. I probably know that Stripe equity benefits, but I don't know what that means for the smaller businesses we're looking at, which is the other thing worth mentioning: we're looking at much smaller businesses.

I am waiting to see more evidence, and this is why I ended the letter the way I did. As part of this journey, we've been at it systematically for 8 months, looking at many different sectors. As George and my team like to say, it's like turning over stones to see what's underneath. I think that's a really good way to crystallize what we do.

A lot of this was also serving crypto founders. I would go to a lot of them in my portfolio and say, “Hey, guys, have you seen evidence of businesses adopting your solution? Are there any clients or potential clients that you think we should be looking at to invest in?” You'd be surprised, but a lot of them were like, “No, I can't think of any.” I said, “What do you mean? None?”

I think this is the nuance: a lot of these companies are starting to implement deep-tech solutions. But I wonder, if every company implements them—if I do it for a company that I buy and lower the costs—my competitors are going to be able to do that as well. You're back to square one, because if everyone adapts to the new normal, it goes back to this: if you haven't bought a quality business, you're in a pretty bad spot. Everything just rerates down.

The benefit of having many of our investors be crypto-native funds and founders is that I'm constantly talking to them and saying, “Hey, what are you seeing in your venture book that's worth paying attention to that I could implement?” So, I think I'm bearish on crypto in the sense that—I’ll say it differently—I don't think the ROI of implementing crypto in a business is big enough today.

I think there are still going to be huge successes, like Hyperliquid, that really profit from trading and hyper-financialization. Those are not the businesses we looked at. We looked at the more traditional use cases. I definitely think the ROI of implementing crypto is nowhere near the ROI of implementing AI.

One of the other things worth mentioning, which I talked about in the letter, is that when we talk to founders of traditional businesses now and ask, “What are you going to do for the business?” they're like, “We've implemented AI in our workflows and in our company.” It's incredible. I'll let you talk about it, but that is very real.

I'll give you an example. We've been super capital-efficient. We have a really small team, and that's deliberate. I couldn't be doing what I'm doing now without AI. I would need a much larger team and many more resources.

I remember the days when I was at JPMorgan. Think about it: today, you and I can say, “Hey, you should look at this sector,” and within hours I have a very enriched CRM of some of the best companies I can go buy. A lot of it is super-filtered. Agents scrape industry databases and bring me really high-quality information, to the point where I can get on the phone with a founder and have call notes that are super-enriched. That would have taken weeks, an army of analysts, and subscriptions. Now that's no longer the case.

6. Finance For The Next Generation

Yeah. I mean, you can talk about what you guys are doing, right?

No, no, no. I mean, anything I say about it will be because Fable is just—I mean, we've got 5.6 coming out from Chat soon, which is supposedly mind-blowingly good, according to the people who have used it. I haven't used it yet. And then Fable is just crazy. But is there anything else on Inversion that's worth touching on?

I don't think so. Obviously—

There's a lot to talk about. Let's talk about Robinhood.

Let's talk about some of that. I'll probably share some versions of this letter for the benefit of folks, because publicly I'd push you to—

I want to open-source it to people—

Because I think it would be helpful for the ecosystem.

Yeah. The one thing I'd call out on the AI stuff is that we have really leaned into it. Fable is insane. If you haven't used Fable yet, it is by far the best model. It is way better than 5.5. It is nuts. Where do you notice that?

In what it can actually build?

The thing I've become obsessed with over the last week is Claude Design. Our product designer, George, uploaded our design files and design system into Claude Design. Then someone on our creative team, Zach, uploaded all of our creative assets into Claude Design. We can now make any landing page or slide deck in it. It literally takes—I'm not kidding, Santi—5 minutes to build.

We bought Messari, right? We need to update all these landing pages. We need a new landing page for our API, a new landing page for the monitoring product that Messari has, which is phenomenal, a new landing page for investor relations, and a new landing page for our contact-us form to book a demo. We need all these new landing pages.

Previously, I would have told our head of product that we needed this. The head of product would have gone to the product designer, the product designer would have mocked it up and sent it back, and I would have said, “We need some work.” Then you would have to send it off to someone who's good at copy. I'm not kidding: this took 24 hours.

The whole thing took 24 hours, and then the product team built it and sent it to the engineering team. The engineering team just took the ZIP file from Claude Design. We designed this whole thing without even going into Figma. Our design team doesn't use Figma anymore. I might be speaking out of turn—maybe they use it a little bit, like 5%—but everything used to be designed in Figma. It's now 5% of the usage that it used to be.

Claude Design has also changed things for the sales team. We took all of our sales decks and combined them with our creative design system, putting everything into Claude Design. Now our sales team can literally go into Claude Design. If you're a seller who hates making decks, sellers hate making decks because they say, “I should be closing,” and I agree with them.

For years, we had the idea of creating a product marketing team or an in-house agency that just builds these assets, because we have so many decks that need to be created.

It’s such a pain in the ass. They could go in and upload their Granola notes.

Best thing ever, by the way.

You upload your Granola notes, combine them with the HubSpot record, and combine them with the internal Slack. For every sales deal we have, we have a Slack channel. It might be called “Customer Inversion” if we’re trying to sell to Inversion. You then upload all those things—you can do it automatically with a connector—and say, “I’m pitching Inversion. It’s a $275,000 deal. We’re pitching them investor relations, and they want to buy $50,000 worth of credits for our API.” It will make you a deck in 2 minutes. That’s better than anything we ever could have had.

Okay, by the way, for all the old-timers—

It’s messed up. It’s really messed up. With Fable 5, if you haven’t used it, I’m not debating whether we have AGI or not, but this is the first thing that’s made me think, “You will definitively spend less time at a computer.”

7. Ads (Peaq)

You will just have these—Fable 5 is really what has made everything click for me.

Robots and machines will outnumber humans onchain and peak is how they get there. Peak OS enables any machine to do business on any chain. It turns machines into autonomous actors and liquid assets with ease, giving robots access to capital and the ability to compound their value. That means funds, allocators, and institutions can underwrite, finance, trade, and route capital to machines the same way they handle any other asset. Millions of new potential consumers for any web3 service, from compute to storage to anything a machine may need, without a human in the loop. Head to peak.xyz or click the link in the show notes to find out more.

So, 2 things on Fable. First, it’s interesting to note that Cursor and Meta released 2 models that are back. The performance of those 2 models is very interesting. Anyone who was short Meta—it’s pretty remarkable what Cursor pulled. They announced it yesterday, right? It has very similar performance and dramatically lower token prices, like an order of magnitude cheaper.

8. AI Pricing Hits The Limit

Yeah. Well, the other thing that has happened with Fable is that the cost of Fable is ridiculous. It’s ridiculous.

It’s hot. When you say ridiculous, it’s ridiculously high.

It’s ridiculously high. It’s super expensive, and it chews through tokens like no model I’ve ever seen. Everyone at Blockworks has access to Claude, and we’re now getting more internal requests than we’ve ever gotten from people saying, “Hey, I just ran out of credits.” I think this is something that companies have to figure out: How do you normalize your AI spend while your token usage is going to grow exponentially?

I think the model Meta just came out with today is 75% cheaper than Claude and OpenAI. This is where things like GLM 5.2 or Kimi 2.7 come in. You basically need an LLM gateway, because probably 80% to 90% of the work at Blockworks could be done with Opus 4.8 or ChatGPT. I forget which model came before GPT-5.5—I guess GPT-5.4. GPT-5.4 sucks, but maybe one of the previous models.

There’s probably 10% of the work that has to be done with Fable. So how do you tell the team, “No, your work is not worthy of Fable right now”? I think this is something many founders are thinking about. I want everyone at Blockworks to be using AI exponentially, but we can’t route every single question through Opus. Do you want me to share your screen?

9. Robinhood Chain Takes Off

Yeah. Here, let me pull this up. All right, Gavin Baker. By the way, if you want to be at the forefront of what’s going on, Gavin’s a great follow. This doesn’t include Meta’s latest model, but I’m sure the benchmarking data will come out today, and we can talk about it next time. Do you want to talk through this? If not, I’m happy to.

Is this correct? What you see is that Grok has similar performance to Fable, and the average cost per task is—

Fascinating.

Basically, just dramatically lower. It seems like Meta also pulled off something similar. On this point alone, I agree with you: If you’re just prompting Fable 5 with questions and answers, you’re using a Ferrari for something that isn’t meant for that. That’s why I asked you earlier what you’re using it for—to create things, to code, and stuff like that.

There’s actually an interesting prompt you can use. You can ask, “Am I using AI correctly? Am I utilizing AI correctly?” I prompted it, and it said, “You’re definitely using Fable for things that are just not—”

Right.

Necessary, right?

I think the biggest criticism—the biggest bears of AI—has come from people saying that the ROI isn’t there. Token prices are really high. Companies like Uber, Amazon, or Microsoft blew through their budgets, and employees are spending 10% of their salaries on tokens. That’s untenable.

This is the Jevons paradox argument: The price of tokens has come down, while performance is still high with Grok. You’ve seen the chart of the percentage of the population using AI—it’s one red dot. Penetration is still extremely low. My reading of this is incredibly bullish: adoption, enterprise and consumer-grade creativity—all of that. Whether Anthropic or OpenAI wins that game is still TBD, because it obviously puts that into question.

But I think the most important thing we’re talking about is that every founder has to get their company as AI-pilled as humanly possible, but you also can’t let these costs get too far out of hand. You have to figure out what’s happening internally. You might need custom harnesses to do better routing, better caching, and better visibility.

We use Ramp, so how do you see who’s spending money on tokens and how they’re actually spending it? Cache misses are the easiest way to drive your costs up. You need to keep the context lean, and you need better visibility.

I pulled up our AI channel on Slack while we were talking about this. I saw one of our data analysts say, “I created a skill so Claude and Codex can delegate their tasks to Grok.” A lot of the comments were, “Yeah, this is the way.”

I’d recommend that people put a full-time employee on solving this internally—an AI engineer. Not necessarily to build AI into your product, which you should probably also do, but to figure out what software you’re paying for internally that you don’t need, what repetitive tasks happen at the company that you can automate, and how to get your costs down as a company.

Yeah. Yeah.

We have that. We’re a super-lean team, but we have 1 person who is constantly looking at our workflows and asking, “How can we codify skills and memory, reduce dependencies on 1 model, and make sure we have a memory layer?” They’re basically following what Andrej Karpathy has been saying about Obsidian and memory: Let’s implement this.

The modifications we’ve made in a short period of time have had a huge downstream impact. So, yeah, I definitely agree with your recommendation.

Should we talk crypto? We’re 35 minutes in.

We’ve been talking about crypto. This is the crypto news, the weekly roundup. We’re supposed to round up—

The weekly roundup. I alluded to it, but, yeah, let’s run through it.

I think Securitize officially went public. There’s also this whole VVV thing, which I’m a little tired of at this point, but we can talk about it because I’d be curious to get your take on this dual-class equity structure at some point.

The biggest thing that happened this week was Robinhood Chain. We talked about it last week—they had just launched the chain—and I would say it’s been a very successful week. My take is that Robinhood Chain is a very real thing.

Let me pull up the numbers here. There’s been over $500 million in volume on Uniswap on Robinhood Chain. That’s more than any other chain besides Ethereum mainnet. Robinhood Chain is now doing more volume on Uniswap than any other chain in the world. It’s up 10x day over day.

Let me try to pull up some other stats. Let’s look at new wallets. There were about 150,000 new active wallets yesterday on Robinhood Chain. These are real numbers, folks. These are bigger numbers than I think many of the biggest layer-2 networks have ever even seen.

That's not doing much more than BNB, because to me, immediately, I think of Robinhood Chain.

No, no, I don't think it's doing more than BNB, but I think this is telling folks something. I think the “everybody needs their own chain” narrative was dying, and I think Robinhood Chain is going to be quite successful here. It's going to tell other people that they need their own chain, too.

By the way, I don't think everyone needs their own chain, but I do think that Robinhood Chain, Base, and BNB are going to be some of the largest chains in a couple of years.

Yeah. I mean, user-aggregation theory in effect—

They control the users. They can point to their own databases.

I thought, Mike—

Go ahead.

Well, let's go to Mike. I want to get your take. What's the benefit of Robinhood launching its chain?

Sure. What's the point of verticalizing the stack? What's the point of Robinhood working with Kalshi for prediction markets and then eventually getting rid of Kalshi and doing it all in-house? It's to verticalize the stack, which means you have more control and better margins.

More fixed, more upfront cost to do it, but ultimately better margins. My theory of where all this goes is that all centralized-exchange trading will eventually move on-chain. If you agree with that—and you might disagree—but if you agree, I think most trades that route through Coinbase will eventually just settle onto some DEX.

By the way, Coinbase might own the DEX, or they might own the—

To put it in software terms for people who aren't crypto people, the Robinhood Chain is like a system of record for trade reconciliation and also probably monetizes a user a bit more—someone who was going to go on-chain to another chain. There's a subset of the Robinhood user base that's trading crypto and going on-chain to trade, and you're looking at the dashboard and saying, “Okay, this is growing to a point where we're going to verticalize and insource it, and we can make more money by launching our own chain.” Like, there was—

Some discussion at the board, and someone showed the stats, saying—

Twenty percent of our users are trading crypto, or whatever the number is—I'm using hypotheticals—and some of those users are going to these other places. This is how much they're spending, and that's money we're leaving on the table because we own that relationship.

It's the same reason Coinbase launched Base, although not that many users. What I would actually be curious about is whether Brian wants to come on and talk about it. The initial question I'd be really interested in asking him, and then Robinhood—eventually Vlad—is how it lined up to expectations after a year or 2.

I think Base took some strategic missteps by really pushing into creators and all that kind of stuff. Simple capital markets are coming on-chain. Lean into capital markets. By the way, I think they've adjusted their strategy. I do think they get that now, and I think there's an all-out war between Coinbase and Robinhood to become the leader. I'd put Kraken in this, too. I think Kraken's execution has been phenomenal under Arjun.

I firmly believe Robinhood settles 99.99% of its trades in a centralized database today. I firmly believe that number will get lower and lower over the years, and more and more will move on-chain. If you're Robinhood and you agree with what I'm saying, do you want that to flow to Ethereum, Base, or Solana? You're probably like, “Well, I actually want to own the chain.”

Yeah. You're probably like, “Well, I actually want to own the chain.”

But again, the crypto angle may be a topic for a broader conversation. I'd love to have a debate to really unpack what you just said there, because I feel like Robinhood's internal systems are pretty efficient. They're a database, and they track it. As soon as you introduce a chain, it has more surface area for things to go wrong, and I don't know if there's an efficiency to be had.

What is the efficiency there? Where is it going to show up in their P&L by launching this?

And, by the way, is it a chain? Can we call it a chain? If it's an entirely operated, validator-operated chain, are you just back to square 1 as a database that, I guess, rents some security from Ethereum? I don't know.

Yeah. Yeah. I don't know why Vlad and Brian are so excited about Base and Robinhood Chain if they're just single sequencers, or whatever.

Yeah. I mean—

But what I will say is, I was with the FOMO guys—Paul and Say, 2 of the 3 founders from FOMO. That business is a rocket ship right now. They were in the office the day I ended up recording a podcast with them; it comes out on Monday or Tuesday.

They showed me their user metrics on their phones. They added, I think, 12,000 or 15,000 users the day before. We're in a bear market. There are probably 100,000 crypto people who trade crypto in the world, and they're adding 15,000 a day.

Who are those users? They're people who want to trade things on-chain. They're not crypto natives. They're people coming from TikTok, Instagram, and similar platforms. As all these assets move on-chain—SpaceX pre-IPO, 5x-levered copper—they all move onto a blockchain.

You want to meet the users where they are.

So I actually think of it more from a user perspective. What does the customer want?

Yeah. The other aspect might be that Robinhood has had a relationship with Citadel. They're becoming Citadel in some ways by launching the Robinhood Chain. They're going to clip fees off the flow, and that might be the answer to all of this. Who knows what—

Flows all the way down. Yeah.

Yeah. Flows all the way down. And I think the thing about FOMO is that my appreciation for crypto is that you can make it so easy for people to create an account with 1 click and fund it with Apple Pay. If you're on a weekend and you want to trade oil because you saw some news, or you're at a party, you can onboard in a second.

That's not the case for traditional finance. Opening a bank account or an Interactive Brokers account takes days. Because markets are so real-time and momentum-driven, crypto's user onboarding is much, much faster. Companies will say—

Today, if you want to go long copper because a buddy of yours at a party told you to do it, you could do it with the click of a button. Then you can complete the onboarding later.

Totally. Totally.

I think that's what Robinhood is probably doing. They're going to onboard many more users who just want to trade options and everything else quickly, and they can onboard them much faster than the traditional route with their chain.

10. Regulatory Arbitrage Goes Onchain

Yeah. The other thing here is that there are regulatory reasons, too. If Robinhood wants to list an asset on the exchange, it's probably a total pain in the ass. I didn't fully know before, but I do know now that it is a pain in the ass. There's a listings team and a compliance team, and they're at odds with each other.

Not just at Robinhood—at every exchange, there's a listings team that wants to grow the P&L by listing more assets, and there's a compliance team that says, “No, no, no, you can't list that asset.” The second you move things on-chain, you can just list any asset that's on Uniswap. Or, if you're on Solana, FOMO can list any asset on FOMO that's on DFlow and Jupiter and routes through 0x. Suddenly, you've got every asset.

We're back to “it's always been regulatory arbitrage.” Someone should create a meme: “It's always been regulatory arbitrage.” It always has been.

Always has been. Oh, yeah.

By the way, it's not just that, but that's not the worst thing. Users should be able to trade any asset they want globally, and this system has gotten too privatized or too archaic.

Let me share this take from Mike, which I thought was quite a good take. Mike is the co-founder of Blockworks. He said, “Base and Robinhood Chain look similar, but it's possible that the early activity on each is very different. One thing that Coinbase has that Robinhood doesn't is a lot of users who sit on unrealized capital gains on Bitcoin and ETH.

“What this actually does—the dynamic here—is make Coinbase, and actually Base, a really great market for borrow-lend protocols like Morpho, where users who don't want to sell can borrow against their assets.”

Robinhood probably doesn’t have the same thing as much as Coinbase does, right? Coinbase has people who were buying Bitcoin in 2013, 2014, and 2015. Robinhood has more retail traders—maybe a little more degenerate than the Coinbase users.

So Coinbase, again, this is all Mike’s take, by virtue of having its L2 on Ethereum and a more conservative listings process—again, that’s the listings process we’re talking about—doesn’t have the same type of memecoin traders that Robinhood does. In this case, Mike’s final conclusion is that Coinbase might look a little bit more like the corporate chain on top of Ethereum, whereas Robinhood looks a little bit more like Solana.

I thought that was an interesting take, and I think I need to think about it a little more because I just read it. But I do tend to agree with him that the underlying users of these platforms will determine what the chain looks like.

Yeah. Does that then—

Which is why I don’t think Tempo will ultimately have any users. I don’t think Tempo will, because Stripe doesn’t have users. Stripe has businesses, and Tempo will build a beautiful blockchain for businesses that need to do payments, but I don’t think they’ll ever have any memecoin activity or DEX activity because there are no individual users.

We should have Dan Robinson or some of the team that has been architecting it. But didn’t they make a very specific design choice to optimize for stablecoins, primarily for B2B use cases, after they acquired Bridge? And so—

Yeah, I think about that stuff.

I think Dan Romero is now running Tempo.

Oh, yeah. But the thing about Robinhood that I found interesting—and I’m curious to get your thoughts—is that I saw a tweet from Vlad saying, “Super-long RWAs.” And then he was like, “Oh, by the way, the chain is also good for memes and memecoins.”

Because he gets it. He gets it.

This is the Robinhood strategy: Come to Robinhood. You should trade your S&P, your Qs. You should trade—just don’t buy the S&P, but also, by the way, I will give you GameStop. I will give you the GameStop mania. That was all on Robinhood. Vlad understands how important the retail—

Degenerate. You know, same—

I think we should stop calling it degenerate, to be honest. I think the reality is that it’s an affordability crisis. These are lottery tickets. Markets are more momentum-driven. Hedge funds are also conceding to that. This is just the new normal: adapt and accept.

I agree. I think that’s a good take.

That’s the reality of things, without too much emotion, because in crypto, the connotation of “degenerates”—I’ve met some really good crypto traders who do pretty good risk management, and they are investing at the frontier, in the trenches of Pump.

Yeah, yeah.

Can I tell you who gets this? Who do you know who gets this quite well? I think people may think of him as a memecoin trader or memecoin streamer, but he is so on the nose about this. By the way, yes, Ansem, but I’m thinking of Thread Guy, actually. I think Thread Guy is more on the nose about what is actually happening in the 25-and-under crowd and how they treat finance.

I saw a Thread Guy tweet about this. He said, “Everyone is all, ‘Robinhood is unserious,’ blah blah blah, but what if memes and crypto fundamentals are foundational to modern finance and never going away? Because clearly, it looks that way.” If memes haven’t died through retail getting absolutely rinsed over the last year from memes—

Yeah, Thread Guy might have a point here.

Sir, any sufficiently advanced technology feels and looks like magic. I would swap “advanced technology” with “memes.”

By the way, Thread Guy has an amazing piece on Twitter. You can just read that, and you’re like, “This guy gets it.” He’s introspective, he understands psychology, and I think in order to be a good investor, you just have to be on there.

There’s a good tweet I saw from someone that said, “If you want to build a retail brand and you don’t understand how 18-to-24-year-olds think and shop, you’re never going to make it—ever.” I think it’s true, and I think Thread Guy just gets it.

Yeah, totally.

Does that make the case that you should constantly be hiring younger people, or is age just not an issue? Can older people really get younger generations?

I think you have to constantly be hiring younger people.

By the way, the banks figured this out. I heard an interview—I think it was with Jamie Dimon, maybe—and he talked about the statistics of the 2-year analyst program. He said that of the people in the 2-year analyst program, only about 3% are still around in 10 years.

But if someone goes through the 2-year analyst program and sticks around for 10 years, they’re then a lifer at JPMorgan, and they tend to make it to the leadership levels. So you almost need some process internally at your company to get the 22-year-olds and 23-year-olds into your company and take bets on them.

Actually, speaking from experience, if I look at Blockworks, the best things ever at Blockworks—and specifically the best products—have come from people under 30 who really have their finger on the pulse. We have people in their 40s and people in their 50s—not to be dismissive—and we could not run Blockworks without them. They’re really amazing at helping scale the business.

Yeah. I’m not going to name names, but there are older people at Blockworks who we could not run Blockworks without. But it tends to be the 24-year-old who understands the market so well that drives forward something new, like the Token Transparency Framework.

I don’t know his age, but this guy Robert—I think he’s probably 23, 24, or 25, maybe—just gets it. He’s driving it forward. And, by the way, they’re so AI-native.

I was on a call with Robert yesterday, the guy who runs TTF, and I was like, “Hey, can we get this? Do you think it’s realistic to have this by Friday?” He was like, “I’ll have it to you by tonight.”

I was asking him for a massive amount of work. Then he was like, “Have you not used Fable?” I was like, “True.” As a manager, you have to update how you think about deadlines and requests. But if someone isn’t very AI-native and you update how you talk about requesting information and materials and deadlines, they’re going to think you’re crazy.

Right, right, right. That sounds like a great employee because he could very easily say, “Yeah, I’ll get it to you by Friday,” and then do it in 2 hours. But he gives you more. Employee productivity goes 10x.

Yeah, exactly. So, VVV—I’m sick of talking about VVV—but maybe just the dual-token equity structure. As someone who has probably angel-invested in more companies in crypto than maybe anybody except Balaji—more than Balaji.

11. The Token Equity Debate

Yeah, I mean, the number in Messari, I think, is understated. Let’s put it that way. I’m over 200—

At least, actually more.

Yeah. So, I mean, you’ve seen so many of these. You’ve seen every structure that exists in crypto investing.

What do you think of this dual model? Are you fine with it? There are 3 camps here that have emerged. One is, “I’m totally fine with it. It doesn’t really matter. You have to do what you have to do as a founder.”

The second is, “I’m not sure this actually makes sense, but in traditional capital markets, you have so many different structures. You have different types of bonds, different types of shares, common A, common B, preferred. Tokens and equity are fine; you just have to be clear with the market about what they represent.”

And then there’s a third bucket, which is that you need 1 vehicle that drives the value of your company, where someone can clearly bet on your company that way. It doesn’t matter that there are pure tokens out there. Because pure tokens exist, you’re in a bind. Anything that doesn’t look like a pure token is going to be discounted.

This is why, if you look at the best-performing projects, for better or for worse, there are 3 pure tokens out there: Ethereum, Solana, and Hyperliquid. There are counterfactuals, right? You could point to Ripple, for instance. They have Ripple-labeled equity. There are always examples that you can cherry-pick to make your argument, but I think unequivocally you want to have 1 instrument.

Carl Malone, for instance, has been really good at Liberty Media. Liberty Media has tracking stocks, and there are very interesting tax reasons why these instruments are useful. You want to go long Formula 1, where you can buy Tracking Stock A and Tracking Stock B and then have exposure. But it’s more for tax reasons, and it’s very niche.

For crypto, because it’s very retail-driven and because it’s very perception-, vibes-, and narrative-driven, you want to have 1 token. The meta is 1 token. It doesn’t matter why; it’s just a reflection of the regulatory environment and the lack of clarity that we still have. We have safe harbors and some of these things, but of course, I don’t think there’s anyone out there who, in a perfect world, would have anything other than 1 entity and 1 instrument.

Berkshire Hathaway did this extremely well. If you want to go long Berkshire, you buy the stock. That’s it. There’s nothing else. You can go buy Coke if you want, or buy the underlying positions. Markets are so psychological. There’s a study out there that says investors can’t understand more than 1 concept—1 narrative—so focus on the Schelling point. It should be 1 token.

I think any other project, whether it’s intentional or nefarious or value-extractive or not—look at all the DeFi protocols. They’re like, “Yeah, yeah, fee switches,” and we all understand why. But the market doesn’t care. They don’t care.

And Hyperliquid—the problem is that you have something like Hyperliquid that is just very clean. This is why I think L1s have been so successful: if you want to express a long view on an ecosystem, you buy Ethereum or ETH, or you buy SOL.

Like, end of story. That’s it. There’s a foundation that has some tokens, and then there are the labs, the builders, and the architecture. But Solana got caught in this whole debate, right?

Totally.

I don’t think we can disagree that there’s a reason why the Token Transparency Act is so important. You want to build trust. Trust is what gives a lot of value, and if you degrade trust—again, not to toot your horn—is there a world where you have 2 different assets, token and equity, and you have a team that has really good communication, really good disclosures and transparency, and abides by your standards? Could it thrive?

Absolutely. Yes, I could see a world where that happens, in the same way that Liberty Media has all these tracking stocks and is super buttoned-up in its disclosures and understands how to make it work. I don’t know if that answers the question, but it’s complicated.

It doesn’t matter, because tokens exist. You’re going to really have a hard, uphill battle fighting against something that doesn’t look like a token. And the market, again, it doesn’t matter what I think. The market sort of already spoke. The token’s down. End of story.

Yeah. What else? This week, Anoma launched a coin and Securitize went public. So there’s your—

And AI is just wonderful, you know.

A Fable 5.6 livestream is going on now. Sam Altman, in addition to the model, announced 3 major product things. One: ChatGPT Work. This is a competitor to Claude, obviously.

Okay.

Number 2 is a new ChatGPT desktop app. Number 3 is hosted sites.

Oh, wow. That’s great. You can just deploy through—

Just deploy. So maybe there’s no need to go through Vercel and Replit. Previously, you could use something like Vercel, Railway, and stuff like that.

Take a look at this. Zuckerberg is tweeting. I don’t think he’s tweeted in a long time.

“Quote-unquote, the pricing from some of the other labs is very extreme and has very high margins. We think that there’s a real ability to offer frontier or very high intelligence at a much more affordable cost.”

An epic pricing war is breaking out between these companies.

Interesting. Zuck tweeting. That is—

Tweeting. That was the last time he tweeted.

Zuck’s last tweet was in 2023. Before that, it was literally just a Spider-Man meme, and the tweet before that was in 2012. So this is only his second tweet in 14 years.

You know it’s real.

Oh, now he’s out here retweeting people. He’s addicted. He’s hooked. He’s hooked.

The CEO of Palo Alto Networks is also doing that. Honestly, let’s go to content of the week, because this goes to something that Jeremy Giffon said on the Invest Like the Best podcast.

I think we’re both having the same content of the week this week, which is Jeremy Giffon on Invest Like the Best. If you want to understand why Mark Zuckerberg and, you know, what the Palo Alto Networks CEO—

Anish?

Oh, God. Anyway, I saw him at the All-In Summit. He’s fantastic. It doesn’t matter his name. Palo Alto—if you want to know why Zuck is tweeting, why Palo Alto Networks’ CEO is tweeting once a day, why all these—why Twitter—

Nikesh Arora.

Nikesh Arora, by the way, is a good CEO and an unbelievable executor. I would try to get your hands on whatever that guy’s putting out, because he’s a phenomenal operator.

If you want to understand all this, and you want to understand why Twitter remains and will remain the single most important place in the world to build a personal brand, go listen to the Jeremy Giffon episode on Invest Like the Best.

By the way, that guy has more aphorisms than any one person. It’s like him and Will something or another.

Munger? Yeah.

Will—him and Will are the new priests.

They’re the new priests. What is it? The new priests of the world, not billionaires anymore. And I love what he just nonchalantly said. Everyone—

Why do you like the episode? Can you summarize it for me? Billionaires are dead, and priests are scientists. The age of the scientist is over. Go listen to it yourself.

You know what I experienced, to be honest, when I was working at the health startup? We had all these Nobel laureates, and we hosted a dinner. We had all kinds of billionaires showing up because they all want to live forever. That’s when you realize that money is not the end goal; it’s relevance.

Everyone wanted to be sitting next to the Nobel laureate—the Nobel Prize winner in chemistry. That’s when I understood that, at some point, you would rather have less money and more relevance. He really gets it.

Is there anything in that episode that stood out to you, or anything that you disagreed with?

Not much that I disagree with, to be honest. I think the guy’s just on it. He really understands it. He also talks about—

The thing I would disagree with is that being a billionaire is still something that 99.99% of the world would be thrilled about. And I think that if you live in the world that we live in, or you’re in the circles and you go to the dinners and the events, the All-In Summit, F1, you’re like, “Oh—

That guy’s just—

You attach the label: he’s a lowly billionaire. But—

I think that’s a very—

I think it’s important to remember that that’s not—

Actually how the world works.

I think, yeah—

What I do think he nailed—sorry, I keep cutting you off—is that most billionaires today do feel like they were important 5 or 10 years ago, and they are no longer. Their billionaire status no longer makes them relevant.

You need to post on Twitter to be relevant.

Yeah.

I would say there’s always been something like that, I think, because I’ve talked to people who were billionaires 10 years ago, and they would probably tell you the same thing. They just deeply—

Like—

Yeah.

And you want—and it’s all the signaling attached to that. I do wonder, in a world of abundance—

Is that it—

Like, in a world of abundance, if you go back to ancient Athenian times—

There’s a whole class—I think he mentions in the podcast—there was a time when people got paid to think. Being a public official was the highest honor.

Yeah. Being—

Being like—you had a class of philosophers like Plato and Aristotle. One of the more important things that he talks about, at risk of a spoiler alert, is this idea that we glorify the grind, and I’ve really experienced that move in Europe. I’ll leave it at that.

I think there is an art, and you really have to start questioning freedom—what it means. I’ve always felt output-driven; that’s what I—

And I think it’s—

But it’s very difficult, because you want to feel like you’re relevant, and for better or for worse, you want concrete evidence of that. There’s no better concrete evidence of that than time inputs—not so much outputs. Time input is such a concrete thing to say: “I work 100 hours a week. You deserve my respect, sir.”

It doesn’t matter. It doesn’t matter in a world where you can do it in 2 seconds.

Mhm.

My thought is that I agreed with most of that section. The counter to it is that I really genuinely do feel busier than I’ve ever felt in my life. And I think the reason for that is because—

Do you have a kid? [laughter] You have a baby at home. You’re cleaning diapers because of a child. No. Well, that, but also because AI is very deceptive in its ability to make you feel more productive.

It is the great enabler of starting new projects. I have a laundry list of things I’ve always wanted to do and build at Blockworks. I’ve never been able to build them because we’re constrained by people.

And now these things can actually take something from 0 to 60 in—and I’m not kidding—3 hours, 1 hour, or 2 days. But it’s very deceptive, because once you get that thing to the MVP stage, you still need a bunch of people, and you need to spend a bunch of time to do it.

So I think I—and many other founders—feel busier than ever because we're actually starting so many things. You probably saw this in our investor letter that I sent out last week, which is the lowlights of Q2. The thing that I think we're not doing well at Blockworks is we're doing way too many things. And this is probably an output of me and Mike thinking that AI can help us.

We're trying to push the boundaries of what AI can do internally, but that's probably the downside. But I think you've just summarized the bull case for AI, which is—you said something there that's so important: you're just going to launch way more creative experiments. And in order to execute them, see them through, and really scale them requires humans. And so it requires new skills. But it's not that you can take it from zero to 100. Zero to one, yeah, maybe. Zero to one to 100 requires humans, requires that.

And yeah, I'll say it is hard for me to go to sleep because I have a cabinet of ideas, and now I don't need an engineer. There's no degradation in the message and the output, and the output doesn't require $100,000, a consultant, and a month of work. It's like 2 hours, and you're like, “Holy, let's go.”

Holy. Yeah. Yeah. It's—wow. Yeah.

Yeah. My wife was like, “Why do you always leave your laptop open, and why are all of our computers still running overnight?”

Santi, good to have you back. You're looking very dapper today, by the way.

Thank you. Well, you know, just—

Seleni conference.

Seleni. Nice. They said business casual, so, you know, this is business.

So, the Solana conference is probably like 15 minutes from your door, huh?

Depends on how fast you drive.

Yeah. Depends. Awesome. Well, good to be with you guys this week. Monday or Tuesday, we’re releasing an episode with the FOMO founders, Paul and Say. Really interesting. There have not been many consumer businesses that have taken off recently, and they are just exploding. Really interesting episode. We talked a lot about marketing and customer acquisition.

What are you doing to that mic?

I’m not leaving.

Have a good Friday. Cheers. Great weekend, guys. Have a great day.