AI让人想起2017年的加密货币：Circle融资2.22亿美元，Anthropic收紧SPV管控

Nothing said on Empire is a recommendation to buy or sell any investments or products. This podcast is for informational purposes only and the views expressed by anyone on the show are solely their opinions, not financial advice or necessarily the views of Blockworks. Our hosts, guests, and the Blockworks team may hold positions in the companies, funds, or projects discussed. What's up everybody? Welcome back to Empire. Um sorry we missed you last week. I was at Consensus. Santi and Rob, I don't know what they were doing, but we couldn't record last week.

Where? Oh, you're at Milken? How's Milken?

Yeah, Milken was good, man. It was all the suits, right? Everyone in New York descends on LA to talk about what's happening in global finance. I will tell you that the conversation was basically about 2 things only: AI and tokenization. It's literally all anybody wants to talk about right now.

Which is good for us, I think.

Yeah, very good for us. Good for us.

Yeah, I mean, the suits didn't just descend on Milken; they descended on Miami, too. I was at Consensus. It was crazy. There was a State Street booth, a Morgan Stanley booth, a JPMorgan booth, a Visa booth, and a Mastercard booth. It was the most—I mean, basically, we had DAS in March and then Consensus in May, and they're both just very, very, very TradFi events.

Hats off to the Consensus team. I actually want to bring Tom from Bullish on the podcast. Tom had some interesting announcements. They actually tokenized the entire cap table of Bullish, which I thought was pretty cool. Hats off to the whole team.

Bought the transfer agent.

Yeah, and they bought the transfer agent for $4 billion. So, Santi, what's your excuse? Why weren't you showing up last week?

I don't know. I was waiting on you guys, and I was sitting there in my chair waiting for you both to show up, which is rare, I will say. Most of the time, it's me that has to cancel and move it around, but I was just looking at the news and some of the clips from both Milken and Consensus.

I got to say, I didn't really feel any FOMO not being at Consensus, especially at the after-party. There was a great video coming out of that.

The vibes were alive and well.

Yeah, what are you doing sponsoring a strip club event in 2026?

Only the AI companies can do that right now. You can't do that if you're in crypto. Rob, I want to go to you about the Milken conference. I saw some of the clips. Putting aside what's going on in AI land—you know, some of these stocks are just going absolutely bananas—and Stan Druckenmiller came out saying he's buying AI stocks. We talked last time on the pod about Paul Tudor Jones calling it a bubble.

I'm curious what the sentiment was, especially around tokenization and crypto. It seems like there's still a lot of apathy around token prices. But anyway, what were some of the more interesting conversations that you had out there?

1. AI Feels Like Crypto In 2017

Yeah, for the tokenization side, it's a lot of the same conversations that have been happening. Larry Fink has been saying on stage, and Jamie Dimon has actually been saying on stage recently, that the whole global economy is going to be tokenized. All assets are going to be tokenized. They're going to be put on a blockchain.

We've seen this start to take hold a little bit, with a lot of the trading firms also pushing for this. The DTCC is obviously doing its sandbox for tokenized clearing, and you're now seeing a lot of these companies, these asset managers, invest in some of these protocols that enable this world. Obviously, Arc and Circle came out with their fundraise on the Arc side. There's been stuff announced around Ondo recently with, I believe, JPMorgan, and with Morpho and others.

It's become very clear that people expect that assets will be tokenized and assets will trade on blockchains. How and where that happens, whether there's a central clearing agent, whether there's still a transfer agent, or whether there can be more direct access to a blockchain itself, is still very much up for debate.

I think it's still very clear to people that the current set of regulations makes them need something like a transfer agent. It makes them need somebody who sits in the middle as an intermediary. But if you're going to trade on the weekends, if you're going to trade overnight, you are going to be on-chain. That's the only way to manage risk during those things.

Every conversation I had was really about those 2 things: AI or tokenization. I will say that the tokenization efforts and conversation now feel more like, “Oh, this is happening. We're doing this thing,” and this is great for business.

Also, we don't care at all about what's happening in token prices. I couldn't even tell you what exists beyond Ethereum and Bitcoin. They don't care about that because it doesn't matter for them. They see this as underlying settlement infrastructure.

The AI piece, I think, has now become a really hot-button topic because there are all these people who are like, “Oh, this is a bubble,” and there are all these people who are like, “We're in a supercycle.” Nobody knows anything whatsoever. It's created a sense of needing to get smarter, right? A lot of people are feeling unsettled about what the future is going to look like on the AI side, whereas tokenization feels settled.

Can we start with AI instead of tokenization? I want to go there in a little bit, but can I hear your take on these AI stocks and what you're doing? It feels like we've entered the part of the cycle where all your friends are getting hilariously rich and you're not.

It really reminds me of 2017, when even the most foolish of your friends were just longing Bitcoin and making a 2× in literally a week. Micron is up 80% in the last month. Intel is up 85%. SanDisk is up, I think, 50% in the last month.

I think you've reached the stage of AI mania where my whole feed, at least, is filled with people who are getting, as they post on Twitter, hilariously rich and you're not. I'm very curious. Santi, maybe starting with you, what are you doing?

Well, I've put in a big position in some of these names. Shout-out to Logan, who's one of the guys I always go to to understand more of the physics piece. Logan from Preston's Lessons and I were chatting.

It's hard to look at a chart and want to put in a position after it's been parabolic. SanDisk was spun out of Western Digital, and I think, in isolation, it's just been in price discovery. But in general, there's this memory shortage, and memory prices have gone up dramatically.

I won't go into the specifics, but the short of it is that memory has historically been extremely cyclical because you have these boom-and-busts. Historically, it's been smartphones: anytime there's a smartphone, you have a huge ramp-up in CapEx, then there's a slowdown, and then you get crushed.

Memory and these memory names—SanDisk, Intel, Samsung, and Micron—have all been extremely cyclical. It's just semiconductors as a category.

The narrative now is that, with agentic workflows that require more memory, anyone who's used AI understands that the context window, what's called KV cache, is very limited. There are a lot of issues around that, and I think anyone can sympathize with that. That requires a lot of memory.

They call it hot memory. You want memory that is easily cached, with a lot of the weights and the context window. A lot of the run-up and excitement around memory stocks is that what has historically been a cyclical industry is moving away and becoming less cyclical.

If you believe that agentic AI is just going to become more prevalent and agentic workloads are going to require more memory, then this starts becoming more interesting. Micron reported earnings last week. I think SanDisk also reported earnings. The more interesting data points there are that a lot of these guys have 5-year contracts.

If you look at the growth baked into these things, they're trading at sub-10× forward P/E. You have everyone coming on board. You have Brad Gerstner from Altimeter coming out saying they own a bunch of these, that memory is no longer going to be as cyclical as it used to be, so it requires a re-rating.

You have the Coatue guys coming into the fold. Everyone's talking about it. I do think that it's real. The things that I'm paying attention to are obviously CapEx. Historically, there's been a huge ramp-up in CapEx and building factories to fulfill orders.

And then you hit an oversupply, and prices collapse. I’m also kind of monitoring the price of memory. There are some really good accounts that I can just post in the chat, but it’s certainly interesting. It’s just a classic rerating.

Even at these prices—I think it was the Altimeter guy who came out and said, “Even if you look at the chart, you would say, ‘Yeah, this is a bubble,’ but even at these prices, these multiples are absurdly cheap.”

Let’s say you’re listening to this and you don’t have a position in these things. The memory stocks are ripping because people realize you don’t just need the raw GPU processing power; you need massive amounts of physical memory. Let’s say you agree with that and you’re totally AI-pilled, but you don’t have a position. You see that the charts look extremely bubbly, but also that the Nvidia chart looked bubbly 3 years ago and it kept going up. What do you do?

Santi, let’s say you have 0 position right now and you’re just starting fresh. You’re looking at these things. How do you think about putting on a position right now?

Well, the same is true for Bitcoin, too. If you traded crypto or invested in crypto, sometimes it’s best not to look at a chart. You could look at RSI, the 50-day moving average, and the 200-day moving average to scale into your positions if you want to play cute.

The reality is, you either believe that this thing is going to persist, that memory is going to become a huge, persistent demand, and that the demand is there, or you don’t. You just wrap your head around the thesis and then you go long.

I’ll tell you what I did. I didn’t have a position a month ago. The way I structure these things, I go long, and then I also do other structural products to clip in a coupon. I just don’t think you’re ever going to time an entry perfectly. If you’re going to wait for a perfect entry, you’re just going to be sidelined. There’s huge volatility in these names.

One thing about NVIDIA that you said is true, which is that the best take I’ve seen is that NVIDIA went from being highly cyclical to being the most valuable company in the world. The demand for GPUs was very cyclical, and now it’s just up and to the right. The demand is really, really strong, and they continue to innovate.

You have the Cerebras IPO coming up, which has been massively oversubscribed.

Cerebras, yeah.

Yeah, massively oversubscribed. I was reading some really good takes about the difference between Cerebras and NVIDIA. The more nuanced debate that is happening is: Where is the demand for AI going to be? Is it just for training? Is it for inference? A lot of the memory trade has been around inference.

The demand is just going to continue to go up and to the right because of the generative workflows. There’s this DRAM ETF that launched around the memory trade. I think it’s been the fastest-growing ETF in history. I think it’s had over $5 billion of inflows. The thing has gone up about 2x in 1 month. It’s a very concentrated industry.

Clearly, everyone has been sidelined because they’re nervous about what happened in the dot-com crash and the cloud-computing hyper-buildout and crash. It’s been a very, very cyclical industry. Again, I think it’s tough to look at a chart and then put on a position.

I don’t like to time my entries perfectly. I’d rather look at the fundamental case for supply and demand. Is this going to be cyclical, or is it going to persist over the next 5–10 years? I think the same is true for crypto. That’s sort of my take.

Are you playing some of the more decentralized versions of this?

I think so. Venice is up—what is it up, 186% in the last week? Something like that. It’s not just that the token is up because of a meme. It’s over 2 million users now, and it’s growing very, very quickly. I think they said they did almost 30 million visits in Q1.

They’re doing over 2 million—the 2 million number earlier was daily actives. I think they’re at that point now, right? There’s very clearly a demand for uncensored models.

Maybe you guys saw it, but Hermes flipped OpenClaw as well, right? Hermes is on OpenRouter. To the point that you made, Santi, earlier about memory and inference, I’ve been playing around with Hermes, and it is so much better than OpenClaw.

One of the main things is that it pauses every 15 seconds and looks back at the context window and says, “What did I do well? What didn’t I do well? Do I need to improve, and how do I self-learn? How do I do better?” It’s this memory and this context about working with you, the feedback it’s gotten, and self-reflection. It makes it significantly, significantly easier to use for a bunch of different types of skills and workflows. There’s going to be more memory needed because of that.

On the Venice side, too, that’s demand for uncensored models, but their biggest model, I think, is still actually Opus. Even though there’s demand otherwise, that’s again a bet on inference, right? To your point, Santi, it’s a bet that more and more people are going to be using inference over time.

Venice is mostly American, but you’re actually seeing a lot of this happening in Asia as well with Hermes. In that world, I actually think there’s a crypto angle, especially on the retail side, that’s interesting.

Yeah. Ben from Stratechery had a really good piece. I’ll quote him. He says, “The most important aspect for answer inference is token speed. The most important aspect for agentic inference, however, is memory.”

I think that is a shift that is happening. If you look at the evolution of AI, first it was GPU buildout, then it was prompting. Right now, it’s turning over to agentic workflows. For that, you need to have a bigger KB cache.

A lot of the conversations I had with Logan—not to get too technical—were about this DeepSeek paper that came out. They were doing a lot of interesting optimizations that bode well for a company like SanDisk. You have to look at the different tiers of memory, from hot to cold. The pricing is very different.

I’m not an expert. What I try to tell you is that I do pay attention when smart people in my network reach out to me and say, “Hey, look, we’re paying a lot of attention to this, especially in how Chinese-constrained companies like DeepSeek have tried to innovate around bottlenecks in this AI supply chain.”

Needless to say, I rotated a lot of my cash and just blindly bought SanDisk and Micron, and of course my bank—

So you didn’t buy Venice? I don’t understand Venice well enough.

I’m fairly concentrated. Of course, I get a call and they’re like, “Hey, are you sure you really want to do this?” And I’m like, “Just do it.” I think this continues, but I may be terribly wrong. None of this is financial advice.

When you buy SanDisk and Micron, are you buying just the stock, or are you buying some complex—

I’m probably doing it wrong, but I did buy spot and scaled into the trade. Then I bought these autocall notes. It’s a very volatile name, so I get paid a guaranteed coupon of around 50% on something like SanDisk, and the barrier is around 30%.

If it breaches that—if it goes down below 30% in 12 months—I basically buy it at that level. So if it goes down 60%, I obviously eat that, but the coupon is so high that it basically has to go down 40%, and I’m happy to own it at that level. In the meantime, I’m getting paid a nice coupon.

I did both, but that’s not my full-time job. I did that very quickly and said, “I’m just going to forget it.” I think this continues, but I may be terribly wrong. None of this is financial advice.

We touched on an interesting point here, but maybe, Jason, one last point. I was talking to Tom Schmidt, one of my partners here at Dragonfly, and he’s been traveling around Asia for 1 or 2 weeks. He went through China last week, and he was talking about how he’s going to meetups where people are literally going to Hermes or OpenClaw meetups.

They’re in parks and random places, and you just get 3,000 people to show up. Then there are people helping them set up their own agent there. He was talking about how a lot of people in China were saying that this feels a lot like crypto in 2017 in China, where everybody was just meeting up and figuring out what they could do.

It’s super interesting. There’s so much excitement about it, and everybody’s talking about it.

If I think about the stuff I care about, which is a lot of digital payments, right? Digital payments grew so much more quickly initially in Asia, China being part of it, but then more broadly in Asia, and then it went to the rest of the world. Right? And it does feel like there’s already this excitement around Hermes and OpenClaw and this agentic future, but it’s still very much a situation where whales eat up the vast majority of tokens, and then there’s everybody else.

Hermes is still, even though I thought it was a lot easier than OpenClaw, way too hard for the vast majority of people to use, and Brian Chesky’s been out talking about this a lot. I’ve really enjoyed his stuff. It does feel like the early days of having a bunch of agents, right? And then all of the things that are downstream from that—to your point, Santi.

Yianni, you’re pretty at the forefront of this stuff and testing it. I’m curious: I have been tinkering around with agents, but how much do you guys use them? Yeah, I’m curious how you guys are using agents, to what degree, and for what.

So, what we realized at the company was—we did the survey internally at Blockworks. 81% of Blockworks was using Claude Code, while 19% was using chat. 96% of Blockworks was heavily using AI, and 4% were not heavily using AI. That was 3 months ago. I’d guess it’s probably 100% now.

What we realized was that most people were just using Claude Code or Codex, but they hadn’t taken the time to do these setups to make the work really go to the next level. So, we moved one of our research analysts into this position. I don’t know his title, but it’s like—we can call it head of AI, head of AI excellency, or whatever you want to call it.

He’s basically put together a SWAT team where he now looks at every single function the company does. I think Claude Code and Codex are the V1 inside of companies. Maybe chatbots are the V1, and Codex and Claude Code are the V2. We’re now doing a lot of the V3 stuff, which is that we’ve built these agents that run a lot of subagents for all of our tasks.

That could be outbound sales, putting together research reports, or building dashboards. The time it takes us to build a dashboard has gone down by about 95%. The time it takes us to add new data to our APIs is now nearly fully automated. You can’t just do that with Claude Code and Codex. You have to build these agents that are running subagents.

That’s why, Rob, you mentioned Hermes. Hermes is the version of this on the personal front. I think that’s where the future goes here: you have these agents that have subagents. On Hermes, I actually haven’t set it up, but my understanding is it’s backed by Nous Research, right? Nous Research, which—

Yeah, it’s a crypto team. I think Paradigm led their Series A last year, maybe. But I’m curious: I was trying to dig into the docs, and it looks like the Hermes language models are trained on the Psyche Network. I pulled this up earlier. The Psyche Network is an AI training infrastructure on top of Solana, so I’m very confused by that.

For that piece, I think they’re winding that down, as I understand it. Today, when you set up a Hermes agent, you just pick whatever model. I ended up using Opus for it specifically. There is the Hermes model, but the model is not the focus of the business anymore because they’ve had the agentic framework stuff grow so quickly.

I think for a lot of these guys doing agentic infrastructure, they still don’t really know how to monetize their business. They’re definitely more like frameworks, and it’s open source, and there are a lot of people using it, but how do you—what does that mean for an actual business?

OpenClaw figured this out because they had no business whatsoever, and it was one dude. Then they ended up exiting for $1 billion, just because of what they had built around it. How this turns into revenue is still very unclear.

Did you read Toby from Shopify’s Twitter post about what they’re doing there? They’re not allowing anybody to have private conversations with AI. All of the AI is done in a public Slack channel. He’s like, “We’re creating this sort of context window for everybody,” and so people are learning internally. Did you read it, and what did you think of that, if you did read it?

I think that’s the right move. Yeah, I did read it, and I think that’s the right move. We haven’t gone that far. People can still have private AI conversations at Blockworks, because it actually takes a lot of time to do the security of this stuff.

Theoretically, you should give each employee, when they onboard, an API key that shows their level of access. I don’t want one of our salespeople having full access to our financials, but I want the finance team having full access. So, you have to set that up. It takes time—you’ve got to set that up. Candidly, we haven’t done that.

You’re not thrilled?

Yeah, that’s the right model. Look, I think there was this big push after COVID to go back IRL. Everyone’s going back into the office. I think there’s a big benefit to that, but we didn’t do that. We’re still a remote-first company, and I think you’re starting to see the real benefits again of being a remote-first company today.

Every single thing that Blockworks does is documented. One thing that we made a big push about a year and a half ago was to go to public channels instead of private in Slack. What’s happening now is that the AI just has so much more context because it can’t read private Slack channels, but it can read public ones.

I would encourage founders, if they’re listening to this, to make all their Slack channels open and public by default. It doesn’t mean you have to add the whole company into these Slack channels, but just record everything at the company.

I think you’re seeing this with Shopify and Facebook. Facebook is now logging every keystroke of all employees as a training method for its AI. There are these RLHF efforts—they bought Scale. Mercor is another one.

Instead of outsourcing your training, what if you did it in-house with just employees? We’re not doing that, either. But I think that’s where a lot of these companies will go: starting to do more and more training.

2. Circle Raises $222M

Right now, everyone outsources their training to the big companies, but I think you’ll start to see more training pushed to the edges. The way that you can do that is by documenting more stuff and making more stuff open. So, yeah, that’s what we’re doing, to answer your question, Rob.

Let’s shift gears. We spent 30 minutes on AI. Digital Asset, the company behind Canton, came out—it’s raising a big round led by Andreessen. We can talk about all of this; I’m not sure how much is public and what’s not, but we can talk about that. And then there’s Circle—

That was just a rumor, and there was very little that was public around it. So, I’m not sure there’s much to say around it until more stuff comes out.

Well, it sounds like they are, is what I’d say. Circle is—not earnings—but I want to talk about Circle raising $222 million for Arc, the token for their chain.

The broad idea here is that I mentioned last week or 2 weeks ago that it’s a good time to be a public enterprise building a chain. I think I mentioned Canton and Tempo specifically, but if you look at right now, you’ve got Digital Asset and Canton raising at $2 billion; Circle and Arc raising at $3 billion; and Stripe, Paradigm, and Tempo. I’m not sure what’s public there and what’s not public, but I’m sure Tempo will have a token at some point.

So, yeah, it’s a good time to be a suit selling a token for a chain, is what I’d say. I’m curious to get your take.

What’s interesting is Circle raising at $3 billion. Tempo’s last round, mostly led by Stripe insiders, was at $5 billion, if I recall.

That’s right.

I don’t know how much we want to get into Circle and Arc. I’ve heard some takes around it actually doing less good or more harm than good by being less like—

Let’s talk about Circle specifically, because Stripe is a private company, and Paradigm behind Tempo is a venture fund. So, this is kind of the equity of Tempo. It kind of makes sense. The thing that you can bet on if you want to bet on their digital asset is a private company. Canton is a public token. Sure, that kind of makes sense.

Circle is a publicly traded equity, though.

Well, to be clear, the digital-asset fundraises, both this one and the one before, are just into the equity, not into the token. The way that they've described it is that you get token exposure through the tokens on the balance sheet. But the investment—all the investments—have not included token warrants. They've just been traded into the equity.

Got it. Circle's interesting, though, because Circle's a publicly traded stock. If I am a Circle shareholder, I am confused why there's now 2 ways to bet on the company.

Well, I'm not sure public investors are really making that nuance, to be honest. It's more like they're going to have some tokens on the balance sheet, so you still accrue some of that benefit. I'm pretty sure that's the way it's structured, Rob. I don't know if you just—

It's the same story a little bit as Canton, right? Like, 25% of the tokens go directly to the balance sheet, right? And so, okay, well, you get some exposure through—you own 25% of the tokens. And then there's a very big piece that goes to all of the partners and goes to incentivizing economic activity, right?

I think honestly the most interesting through line here is what this means for the Coinbase-Circle relationship, actually. I don't know for a fact that this is true, but here's what I do know is true: in the Circle-Coinbase deal, if USDC sits within a Circle wallet or Circle account, or at Circle, then they don't have to pay the 50% revenue share on the net interest margin to Coinbase, right?

And so the thesis has been, I think, for a lot of people—and I think the article that they put out kind of says some of this, right—which is that Arc is part of Circle. If they can move a lot of economic activity into Arc, that is actually an immediate uplift in economics to Circle because maybe there's—you know, USDC is the most dominant stablecoin in DeFi, and DeFi currently exists on Ethereum, Base, and Solana.

Well, let's say a lot of that DeFi economic activity moves to Arc. They get—call it net zero, right? And if you're not expanding the pie, you're actually just cannibalizing what exists elsewhere. Now they've gotten double the revenue because they've gone from a 50% revenue share with Coinbase to 100%. And that's something that, if you're a public-market investor in the equity, you obviously like a lot, right? Because there's a huge uplift for you, right?

You saw that they beat earnings. The shares traded up—I forget what it was, but I think they traded up quite a bit—after hours and then yesterday. The bet here is almost certainly that they can take these other things that they have been doing around CPN, which has been this sort of Circle Payments Network that I don't think has had the success that they had hoped it would have.

They can take that, more of the work they've done on the cross-chain side, and more of these value-added services, and then move a lot of the activity that's happening on-chain, on which they are giving up a bunch of the economics of USDC, onto their own chain. It's good for both the equity and the token.

The token is needed to incentivize people to come and be part of that ecosystem. They haven't fully talked publicly yet about what use case the token will have, but it is going to have usage—or at least they have intended that it's going to have usage—that will incentivize people to hold it and do things on Arc.

So it's a customer-acquisition cost. It's marketing.

But that's true for a lot of tokens. Tokens are just CAC in a lot of cases.

True. It's showing up in the numbers for sure. They reported earnings—I mean, their OpEx line item was up almost 80% year over year. Stock-based compensation is up 4x. I think, reading into that, my take is they're obviously spending a lot to get this thing off the ground, so shareholders are definitely footing the bill there.

And they're getting some benefit. I actually don't know. This goes back to an earlier point I want to get your guys' take on: Arc on its own, is it going to be hugely profitable as a traditional L1? Because if it's just stablecoin and stablecoin transfers, there's not, in my opinion—if it's just that, I don't think it's that interesting.

I'm curious, of course it depends on the valuation, but would you rather own Circle the stock or Arc the token, if it exists? It's nuanced, right? It depends if it's $3 billion, $10 billion, or $40 billion. Of course it depends, but I'm curious about the correlation between the stock and the token, and what you would rather own and why.

I mean, if Arc is successful and everything I just talked about, it has a direct economic benefit to the equity, right? And so I think the debate that you have to have—which I've been having, actually, with a bunch of people—is, okay, well, what do we think the net-new buyer is for a token today? How many buyers of those tokens do we think actually exist?

In this round, there were obviously a ton of strategics who said that they did the private sale. Does that mean that there's going to be institutional buyers of the Arc token over time if the Arc network does well? That's the question you have to ask yourself, because the amount of capital that could go into the equity over time, if USDC continues to take market share and dominate, is obviously a much bigger pool of capital. There are many more buyers that you can reach than there are today of people who can hold this token.

And so there's a supply-demand question around that, right? Now, does that mean that the token, on that supply constraint, won't perform just as well, or shouldn't perform just as well? I think today you've seen that correlation break down a little bit, right?

Pump.fun's a really interesting example because they did $700 million of revenue last year, and they're doing $350 million of revenue this year. They were doing a 100% buyback and burn before. They came out and announced that they were going from a 100% buyback and burn to a 50% buyback and burn, and the token went up 30%.

And so it was unclear whether the market even understood what was happening. The crypto markets have been a lot more cyclical, and so I think in the equity, anyways, you're betting on more—call it—the long-term trend of stablecoins. On the Arc token, you're betting on their ability to win DeFi and this pie that exists today in the crypto space, which is smaller. How you square those 2, I think, for any investor, that's their job. That's what you have to figure out.

The way I like to frame the question—I want to go to you, Yano—is that Arc is the asset, the token itself; USDC is a commodity it transports. It depends on what you want to have exposure to.

But the commentary I've heard is that you obviously saw Hyperliquid do an RFP, which was very competitive, that Circle also wanted to compete in. Hyperliquid, which I think is at the forefront of this, has just been very good at thinking about go-to-market and marketing and how to build more modes.

They did their own RFP for a stablecoin, and they issued their own native stablecoin. That doesn't mean that USDC isn't going to be on the chain, but I read into that and I think—

Still the largest stablecoin on Hyperliquid by far.

USDC? Yeah. Yeah, yeah. Well, that's what I'm trying to get at: has Jeremy and the team come to the conclusion that USDC is sufficiently entrenched in the crypto ecosystem that they can now go and launch Arc?

I don't think they could have done that in the earlier days, but they're seeing, okay, we're gaining more market share, especially on the B2B side—on the business side. We sort of have the brand and the trust. We can now go out and launch Arc.

Sure, some of the other chains, like Solana and Ethereum and some of the L2s, are probably going to feel the heat, but we're at a point where we launch Arc and still dominate. I think that's sort of the calculus that probably went through their minds.

Yeah, I think that's right. And I think Rob's point around their relationship with Coinbase—I think, Rob, you said this—also Omar from Dragonfly had a really good post on this, too.

This is the beginning of the end. Actually, probably the beginning of the end was probably a year ago, but this is a continuation of the end of the Coinbase-Circle relationship, right? We saw them launch competitive products like cbBTC, and then Circle launched CCTP.

But if you look at what Coinbase is trying to do with Coinbase plus Base, it's to build the everything exchange. I think this is Circle saying we're going to take all that activity, and instead of flowing it into Coinbase and Base—USDC balances, tokenized assets, payments, settlement, eventually FX, which will probably get big in crypto—that should all go onto our chain, not onto Base.

And I think this is Circle saying it should flow onto our thing. So I like it from that point of view.

Yeah, go ahead.

Well, no, I just want to add one point to that before I want you to continue, which is that there was this ANPRM, an advance notice of proposed rulemaking, that came out from Treasury around the GENIUS Act. Obviously, as part of GENIUS, there had to be a bunch of rulemaking that the OCC and Treasury have to do.

Coinbase filed essentially a comment letter when this request came out: Here’s what we think about the rules and what they should be. One of the things that Coinbase put in their filing, which I think I a little bit underreported, was support for one entity that is able to reach the GENIUS Act level of an issuer and then other entities doing multi-token or multi-stablecoin issuance.

What that implies to me is that Coinbase believes in the future they will be a multistablecoin issuer of different types of stablecoins, which also implies to you the same thing you just said, Yano, about the relationship with Circle.

Exactly. They own the distribution, they can have multiple stablecoins there, and just take massive revenue shares on multiple ones. In that case, I like it for Circle; it’s strategic. But I really don’t like it from a token perspective. What’s the value of a token?

It’s like Arc at $3 billion. Look, they’re able to get some smart people to do this, right? Rob, I’d be curious why you guys aren’t included in here and why you guys passed. But they got Andreessen, Apollo, BlackRock, General Catalyst, ICE, Janus Henderson, Marshall Wace, and Standard Chartered.

I have my own takes on why those folks did that, but it was a very successful fundraise, I’d say. A $2.2 billion to $3 billion valuation is nothing to turn your head at, but it’s 10% of Circle’s market cap, give or take.

Yeah, it’s a very successful fundraise, and I like the strategy. I will just say, so many people over the last decade have bet against Jeremy Allaire and have said, “Jeremy is this,” or “Jeremy is this,” and the guy just wins. He keeps being successful. Everyone said the IPO was going to be the worst IPO ever. The thing ripped.

I’m not going to bet against him. What I will say is that I think the trend of the industry has to be that there is 1 single thing that captures the value of your company. It could either be equity or it could be a token. So I don’t like having those 2 vehicles.

My vision for this stuff is that I really do believe in tokens, but I think these things will just converge into 1. I think either all equities will trade on-chain or tokens will become the new version of equities, but you can’t have 2 forms of value capture for a company. I feel pretty strongly about that.

That’s my take. I’m obviously going to debate this with Haseeb tomorrow. I think we’re opening up the conference here about whether tokens are dead or not. Unfortunately, we won’t be able to share that because it’s closed doors, but I’m excited to do that.

You’re arguing for sure that tokens are dead, right?

Yeah, I’m taking the affirmative. Haseeb is taking the negative. So, in any case, would the right analogy here be an oil exploration company, give or take? They have the rails, they own the pipes or the infrastructure, and then they have oil on their balance sheet. I don’t know if that’s the right analogy.

No, no, no. It would be like what they’re doing here is, if you’re an oil company, you’re Exxon and you have Exxon stock—that’s Circle stock—but then Exxon launched another stock just for the oil pipes. I’m like, that doesn’t make sense. Just have 1.

I’ll look into this, but obviously, in the Born to Be Wired book, John Malone experimented a lot with tracking stocks. The way he did that was, you have Liberty A and Liberty B, because investors want exposure to different things. I’m not sure how Jeremy—there is a little bit of financial engineering here.

Again, I think there is a place, a role, for a token to serve. You need to have a token. I don’t think tokens will cease to exist. There’s a role for a token, especially for an L1, to exist, to pay for settlement transactions and MEV, block inclusion, and priority fees. You can’t have a stock to do that, so you’re still going to have a token.

Was that the primary motivation, or was it just Jeremy saying, “Look, there’s clearly interest,” looking at Stripe and what they’ve done, and saying, “Okay, let’s do this ourselves and capture the attention and have some of these new investors subsidize a lot of the OpEx”? Probably, maybe. The question here is, why didn’t Coinbase do this for Base?

Yeah, listen, I think there’s obviously a question around Arc being an L1, while Base is an OP Stack L2 rollup. There’s also a question around whether an OP rollup actually needs a token, whereas, to your point, it’s much clearer why an L1 needs a token.

There are obviously a bunch of L1s that are trying to pay gas in stablecoins, but I think it makes a lot more sense from how to align economic incentives in an L1 than it does in an L2.

What’s also true here is that we’re in this time where you can be one of the first public companies to go and launch a token. What is Jeremy other than somebody who’s wanted to push the edges of what happens out there in public markets, what happens out there in private markets, and what’s happening in technology more broadly? That’s been his whole career.

Well, yeah, and by the way, let’s not forget Circle and Base moved away from the OP Stack earlier this year. So again, is this part of the evolution where they came to the conclusion, “Hey, we don’t necessarily need to be an L2. We’d rather just build an L1 ourselves and own our own sovereignty”?

3. The Clarity Act

Related to this, a piece of news I wanted to cover related to what you just said about perhaps Jeremy and the willingness to be more on the cutting edge of regulations and doing interesting things: We have here the Polymarket odds on the CLARITY Act. When we recorded 2 weeks ago, it was at a record low of 30%, but basically you’re now at 60%—what is it?—a nearly two-thirds chance the CLARITY Act gets signed into law this year.

Wait, wait, Santi, say that again. What’s the percentage?

60%? Okay, cool. Wow, listeners could have made some nice money listening to your boy.

Didn’t you say that, Yana?

2 weeks ago, I said we’re getting clarity. By the way, we should talk about clarity. I’m still at 98% that we’re getting clarity. I think there’s been a big change in the last week. Rob, I don’t know if you’re capitulating on that trade. Rob was saying we’re not getting it. I’m saying we’re getting it. I’m standing by that.

I did not say we’re not getting it. I said it was mispriced in the market and that the market was—

Which, by the way, Santi, everybody—Rob—

No, no, we’re going to have to do a clip here. I think we said both you and I—

Every single insider had the same opinion, right? Now, to your point, there has been a big change in the last week. I don’t know if we want to move to this right now, but Tillis basically came out and was like, “We are going to put this out there. We’re going to be happy to do it in a bipartisan vote if that’s what it means in the Senate Banking Committee.”

It doesn’t matter that the banks are going to continue to throw up a fight, because I’m fed up and I believe that we have a reasonable compromise.

It was unclear that would happen, and a lot of the back-and-forth, by the way, was based on the belief that the banks were not going to get to a compromise and play in good faith, right? What has actually proven out to be true is that exact thing. The banks have not been acting in good faith.

You saw the ABA put out an email blast to every single member over the weekend, trying to get everybody to flood Congress with comments. It says, “Hey, this is bad for consumers. It’s bad for deposit flight.” They’re putting up a huge fight.

Now it does feel a little bit like a death rattle, to Yiannis's point, but the banks have essentially been acting in bad faith this entire time. What we needed was for Tillis to come and say, “Hey, I hear you. I think we have a good compromise, and I’m not going to continue to be beholden to the private interest groups of the banks.”

That’s what he did. The big change was that Congress said, “We think there is a good agreement in place. We have heard everybody. We understand the concerns. We are trying to come to a place where we can address those concerns, and now we’re going to put this on the floor.”

This is not done yet, though.

To Yiannis's point about 98%, that's definitely a mispricing as well, because we're going to get out of committee probably—almost certainly—but it might still very much happen on a partisan basis, which is not typically a good sign for what happens on the floor, right? Then you're going to go to the floor, and 1 or 2 flips on the Republican side could create big issues for this, right? There's going to be a lot of whipping that needs to be done. There's going to be a lot of work that needs to be done. The ABA and the other interest groups on the bank side are going to put up a lot of a fight.

But I will say that the prospects of a Clarity Act now seem much, much larger than they did a week, week and a half ago. And it does seem very likely that, if not close to 98%, to Yiannis's point, we at least get a vote on the Senate floor. Which is really not being talked about enough, I think, right? I mean, I don't know, I think crypto asset prices did kind of rally on that basis. I saw a lot of news out there, and I think you saw Bitcoin above 80K; some of the other major networks caught a bid. And I think people were willing to be a little bit more risk-on.

I heard it from, I think, Rebecca, that said a bill dies like 7 times before it gets passed. I'm actually now of the mind that it does get passed in 2026. Maybe not the version that you wanted, but it will get passed in some way, shape, or form, and that's going to be positive. So I think it's still mispriced at 60% off. I would be operating under the assumption it does get passed.

Yeah. I mean, I really think it'll get passed. It might be helpful. Did you trade it, Yiannis?

What matters is, I didn't trade it. Unlike most of crypto, I'm terrified of insider trading. It's not worth the risk. So, could you trade it? Probably. Is it worth it? Probably not.

But I think I did say on the podcast, “I think we're going to get clarity. People aren't expecting it. I think we're going to get clarity, and I think we'll get a pump into a re-rating of some crypto assets into positive news about clarity.” And I think we got that. I didn't trade it, sadly.

Do you want to give an overview of what Clarity—the impact of Clarity, and specifically the impact of Clarity on tokens—or I'm happy to as well?

Yeah, I mean, I can. What matters to us, I think, is that a lot of what is going to end up happening with Clarity—and the same thing that we've had happen with GENIUS—is that we're going to punt a lot of rulemaking still. So we're not going to know the full effect until the SEC and the CFTC get together and take this bill a layer deeper and say, “Okay, now here's how we define a commodity versus a security,” right?

But what it does do is provide us a framework to say, “This is a separate category, and now we're going to go define what is a commodity versus a security.” There are some guidelines in Clarity already, but the devil's in the details, right?

What's really good about that specifically is that now all of these—call it banks, traditional finance companies, tech companies—who have been saying, “I want to be able to do something where I hold different types of crypto assets on-chain,” aren't going to have this concern over what this thing is anymore, right? Which is going to provide them clarity to be able to trade them, hold them, create new things themselves, and potentially create—or at least, with some certainty, create—an on-chain security, which you can do today, obviously, if you go through traditional intermediaries. But maybe there's going to be a lot more flexibility around that over time. Or maybe you don't need a transfer agent, or you don't need a central clearinghouse.

The stablecoin yield point that's been talked through to death, that again kicks a bunch of work back to the rulemaking side. A big thing that we care a lot about is what comes out of the House, what they call BRCA. What this really does is protect developers from criminal charges if their software is used for nefarious reasons, right? We didn't go and say, “Okay, North Korean hackers are using Google, and so we've got to go—or maybe they're using whatever they're using, Airtable, or whatever your software is—and say, ‘Okay, well, this was used for a nefarious reason, and so we're going to put criminal sanctions against the founders and the company.’”

Whereas we've seen in the past—Tornado Cash being one of them, right?—where Roman Storm is being prosecuted for criminal charges based on a piece of open-source software which he launched out into the open and had no idea how it was going to be used, right? And then it ended up being used, in some sense, by North Korea, but not entirely, right? And so that one matters a lot to us because we want—and we hear from our developers or founders—that they are concerned about building a DeFi protocol that may end up being used by the wrong actor and what that means for them personally, right? And so that protection, I think, is really, really important.

There are a bunch of other nuances in there that matter, but for this audience, those are the things that I think are probably most important. Because at the end of the day, what does Clarity actually get you? It just gives you comfort and clarity that you can participate in this space in frameworks that will be specifically made to answer the questions that we're asking ourselves every day.

Yeah, to tie it all back into this circle of conversation, my one line on Clarity is it creates room for tokens to have utility, value creation, and value capture for really the first time in the industry's history.

Now, Rob, you're spot-on. I think rulemaking is where the real details are going to get decided, right? That's probably where the SEC is going to define what that actually looks like. But I think what we're going to have is, for the first time, I think Clarity is the beginning of a beautiful, multi-year bull market for crypto tokens. Santi, this is where I'm going to really disagree with, I think, the premise of your debate with Haseeb. I'm probably more on Haseeb's camp here because I think Clarity is the beginning of a beautiful, multi-year bull market for crypto tokens.

For a very select few who decide to do it, there are only a few who I think are doing it the right way right now. And I'm talking my book a little bit here because token transparency, investor relations, and token disclosures and transparency—I don't care if you use the Token Transparency Framework or something else here to do it, but there's a select few tokens that I think will do incredibly well. And I didn't buy Polymarket odds on Clarity, but I am starting to bid tokens for the first time in many, many, many years.

Yeah. No, look, I actually don't disagree. The premise of my argument—spoiler alert, Haseeb, if you're listening, I guess you'll have a few minutes to catch up, although I go first—is that tokens as we know them are dead. Tokens will end up resembling equities more, for the reasons that we've talked about at length: more transparency, more value-accrual mechanisms that allow token holders to participate, and then less value leakage.

The first point is the reason why public equities just have way more liquidity behind them. We talked about this in a prior episode about prediction markets: you need to have a certain set of ground rules that provide market fairness and integrity to incentivize people to actually bid these things. In the absence of that, it just becomes harder to really see much more liquidity entering the space.

But I don't think tokens are dead. Tokens as we know them are dead, but we're moving more in the direction of creating certain rules in place, like Clarity will set in place. And again, we touched on ARK. There is a time and a need to have tokens. I'm not in the blockchain-not-crypto camp. I actually strongly believe that tokens serve a really good purpose.

Rob, you said it: customer acquisition cost and go-to-market, providing validation and consensus and incentive alignment—all of that, like crypto-economic design. Tokens are essential substrates to spin up and maintain these systems. Without them, it becomes incredibly hard, right? If you really want to achieve decentralization and distribution, you want to—and need to—have a token.

But I do think—have you guys put out a correlation? It might be too early because there's a lot of noise in that, but it would be interesting over a couple-year time frame to say, “Look, the tokens that moved and had a higher...” What is the R-squared between token transparency and performance? I think there was a blog post by One Token Fund, which is more anecdotal than not, like, “Hey, look, we...” But I do certainly think it's true. You hear from VCs all the time: founders that respond really quickly to emails is kind of a thing that people latch on to. Teams that send out quarterly investor updates on a regular basis—these are all things that give you confidence as an investor that the team is high-integrity and operates at the highest level, kind of thing.

4. Anthropic Cracks Down On SPV’s

Anyway, we're not necessarily as serious an industry as we can be. The last interesting thing from the week is Anthropic. There are all these Anthropic SPVs that I'm sure you guys get pitched as well, and not just Anthropic—Rippling, OpenAI, Stripe. The SPV craze has hit an all-time high, and people want to get pre-IPO shares in a lot of these things.

Anthropic came out this week and said, “No, no, you cannot. We will not recognize any of these.” I know friends who have put probably 90% of their net worth into these SPVs in the last 2 years. The amounts involved in this craze are crazy. I think the craze that's happening in SPVs is not talked about or seen as much as some of the public-market stuff, but it is mania in SPVs right now.

Anthropic came out—I think OpenAI might have too—and said, “We will not recognize these.” I think you're going to see a lot of other folks start to do this. There are a lot of other SPVs that are very hot. Brex SPVs were hot, Lovable SPVs were hot, Stripe SPVs are very hot, and SpaceX SPVs are extremely hot.

The reason that's relevant for Empire and for this podcast is because we talked about this maybe a couple of months ago, but pretty much all of the tokenized-equity trading that you see today on Ondo, Hyperliquid, and a lot of these other platforms is just trading SPVs. It's not a dig at Ondo or Hyperliquid; I think it's just stating the reality. You're throwing a bunch of shares into an SPV and then throwing it onto a blockchain and trading it.

If the SPV mania blows up, with these companies not recognizing them, a lot of people are going to get hurt. As for the tokenized equities, I'm unsure how this will play out, but if you're trading tokenized equities, it's definitely something to be aware of. So, Rob or Santi, I don't know if you guys have takes or have thought about that.

I tweeted about that before this actually came out. I tweeted, a week or a week and a half ago, that the fraud in the SPV market hadn't really been digested by people yet. I got probably 10 pings from reporters after that, saying, “Oh yeah, we're talking to a bunch of people, and everyone agrees.” It's one of those things where we've seen a couple of things happen.

One is that there are a lot of smaller VC funds that have built their business on these SPVs because one of 2 things is true: they've struggled to actually win allocation in deals and be directly on the cap table for their own fund. Sourcing some shares of an SPV of an SPV—a second- or third-nested SPV—has been much easier to do. I think the TBPN guys said it, but we're almost at the phase of your cab driver shilling you his Anthropic SPV. It does feel that way a little bit.

I think it's important to note that there are a couple of ways these things happen. One is that the company is selling primary shares. In the OpenAI case, there was a Wall Street Journal article that came out about them actively allowing and facilitating secondary transfers specifically. Sometimes people just have an SPV that tracks this cap-table ownership because they don't have enough money themselves and want to source some extra capital for it.

Those are obviously fine. I think a number of nested SPVs into those are also obviously fine. The bad thing that people don't understand about these nested SPVs is that oftentimes they're getting 2 or 3 times the fees, because you're paying fees on an SPV and then you're paying fees on the ownership of the other SPV. But those are fine, right?

There's also a lot of employee transfers happening where they're not actually transferring the equity, but they're agreeing to a forward agreement and saying, “In the future, when it's liquid, I will give you the shares.” That's where this question becomes really tough.

I think Thomas Braziel, who runs a fund that does a lot of distressed stuff and did a lot of the FTX claims, has also been retweeting the other side of this. A lot of these companies really knew that these SPVs were happening. They may even have encouraged some of them, and they didn't stop them or come out earlier saying that they weren't okay.

Will Delaware take the point of view that they were complicit, and therefore they can't actually go back and disallow them? Can they now say that transfer restrictions were in place? Depending on what happens, there could be a significant number of lawsuits around this in Delaware. I expect that to be the case in the future. I don't know how this plays out.

There's also a world in which, if you're an Anthropic employee and it goes public this year, you get your shares 6 months after, and they're unlocked 6 months after the IPO. There's nothing stopping you from actually transferring them then. They can say this is true, but 6 months after the IPO, people can do whatever they want with their shares. I don't know how this plays out.

I've said this a bunch of times publicly, on this podcast, and on Twitter: on-chain spot exposure to equity that is an SPV of an SPV of a U.S. broker-dealer is not scalable and is very risky. It's illiquid, and buyer beware. Synthetic exposure is much better. That's why perps have been doing well.

We saw it on the Anthropic on-chain market—I think it was on Hyperliquid, or one of the other platforms. It was down to something like 400 billion after this came out, despite the fact that they're raising at 900 right now. Buyer beware.

I think the last point you made, Rob, is excellent. I would feel much more comfortable getting synthetic exposure to whatever the underlying weighted-average price in the secondary market is for some of these things. Of course, the implication is less liquidity if transfer restrictions are now enforced. Less liquidity means less price discovery.

Omar from SecondLane has really come out with this. The teams in crypto have had a lot of this for many, many years. Omar said it to me, and I think he's published it in his newsletter at SecondLane: teams that actually allow transfers and have a healthy pre-TGE market tend to do better. Solana did this, and many other teams have done this.

The difference between public and private markets is just liquidity. The Collison brothers, and probably Stripe as a company, have been the most adamant about saying, “We're never going to allow transfers. We just don't want people we don't know on our cap table.” I get that argument, but they have supermajority voting, and the structure is very much in their favor. They do a lot of tenders.

If you're not actively doing tenders, I actually think it's good for employees and whatnot to get liquidity if they want to. Elon, for instance, in SpaceX—you know, probably the most sought-after shares in the secondary market now are SpaceX, Stripe, Anthropic, OpenAI, and maybe some of the robotics names here and there.

Databricks is in the top 10 right now. Polymarket is actually there, and Kalshi is there. Those are the top 10. I don't know. I feel like maybe Anthropic has taken more of a hard line, and I don't necessarily agree with it. But if they don't have an active tender, then why allow it? Maybe just play ball to some extent. In the absence of that, I think that does raise the question.

What are the risks of synthetic exposure? I guess just a pricing fault and an oracle issue?

An oracle problem, right? That's a problem with a lot of these, call it TradeXYZ and HIP-3 markets, right? The same is true for anybody else who's doing this later; there are a bunch of others.

There is an oracle problem, and there's a question of how you manage risk at times when there isn't a publicly open market that you can actually reference. Everyone is trying to figure that out, because you can definitely provide some exposure, but how does that scale? That becomes really tough.

I think the private-equity market, even with synthetic exposure, has a real scaling problem and a real scaling question if the demand is really there—which it is. I think it brings up this other point: should we be doing things to encourage companies to go public sooner? That is probably almost definitely a yes.

But there’s real demand, to the point that, for every cab driver who’s in an Anthropic SPV now, if you’re an employee, wouldn’t you want to hedge your exposure if you have 90% of your net worth in it? I think SpaceX employees are about to make the Austin economy rebound because they have 12 people who are going to become billionaires, and over 100 people who are going to become centimillionaires.

If you want to hedge your massive concentration of wealth, you’d want to maybe get synthetic exposure, hedge your concentration of wealth, and lock in gains. They serve a real purpose, is what I’m trying to say. In the absence of that, not every company has a tender program like Stripe, where they repurchase shares.

There was a Wall Street Journal article over the weekend—and we’ve probably beaten this topic to death—that said OpenAI allowed employees to sell up to $30 million. Almost all the employees participated. The average per employee was $11 million at OpenAI. I think 75 took the entire $30 million off the table, and something like 600 and some took over $1 million.

People are hedging. We’ve seen all the tweets about the San Francisco housing market, the high end of it, anyway, and the Austin market and stuff. That’s going to happen, but there has to be a way for all of these employees who have all this newfound wealth to be able to hedge out their exposure. This is why all of these companies now have a head of IR and run these tender programs: They’ve decided they’re going to stay private for longer, but there has to be a way for employees to hedge out their exposure.

5. Content of The Week

All right, gents. It’s a beautiful day out there. Let’s go to Contact of the Week. Robin, I know you have to hop. Roberto, you’re first.

All right. I’m actually not going to do fiction this time. I just caught up to Fred Wilson. They put out a USV podcast channel, and they just did an hour-long episode about Fred’s career and USV’s growth. I found that super interesting, especially as what I consider a New York VC.

I think there’s this divide happening right now between New York and San Francisco. It’s always been there, but it’s getting even worse. There have been a lot of these debates, and I don’t know. I love Fred. He’s amazing, and I really enjoyed that.

Then, to your point around—or at least, I mentioned Brian Chesky earlier—his Invest Like the Best episode last week was really good as well.

Let me share a 30-second Fred Wilson story that’ll make you like him even more. Mike and I were hosting the second-ever Blockworks event. We had just quit our jobs. We didn’t realize that you had to sign a contract with a venue, so we just did a handshake deal with a venue, not realizing that we had to follow up with them. We thought the date was confirmed.

Three days before our second-ever event, we called the venue and said, “Hey, so you guys ready for us? We’re going to come in a day early and set some stuff up.” They said, “You don’t have the venue. You never signed the contract with us.” We panicked. We had about 300 people coming to this thing.

I used to read AVC, Fred Wilson’s blog, which he wrote for many years. It was a favorite.

Every day.

I still think he writes it every day. I don’t read it anymore, but I used to.

No, he doesn’t write that much anymore. He doesn’t write it anymore, but it’s an amazing blog.

It was just a paragraph every single morning. We were living on 18th and Broadway, and Union Square Ventures was a block away. I was like, “Let’s go see if USV is around.” I read this guy’s blog. I didn’t realize how big of a deal he was.

We went into their building and told the doorman we had a meeting with this guy, Fred Wilson. He let us up. We went into USV’s office. Mike and I were 23 years old at the time, and we said, “We have a meeting with Fred. We’re here to see Fred.”

The USV EA, the person at the door, said, “No, you don’t.” We said, “We’re meeting with Fred right now. We have a meeting with Fred. It’s confirmed.” Fred came out and said, “Guys, you don’t have a meeting with me. What’s the deal?”

We said, “I’m so sorry. We just need a venue. Can we use your office? I read your blog every morning.” He said, “No, you can’t.” But he connected us with Nick Grossman and Albert Wenger, another partner at USV. I have a lot of love for the USV people.

That’s a great story. That’s a real, real story.

Yeah, early days. We’re grinding. Grinding.

Two for me. One: Founders Podcast, the Dana White episode. Phenomenal podcast. So good. So good. It just gets you—it’s electric. I’m just going to leave it at that.

The second one is going back to the memory conversation. If you want to learn more about it, there’s a lot of content out there on Substack from really smart folks. There are 2 pieces. One of them is The Circuit. It’s actually 2 guys who are very deep in the semiconductor space, and they put out really good content.

There’s one with a Micron SVP that talks about the shortages they’re seeing and the trends. It’s just a good overview. The second one is Ben from Stratechery’s most recent blog post, called “The Inference Shift,” which is also pretty useful. For folks who want to learn more about that, it’s pretty good content.

I’ve got a political podcast, actually. Tucker Carlson went on a New York Times podcast called “The Interview,” which I thought was very interesting because The New York Times is pretty left-leaning. It’s the best 1-hour podcast—maybe it’s 2 hours—to understand the divide that’s happening with the right, which I don’t know if anyone on the podcast or if Empire cares about.

We never talk politics, so you guys can skip this recommendation this week, but I did find it pretty interesting to hear. Tucker Carlson was very pro-Trump, rallied and campaigned for him, and he’s very anti-Trump now. He talks about the impact of the war in the Middle East and what it’s done to divide the right.

It’s relevant because it has implications for what happens in the 2026 midterms. The midterms are this year. There are implications for crypto. If the House flips, you better believe that the House is going to be investigating a lot of crypto stuff.

My fingers are crossed that the only thing they’ll focus on is World Liberty Financial investigations, which have already started. It’s probably content for another podcast, but the House is getting fed a lot of information right now on World Liberty. I would guess that they launch a big investigation if the Democrats win.

Anyway, go listen to it. Santi, Rob, good chat. Thank you, folks. Let’s go enjoy Tokyo. Sayonara. Have a good weekend, guys. Love you guys.