Conversation with Patagon Founder: Revealing the Inside Scoop on Anthropic's Secondary Market

Organized & Compiled by: Deep Tide TechFlow

Guest: Dio Casares, Founder of Patagon

Podcast Source: Bankless

Broadcast Date: May 14, 2026

Editorial Introduction

In this podcast episode, Patagon founder Dio Casares unveils the inside story surrounding secondary-market trading for star companies such as Anthropic. Patagon is a company focused on digital-asset investing and private secondary-market matchmaking. Dio Casares says that even secondary transactions related only to Anthropic—here “secondary” means a primary-like secondary, i.e., shareholders or employees privately transferring equity to others; the detailed explanation is mentioned later, and all Anthropic secondary transactions in the article are meant in this sense—amount to hundreds of billions of dollars. The fee rate can be as high as 10% per deal. About 10%-20% of executed orders involve fraud or forged equity. Even the money fund professionals make from these kinds of transactions already exceeds what they earn from their core investing business.

Even more alarming are nested SPV (Special Purpose Vehicle) structures, “forward contract”-type employee equity, and “tokenized” private equity. Once Anthropic goes public, delays in the distribution of multi-layer SPVs within the DTCC system, whether GPs at each layer choose to hold, and the possibility that some equity may be invalidated at the company level will trigger a wave of lawsuits that can last for years.

Key Quotes

Market Structure and Arbitrage Space

“You can’t just walk up to Anthropic and say, ‘I want to buy $1 million worth of stock in this round.’ It’s a market built on insider connections.”

“If someone has shares, they sell shares. If someone has buyer resources, they sell buyer resources. A small number of people do both sides—that’s the structure of this market.”

“Even people inside funds make more from this kind of secondary trading than from doing their main investment business, so a large number of people are moving into this market.”

Market Scale and Fee Rates

“In the private placement market, the financing amounts in the past few years have already surpassed IPO fundraising. With recorded secondary-market deals plus financing rounds, the scale is over $200 billion.”

“Many Anthropic deals we see involve an upfront 10% fee plus long-term commissions. If $10 billion comes into the channel like this in one round, the fee pool alone is $1 billion.”

Company-Approved vs. Unapproved Secondaries

“Anthropic generally supports direct trading: company approval, being added to the shareholder register, and then joint distribution by the connecting funds.”

“What the company hates most is platforms like Hive and Forge. They see a large block of shares, then send mass emails to hundreds of thousands of people on the platform who haven’t done KYC, saying, ‘I have discounted shares here.’ This directly interferes with Anthropic’s fundraising for this round.”

“OpenAI and Anthropic recently made employee tender offers, allowing employees to directly sell up to $30 million under this round’s valuation. This is essentially the company ‘cutting in’ on those sellers who otherwise would have gone through gray secondaries.”

Fraud and Bad Debt

“About 10%-20% of the deals we’ve seen are fraud. Equity certificates can be forged—that’s direct scam.”

“More common than outright scams is when someone claims they have shares but actually don’t. They collect the money first, then circle back to look for the stock, and often they can’t find it.”

“In the U.S. legal system, you’re ‘presumed innocent,’ so the problem is this: if a position goes from $1 million to $50 million, and it costs $10 million to sue and get it back, the seller might simply default. After all, they can still net a $40 million profit.”

Nested SPVs and the Settlement Hell After IPO

“Why are there two layers, three layers of SPVs? Because buyers and sellers often can’t ‘match up’ just right. A $8 million seller might be matched with three buyers pooled together.”

“Anthropic calls out Sidecar explicitly. They think Sidecar’s due diligence isn’t sufficient. Basically, once they look at a document and think ‘looks fine,’ they approve.”

“The real chaos after the IPO is this: first-layer SPVs take days to two weeks to receive the stock. Then they ask the LPs whether they want cash or stock, and pass it to the second and third layers… at any layer, if the GP wants to continue holding and lock the position without distributing, the downstream is completely blocked.”

“After the IPO, the company basically won’t go back to reclaim any problematic shares. They won’t do private rounds again either, and the incentive to play games that maintain market order disappears.”

Advice for Small Buyers

“If you’re a small buyer, putting $100,000 to $1 million into some ‘tokenized version of Anthropic’ or a similar vehicle—most of the time you can’t even lift the lid to see the underlying layer. At most, you can see the vehicle where the money goes, and that is typically the second or third layer.”

“I trust my intuition. If you feel very bad about this position, you should get out.”

The Real Operating Mechanism of Anthropic Secondary-Market Trading

Host: Regarding Anthropic’s secondary market, and even the private market more broadly, many people have a lot of questions. Before we begin, could you introduce yourself and explain why you have a unique perspective on Anthropic’s secondary market?

Dio Casares: Patagon has two core business lines: proprietary investing, and client-facing services. We’ve invested in secondary trades ourselves, and we also treat secondary trading as a product for clients—helping people find channels to access share allocations.

Host: So, as a client service, you go into the market to find popular secondary share allocations, package them, and sell them to your clients.

Dio Casares: Exactly.

Host: That puts you in the front row for observing this market. Right now, the hottest pool of capital is the secondary market—especially Anthropic, as well as SpaceX and OpenAI. Could you tell the audience what’s happening here? Most people have no concept of this at all.

Dio Casares: Broadly speaking, secondary deals can be divided into two types. The first type is primary-like secondaries. The name itself is a bit contradictory. It means that instead of letting a fund directly put money in, there are people in the market who set up an SPV (a special-purpose vehicle), then set up another SPV on top of the SPV, and inject the money. This is essentially new money for the company—yes, the company is indeed getting the financing.

Employee stock sales fall into this category as well, because they are approved by the company. The value is obtained by the company from issuing stock to employees, and then employees are allowed to cash out the stock.

The second type is genuine secondaries in the truest sense. You buy the allocation from someone who has already bought the shares from the company. Historically, this has been relatively troublesome. In traditional thinking, VC exits wait for an IPO or an M&A deal, but now funding rounds can be hundreds of billions, far beyond the $10 billion scale of companies’ IPOs in the past. The liquidity timeline has changed completely. When FTX went bankrupt, a large block of Anthropic shares was forced to be sold—pushed by the bankruptcy process itself.

So the secondary market needs to be built. But at the same time, it is viewed with suspicion by management, because they believe it may compete with the stock sales they use for fundraising.

Host: So besides Anthropic’s own appeal, this phenomenon is driven by two structural inputs: first, the market size itself is already enormous, with even more capital flowing in; second, these companies stay private for longer, giving the secondary market time to mature and for more participants to enter.

Dio Casares: Yes, I agree.

Host: Can we first talk about the normal situation? Anthropic knows that the secondary market exists—some of it is company-approved. How is a company-approved secondary deal completed?

Dio Casares: Perhaps the more accurate phrasing is the SPV market. There are people in the market who want to buy into Anthropic. They’re not in funds, and they don’t have special loyalty to the company. They’re purely there to make money. Anthropic in general supports direct trading: company approval, inclusion in the shareholder handbook, and then joint distribution by the connecting funds. Those funds make money by helping the company raise capital.

Anthropic is doing this with several large PE (private equity) firms right now. These institutions are discreet, but they’re connecting with many people and allocations. They’re not in Anthropic’s publicly announced list of “unauthorized institutions,” so we can basically treat them as company-approved.

Another type management strongly dislikes. These companies often send legal letters to them directly—for example, platforms like Hive and Forge. Their play is: once they see a large block allocation, they mass-email tens of thousands of people on the platform who haven’t done KYC, telling them, “I have discounted shares here.” This directly disrupts Anthropic’s fundraising for this round. What they’re doing is a “leak-and-make-a-profit” business model: trying to find shares cheaper than the current secondary market price or this round’s valuation.

As a result, on Anthropic’s side, family offices and major clients will go around saying, “Over at Hive and Forge, they say I can get a 20% discount—so why should I invest directly in this round?” This makes fundraising harder for Anthropic. Worse still, psychologically, once the market shows a clear spread—“selling prices low, buying prices high”—it usually means the market isn’t active, which is a bad signal. The company wants to eliminate that signal.

Recently, both OpenAI and Anthropic have made employee tender offers, allowing employees to sell up to $30 million at this round’s valuation. This is essentially the company “cutting in” on sellers who would otherwise go through gray secondaries. Many people who wanted to sell already sold enough in the tender offer, so they won’t go sign those private agreements promising “I’ll buy your stock next year.”

Host: So company-approved deals fall into two categories: first, non-competitive ones—where funds raise money for the company, and the money goes into the company; second, improving the future market structure—so that employees or others in the ecosystem who want to sell can sell before the IPO, releasing selling pressure. These are healthy, positive-sum transactions that align with Anthropic’s interests. The bad category is a bunch of intermediaries “siphoning off value,” which makes the company worse off and makes things look bad.

Dio Casares: Exactly. In the U.S., there’s a rule for unpublicly listed securities: a six-month holding period. So some “tokenized private equity,” theoretically, could violate this law if someone keeps buying and selling back and forth, repeatedly. Maybe they do some kind of workaround in the background, but historically U.S. regulators tend to claim jurisdiction as long as the assets are linked to the U.S. Another thing Anthropic doesn’t want is to be accused by regulators of “knowing and not acting.”

Host: So legally, Anthropic has no room to “pretend not to see.” Once they know about these markets, they have to take action.

Dio Casares: Right.

Host: How big is this market, exactly? Just Anthropic-related activity is in the hundreds of billions. Within that, what proportion is unhealthy dark-market activity, and what proportion is the whole market?

Dio Casares: Basically, it’s the entire private equity side. Private equity comes in many forms. Some family offices co-invest. Some work with brokers and with institutions like ours to raise funds and charge fees—completely different from each other. On top of that, brokers are layered: first-layer brokers know a bunch of buyers, and also connect with another broker—where that broker actually holds the allocation. So the market structure is complicated, and the amount of money involved is huge.

One interesting data point: private fundraising has exceeded IPO fundraising for years now. With recorded secondary-market deals plus financing rounds, the scale is over $200 billion. And considering that fees aren’t just a few basis points, but rather the kind of structure we see in Anthropic deals—an upfront 10% plus long-term commissions—if $10 billion comes through this channel in one round, the fee pool alone is $1 billion.

Host: I recently saw two pieces of content on social media describing how insane the market is. One was on the Hing profile of a guy in San Francisco, who wrote, “I know people at Anthropic, dating without commission,” using Anthropic stock allocations to lure dating prospects. The other was a woman tweeting that, “I brokered a single Anthropic secondary deal, and the money I made is more than all the money I earned working part-time in my 20s—this is ridiculous.” This is what the San Francisco social elite looks like when they’re competing on social platforms around Anthropic allocations. How did this happen?

Dio Casares: I actually talked with the person who posted that tweet. From a buyer’s perspective, you want to buy Anthropic, but the company bylaws and agreements aren’t public, and it’s hard to get access. You can’t just run up to Anthropic and say, “I want to buy $1 million worth of stock in this round, thank you.” That’s because it’s a market built on internal connections: if someone has shares, they sell shares; if someone has buyer resources, they sell buyer resources; a few people do both. That’s the market structure.

People inside funds doing these kinds of secondary trades make more, so many people are moving into this market.

Host: So everyone sees Anthropic equity like a gold mine, and then there’s a whole lot of people selling picks and shovels.

Dio Casares: Yes—and competition is fierce now, which is a good thing. A few months ago, there wasn’t real competition. Most people were just flipping through intermediaries rather than directly connecting with sellers. Now more and more people can find both buyers and sellers at the same time, and handle the entire process more professionally. But at the same time, the fee rates they can charge have come down.

There’s another risk that many people don’t realize: in some cases you can’t obtain the shares held by investors, so you can only buy employees’ forward contracts instead. This recently blew up. A well-known institution sold an xAI employee forward contract. Later, that employee was named in a lawsuit against OpenAI related to corporate espionage, and their entire equity was reclaimed by the company. The result was: the money was paid, the fees were collected, but it was a complete mess. All the buyer brokers were left hanging. The institution’s stance was, “If you paid the fee, that’s your problem, not ours. We can only return your original principal.” I think these “fake SPVs” will increase. In the future, this will evolve into a game of reputation—who can make an investment vehicle that doesn’t blow up.

10%-20% of trades involve fake stock certificates

Host: Let’s talk about why an investment vehicle might blow up. I understand that it’s a nested SPV setup—a second layer, a third layer, a fourth layer—each layer extracts fees. The further down you go, the more uncertain it becomes whether the equity truly exists.

Dio Casares: Second- and third-layer SPVs exist because “buying and selling intent doesn’t line up.” A seller with $8 million usually doesn’t face a buyer with exactly $8 million. It might be that three buyers pool together. Most people in this space aren’t licensed brokers, so they can’t charge fees by passing shares through the middle. But if you set up a fund, you can charge upfront management fees for running that management structure. The fees are collected at the SPV level.

Host: Does Anthropic like these funds, or are they explicitly against them?

Dio Casares: Better than having none. Because at least you can do tax reporting; if managed properly. Anthropic also publicly says which fund administrators they approve for administrative services. They specifically named Sidecar, which is interesting. Everyone else is a broker for funds or SPVs, but Sidecar is simply a fund administrator. They named Sidecar because Anthropic thinks Sidecar’s due diligence is insufficient. Basically, once they see a document and think “looks fine,” they greenlight it.

Returning to your risk question: first, the equity may not actually be real. Stock certificates can be forged—that’s direct fraud. We’ve seen at least 10 orders like this. After checking the share transfer records, you can confirm they’re fake, but there’s not much you can do besides whistleblowing. Sometimes you also can’t tell whether the person forged it themselves or whether they’re reselling fake goods. There are definitely plenty of scams in the market, but I don’t think they’re as common as people claim from outside. Roughly 10%-20% of executed orders involve fraud. What’s more common isn’t scams—it’s people claiming they have allocations when they don’t. They collect money first, then try to put it into the company afterward, and it often fails.

Host: Is there “unintentional fraud”—people try their best, but because the market is like this, they actually end up not delivering the assets they promised? Is there a gray area?

Dio Casares: That’s called “gross negligence.” The gray area isn’t that large. Resources like Pitchbook, shareholder handbooks, and other due-diligence materials are supposed to be used when dealing directly with sellers. If you don’t do due diligence on your buyers or clients, that’s negligence. If you buy from a reputable seller with a shareholder handbook and they still do something shady, then that’s a different situation. But the market values reputation—people who are unreliable have their reputations in the circle.

Litigation and Locked-Share Disputes After IPO

Host: After Anthropic’s IPO, how does this pile of speculative markets “collapse”? Not a bad collapse, but in terms of settlement, allocation distribution, and cash changing hands.

Dio Casares: It mainly comes down to two things: first, broker accounts and AML (anti-money-laundering) procedures at the DTCC system level; second, the distribution terms of each fund. Some funds have complete discretion over timing; others state that once the IPO happens and shares are tradable, distribution must occur immediately in physical or cash form.

Imagine a three-layer SPV: the first layer receives the shares, then asks the underlying LPs whether they want physical shares or cash. If the LPs want shares, it gets passed up. This depends on DTCC. Normally it takes a few days, but if banks are slow, it can take two weeks—that’s the two-week delay. Then the second layer again asks its LPs whether they want cash or shares, and passes it to the third layer, which is another 3 days to two weeks.

At any layer in the middle, if the distribution rules allow the GP to decide freely—for example, if after Anthropic opens, the stock price skyrockets. The first-layer GP might say, “I have long-term commissions. I want it to go up a bit longer.” Or vice versa, if the stock crashes after opening, they might not want to deliver immediately and instead want to hold for a few more months. Once this happens, everyone downstream can’t get their shares. There may also be someone who hedges their long positions in the public market. Technically that’s a gray area. You might have thought delivery would take six months, but then it drags on another month—this is where a lot of lawsuits emerge.

Host: It sounds like Anthropic itself probably isn’t too concerned. For them, once the shares are sent out, it’s done—the top-layer SPVs handle it.

Dio Casares: Right. After going public, the company no longer needs private transfer agents. They only used transfer agents at the initial issuance. After that, everything runs through the DTCC system, and they basically don’t get involved. But many brokerages and banks might watch these trades and think, “Anthropic says this transaction is invalid. We need to find out whether we can help you sell.” That can be very troublesome.

But from a game-theory perspective, after the IPO, the company generally won’t go back to pursue those problematic allocations. They won’t do private rounds anymore, and the incentive to play games that maintain market order is gone.

Host: How big can this get? How many lawsuits? How many dollars get pulled into it? How long will it take to clean it up?

Dio Casares: The lawsuits will drag for years. Some cases will definitely take years. I can’t say the exact total—nobody can. But it will be this market’s “wake-up moment.”

A few days ago, I spoke with a small European family office person. It really upset me. I believe they invested in the problematic deal I mentioned earlier, and ultimately the money was returned. But I believe that GP didn’t tell the LPs that they kept the returned money and wanted to keep operating—trying to bet on Anthropic’s appreciation. This is pretty common: taking returned money as trading capital to chase gains. Unless they can generate a 500% return, they can’t plug the hole. I’m not optimistic they can pull it off. That accounting loss will be borne by the fund itself.

Host: What you’re worried about is that some people subjectively want to do well, but they mess up—like buying fake equity. But why does the client’s money still remain after things go wrong?

Dio Casares: Yes, or because of negligence. My intuition is that the fee structure on these large blocks is heavy: the GP takes the money, leaving little to return to the LPs. Or for some reason, they believe they can’t return it. But finance doesn’t work like that. Once something goes wrong, someone has to step up and say, “I’m very sorry this didn’t work out. Here’s your money back.”

Host: So the error path looks like this: you raise money from friends and family, set up an SPV, and you have money. Then you receive a verbal promise from another person that shares will be delivered. At that point, you have two options: do nothing, keep the money in the SPV and wait for the shares to come; or count your eggs early—“I just made a big profit; I’m buying a house, buying a Porsche”—and then on the delivery day you find out you didn’t get the shares, but the money is already spent, and there’s nothing to return.

Dio Casares: That’s exactly right.

Host: Let’s zoom out more. The private market is huge. Companies IPO later, and capital changes hands privately, gradually turning into an internal market of its own. That’s the opposite of public markets—but strangely, now the coolest companies are staying in this market longer. How will this market evolve in the future?

Dio Casares: Saying it’s “completely unregulated” isn’t fair. Regulation exists, but it’s definitely “wild,” and it isn’t strictly enforced. Unless there’s obvious fraud, regulators generally don’t really manage it, and they can’t keep up. For example, should the U.S. financial regulators go arrest someone for not filing the correct paperwork, or go after illegal financing? Most people would say, of course, go after illegal financing. Sometimes it’s the same batch of people doing both.

The market keeps repeating similar patterns. This is similar to the period in crypto with low liquidity and high FDV. When supply is restricted, it creates crazy price action, which makes fundraising easier. Behind this cycle, there is real technology—like I use Claude. Their revenue is already substantial.

What’s interesting is that the big existing institutions—banks that either have their own secondary departments or partner with others—are cautious and can’t keep up with the market’s pace. So you see a wave of new companies coming in to fill that gap. Meanwhile, large funds also set up SPVs, but structurally different, mainly serving their own LPs. The trend is shifting from “investing through a fund with pooled management” to “directly managed capital.” I think this will last for a while, until this cycle ends. There will be a batch of people who buy vehicles equivalent to “locked-token” structures, lose a lot of money, and finally say, “Fine, I’ll put the money back into VC funds.” This hot money will go elsewhere, but the U.S. secondary market will become more professional.

Patagon’s Strategy and Philosophy

Host: Coming back to what you do at Patagon. Based on your experience and understanding of the secondary market, please introduce Patagon’s strategy and philosophy.

Dio Casares: We initially only did proprietary trading—entering deals ourselves. Later, one time a friend paid me a fee. I asked him why. He told me another broker was going to charge him two to three times as much. What he paid me amounted to the portion he saved. That made me realize: I grew up in the Bay Area, know a lot of people, and I know who to call and how to verify backgrounds. Many of my friends have international backgrounds, and they don’t have that much local social network in San Francisco. I started doing it as a side job, and then gradually I found out it could become a business—especially with branding and process.

Look at platforms like Forge and Hive. They don’t verify whether equity is real. They don’t audit buyers. They don’t collect KYC information (here, I’m only talking about their marketplace business; their own directly handled investment opportunities are another matter). Yet they still charge 3.5%. To give you only an introduction, a fake order book, and you still have to negotiate prices by email—then they still charge 3.5% on the transaction. We think that’s outrageous.

What we do is: we find deals ourselves, set up investment vehicles ourselves, and do due diligence ourselves. We ensure that the equity is real and that the structure is compliant. Clients can directly invest in these deals on the platform—without negotiating prices first, requesting investment-vehicle documents, signing, and then going back and forth by email for bank transfers. Everything is completed in one place. In the end, we can even help clients use their positions for credit financing. The value we want to provide to clients goes far beyond “get you into this deal and then leave you.”

We’ve handled some complex deals. For instance, a crypto company where everything was employee forward contracts. In the due-diligence process, we did background checks for each employee: whether they had gambling problems, whether they had acquaintances with negative references. For any employee we found to be problematic, we didn’t work with them; for the others who were fine, the entire transaction went through smoothly.

Host: This also helps you build credibility. When you’re trying to get Anthropic secondaries or allocations of other companies, you can say, “Our client base has been quality-filtered.”

Dio Casares: Exactly. We can also tell clients, “We’ve handled hard deals.” At the time of that transaction, there were no other channels that could access authorized allocations. We got clients into deals that others couldn’t access. Clients appreciate that, and next time, they naturally come back to you.

Tokenized Equity and Legal Risks of Perpetual Contracts Before IPO

Host: If there are listeners who have already bought Anthropic secondaries or other companies’ secondaries, but don’t know whether the underlying story is real or fake, what advice or actions do you have?

Dio Casares: It’s hard to generalize, because the differences in market structure are huge. Some people currently hold perpetual contracts. Personally, I don’t recommend them. But ironically, perps are derivatives, which fall into a completely different legal category, and the risks aren’t as obvious. Funding rates might be very aggressive, but that’s the price you pay to keep them aligned with the IPO opening price.

If you’re a small buyer, putting $100,000 to $1 million into some “tokenized version of Anthropic” or other similar vehicle—most of the time you can’t even lift the lid to see the underlying layer completely. At most, you can see the vehicle where the money goes, and that’s usually the second or third layer. I would recommend not adding to your position. If your gut feeling about this position is very bad, then in general I trust your intuition—get out.

Host: The tokenized perpetual contracts you mentioned—do they have real claims to the underlying equity, or are they just projections, like predictions or subjective mappings?

Dio Casares: A number of institutions are doing this now. Although the mechanisms differ across parties, the idea is that once these products go live, the funding rates on pre-IPO perpetual contracts will become extremely crazy—because pre-IPO perpetuals aren’t quite the same as ordinary perpetuals. Market makers already have underlying trades as hedges, and the hedging approach is different from the structure in the U.S. stock market. But ultimately, it will converge to a real stock. That means arbitrage becomes possible. So as the IPO approaches, the perpetual price and funding rate will move toward “normal market” levels.

Host: Are there any other topics you think I didn’t ask about?

Dio Casares: I think we’ve covered things pretty comprehensively.

LAYOUT REFERENCE (source): total_lines=211, non_empty_lines=106, blank_lines=105