PRIM3 Brief #20: Treasury Companies Were a Distribution Channel, Not an Asset Class
In the second week of August, four of the better-connected names in Asian crypto handed back $110 million they had already raised.
Huobi founder Leon Li, HashKey chairman Xiao Feng, Meitu co-founder Cai Wensheng and Fenbushi's Bo Shen had spent months assembling a $500 million ether digital asset treasury company. Commitments were signed. Then they shelved the whole thing, pointing at market conditions.
Most of the coverage filed that under price. Crypto sold off, the vehicle stopped penciling, capital went home. True, and not especially useful if you're a founder trying to plan next year.
Here's the read that matters. Digital asset treasury companies were never primarily an asset class. For token projects they were a distribution channel — a way to move a large block of supply off a foundation's balance sheet into a listed equity wrapper without ever printing a trade on an order book. That channel is closing now. And it closes unevenly: bitcoin vehicles get to grind on as levered holders, while the single-token DATs built around specific protocol treasuries lose the only thing that made them work.
Key takeaways
- One in three of the 156 listed crypto treasury firms tracked by Capriole Investments trade below an mNAV of 1 as of August 2026, with the sector average falling since May.
- What founders actually bought from a DAT wasn't the premium. It was supply placement without market impact.
- Below 1.0x, that trade inverts. A foundation that swapped tokens for shares now holds illiquid equity worth less than the supply it handed over.
What a Treasury Company Actually Bought
The public version of the DAT trade is simple enough. A listed shell raises money, buys a crypto asset, and trades at a premium to the value of what it holds because equity investors will pay for access, leverage and index inclusion they can't get directly. The company then issues stock into that premium, buys more coins, and coins-per-share goes up. Forward Industries ran the largest version of it on Solana, with a $1.65 billion PIPE in September 2025 followed by a $4 billion ATM programme.
The private version — the one that showed up in our deal conversations through 2025 and the first half of 2026 — was different. Protocol foundations weren't looking at DATs as investors. They were looking at them as a place to put supply.
StablecoinX makes the mechanism legible because it's in the filings. The company reported an ENA treasury of roughly 3.0 billion tokens at the end of Q2 2026. Of that, 284,954,407 tokens came directly from the Ethena Foundation as a contribution. The rest, about 2.75 billion, came from PIPE investors in cash and in kind, carried at $218.4 million as of June 30, 2026.
Read that again as a founder rather than as a trader. A foundation converted a nine-figure slug of its own token into a position in a Nasdaq-listed vehicle. No exchange print, no depth problem, no unlock headline. For a team sitting on 40% of supply with a community that punishes every treasury transfer, that's not a financing. It's a pressure valve.
Why "Forced Selling" Is the Wrong Thing to Watch
The consensus worry right now is a liquidation cascade. Treasury companies fall below NAV, debt comes due, they dump coins, price falls further, more of them fall below NAV. The spiral framing has been everywhere since DL News reported the one-in-three number, and the individual data points are real enough. Nakamoto Inc. sold 600 BTC for around $48 million to retire $45 million of lender debt, then authorised a $25 million buyback of its own stock. Satsuma Technology is heading into a proxy vote, pushed by Pantera Capital, on liquidating its entire 668.48 BTC reserve.
But the aggregate is smaller than the noise suggests, and the market already knows the unlock calendars — which is exactly the problem we described in Brief #16 on unlock cliff design. Positioning ahead of a scheduled, public sale is a crowded trade before it starts.
The structural loss is quieter and lasts longer. Capriole's Charles Edwards has laid out where a sub-1.0x company can go: sell the underlying, get acquired, or add leverage to manufacture a yield story. None of those routes involve buying more of a partner protocol's token. So the marginal non-VC, non-retail buyer of large token blocks — a buyer that absorbed supply at negotiated prices with no slippage — has stopped bidding. Nothing blew up. The bid simply went away, which is the harder kind of loss to notice.
And the exit is worse than the entry was good. A foundation that contributed tokens in exchange for restricted shares in a vehicle now trading at 0.7x NAV is holding a discount on a discount, with a lockup on top. Forbes made the accounting version of this argument on August 21, 2026, pointing out that the metrics the sector reports on itself don't capture what these positions are worth. Foundations that took paper are finding that out in their own quarterlies.
What the Data Actually Says About Where the Damage Lands
Split the 156 companies by what they hold and the picture stops being uniform.
Bitcoin vehicles have a defensible reason to exist below par. They're a levered claim on a liquid asset with a deep options market and, for some allocators, a wrapper that solves a mandate problem. Painful, dilutive, survivable. Altcoin vehicles are a different animal. Their premium was always a function of the sponsoring protocol's narrative, their float is thinner, and their underlying often can't be sold in size without moving the market it's marked against. When those trade to a discount, there's no natural buyer of the equity and no clean liquidation of the assets.
That asymmetry is why the ether vehicle got shelved and the bitcoin ones are restructuring instead. It's also why we've told portfolio teams to stop treating "a DAT is interested in us" as a validating signal. In several cases through the first half of 2026, the DAT's interest was the clearest evidence that the token had a liquidity problem the team hadn't priced.
Work through the arithmetic on one of these deals and the asymmetry gets uncomfortable fast. Suppose a foundation contributes tokens marked at $60 million into a listed vehicle and receives restricted equity at a valuation struck against net asset value on the day of signing, with a twelve-month lockup and a registration process that realistically runs longer than that. The vehicle then drifts to 0.75x, which is not a distressed print in the current tape but roughly the median for the discounted cohort. The foundation is now carrying something in the mid-forties on a position it could have sold in tranches on-venue for more, and it can't move it, and the underlying token has fallen too because the whole sector re-rated at once. Every one of those effects is correlated with the others, which is precisely the property a treasury team was trying to avoid when it picked the negotiated route over the order book in the first place.
Anyone can check the receipts here without waiting for a filing. Bubblemaps, which we backed and which turned on-chain investigation into something you can look at rather than reconstruct, makes contributed-supply clusters visible. If a foundation moved a block into a corporate vehicle, the wallet graph shows it before the 10-Q does.
What This Means for Founders
If your 2027 plan has a treasury-company placement in it, rewrite that line this quarter. Not because the vehicles are all going away, but because the terms have inverted and most teams are still modelling the 2025 version.
Price the equity, not the headline. A $50 million placement into a listed vehicle at 0.75x NAV isn't $50 million. It's restricted stock in a company whose own shareholders are voting on whether to liquidate. Mark it the way an LP would mark it.
Assume the placement is public and permanent. Every transfer of that size is legible on-chain within hours, and the community reads it as a sale regardless of the structure. Teams that treated a DAT deal as a quiet alternative to a market sale generally discovered it wasn't quiet.
Then go back to the boring instruments. Over-the-counter blocks with real vesting, revenue-linked structures, and equity rounds that leave the token alone entirely. We wrote in Brief #19 about how much of the headline funding number is now non-cash token contribution rather than money into an operating business — a lot of that flow ran through exactly these vehicles. Strip it out and the actual cash available to early-stage teams looks the way it did in Brief #12's barbell: fewer rounds, larger checks, less middle.
Where PRIM3 Is Placing Bets
We're underwriting the plumbing that the last two years of treasury-company activity exposed as missing.
Custody, reporting and valuation infrastructure for institutions holding token positions they can't mark cleanly — that's the thesis behind our position in Crymbo. On-chain forensics and supply visibility, where Bubblemaps has become the default reference for allocators trying to see who actually holds a float. And compliant issuance rails through Kor Protocol, because the projects that will place supply with institutions in 2027 will do it inside a securities wrapper rather than through a shell company's balance sheet.
We're also spending more time than usual on token designs that never need a placement channel in the first place. If the treasury allocation is small enough and vests long enough that no single transfer moves the market, the founder never has to negotiate with a vehicle trading at a discount to its own coins.
That's the structural version of the same lesson the unlock cliff taught: supply problems are design problems, and they show up years after the design is locked.
The DAT window was open for about eighteen months. Teams that used it to solve a liquidity problem bought themselves time. For everyone who reached for it instead of fixing tokenomics, the result is illiquid equity in somebody else's unwind.
That's where PRIM3 is placing capital this quarter. More on our thesis at prim3.vc.