OpenAI was last marked at $157 billion. SpaceX at $350 billion. Dragoneer is taking its holdings in both companies and placing them into a new fund targeting $2 billion. The fund is being marketed with a ‘minimal discount’ — not a standard control premium, not a fire-sale haircut, but a price that signals scarcity.
That is the anomaly. When a seller holds two of the most sought-after private assets in the world and still asks buyers to accept minimal downside protection, the market should stop and ask who is really selling what.
This is not an ICO. It is not an airdrop. It is a continuation fund — a private-market liquidity event that uses no blockchain, issues no token, and creates no public order book. Yet it is the most relevant piece of crypto-adjacent financial engineering you will read this quarter, because it reveals how the real world is solving the same problem crypto has failed to solve: why does a shareholder in a high-growth asset need to exit at zero liquidity?
The Context
Dragoneer is a growth-oriented investment firm with a Bay Area address and an SEC registration. Its portfolio has included Stripe, ByteDance, Robinhood, and NVIDIA. It is not a blockchain company. It does not run a validator. It does not need one. But its latest transaction matters to this industry because it is a liquidity event with no public ledger.
A continuation fund is a private equity mechanic. The general partner takes assets out of an older, soon-to-expire fund and transfers them into a newly created vehicle. Old LPs are offered cash or a rollover. New LPs get access to assets that would otherwise be locked away for another five to eight years. The GP gets something too: a new fee clock and a new carry period.
I learned to look for these clocks the hard way. In late 2017, I spent 40 hours auditing an ICO distribution contract. The token arithmetic was correct. The vulnerability was in the permission model: a subtle integer overflow in a bonus allocation function allowed a wallet to drain funds after the fourteenth reward cycle. The fix was a one-line boundary check. The lesson stuck. If a contract or a fund only works when everyone behaves, it does not work.
Continuation funds work only when the GP is honest about the fair value of the assets being moved. That makes the ‘minimal discount’ the most important variable in the entire design.
The Core
Let me be direct about what a continuation fund does to fees. Dragoneer’s older LPs may have already earned a large multiple on OpenAI and SpaceX. If those assets are transferred into a new fund at near-par value, the GP resets the carry clock on billions of dollars of embedded gains. The old LPs get liquidity; the new LPs get access; the GP gets a fresh 20% carried interest schedule on assets that were already accumulated.
This is not fraud. It is structure. But structure is a code, and code has bugs.
The first bug is valuation. OpenAI and SpaceX are Level 3 assets. There is no public tape. No continuous auction. A fair value is determined by the GP, its advisers, and whatever window into private secondary markets it cares to pull from. Forge Global and EquityZen data are useful, but they are not exchanges. They are negotiated, thin, and prone to sampling errors. In this context, the phrase ‘minimal discount’ is not a price target. It is a statement about negotiation leverage.
The second bug is concentration. A $2 billion fund seeded primarily by OpenAI and SpaceX shares will naturally hold 40 to 60 percent of its NAV in two assets. If either company suffers a 30 percent markdown, the fund drops 12 to 18 percent in a single valuation cycle. That kind of volatility is acceptable in public equities because you can sell. In a continuation fund, the LPs have signed a five-to-eight-year lockup. Volatility is not risk; impermanent loss is. In private markets, the equivalent of impermanent loss is being locked into a falling Level 3 price with no buyer.
The third bug is the discount itself. The press release is not going to say “we are selling our winners at a premium to a naive buyer pool.” It says “minimal discount.” That is a marketing phrase designed to appeal to FOMO and fear of missing out on OpenAI. A rational institutional LP should run a simple sensitivity table. If the true fair value is 10 percent below the transfer price, the new fund starts with a negative cushion. If the discount is truly minimal, the fund has no margin of safety.
I have reviewed enough audited financial statements to know that audited does not mean true. Ledgers do not lie, only the auditors do. In a continuation fund, the auditor is paid by the GP. The valuation committee is appointed by the GP. The fair market value is defended by the GP. The new LPs are the only party in the room without a seat at the pricing table.
Now add the regulatory layer. The SEC has raised its scrutiny of continuation funds, especially around fee disclosure and conflicts of interest. OpenAI and SpaceX are also sensitive US technology assets. If Dragoneer raises money from sovereign wealth funds in the Middle East or pension funds in Asia, it has to consider CFIUS review. A foreign LP is not just an investor; it becomes a counterparty to the US national security apparatus. That risk does not show up in the private placement memorandum unless you look for it.
The competitive context matters too. The private equity secondary market has grown from roughly $40 billion in 2015 to more than $100 billion in recent years, with continuation funds now accounting for over 30 percent of the volume. Carlyle, Hellman & Friedman, Lexington Partners, and Ardian have all normalized this structure. Dragoneer’s only edge is not its legal architecture; it is its asset list. OpenAI and SpaceX are the two most exclusive private tickets on the planet today. Everything else in the memorandum is a supporting argument about why those tickets deserve a premium.
The macroeconomic window also favors the GP. Global central banks are moving toward easier policy. When the risk-free rate drops from five percent to three percent, pension funds feel the yield gap and start calling private equity managers. Dragoneer is raising into that window. It is the same mechanics as a crypto project launching a token during a liquidity bull run: the window flatters the sales process, but it does not change the underlying valuation risk.
And then there is the tax angle. Carried interest tax reform is a recurring political target. A GP facing the possibility of a higher top marginal rate on carried interest has a strong incentive to roll existing gains into a new vehicle now, before the rules change. A minimal discount minimizes the taxable gain at the moment of transfer while preserving the economic upside for the new carry schedule. That is not a conspiracy. It is a calendar.
The Contrarian Angle
The crypto world will read this story and immediately imagine tokenized shares of OpenAI and SpaceX. That is the wrong takeaway. Dragoneer is not building toward a security token offering. It is proving that traditional finance can manufacture private liquidity with documents, legal opinions, and GP relationships alone. No chain, no oracle, no settlement layer. Just a limited partnership agreement.
That is a problem for crypto. If the smartest money can get access to the most coveted private assets without using blockchain, then tokenization is not a functional requirement. It is a marketing overlay. The crypto grants, the open secondaries, the so-called democratization of liquidity — all of that sounds revolutionary. But Dragoneer is solving the same problem with a closed group of qualified purchasers. The real value in this fund is not the shares. It is the exclusivity.
So the contrarian angle is not “this fund is bad.” The contrarian angle is that the buyer needs to know the fee tax they are paying. Beta is the tax you pay for ignorance. A new LP in this fund is paying a carried interest premium for the privilege of holding two hyper-momentum assets that everyone already knows are good. That is not alpha. It is beta wrapped in an obscure legal structure and sold as access.
If Dragoneer’s true edge is the ability to source future OpenAI shares, then new LPs should demand a larger discount to compensate for the opacity. A minimal discount says the GP does not think it needs to share that edge. Maybe it does not. But the phrasing should still set off the sanity check: Sanity checks before sanity wins.
The crypto-native reader should also ask why this story is on Crypto Briefing. Dragoneer is not running a token sale, but it is marketing to a community that assigns high value to narratives. The Web3 investor class, family offices, and crypto fund-of-funds are all potential LP candidates. They are also the least likely to demand GP-level transparency, because they are used to relying on TVL numbers and community hype. If a Web3 fund buys a continuation fund slot based on the OpenAI logo, it is doing the same mistake as buying a farm token without reading the harvest contract.
The Takeaway
Watch the offering memorandum, not the press release. Look for the independent valuation process. Look for the note on CFIUS. Look for the specific discount range. If the discount is truly minimal, then the GP has extracted maximum control over an asset that will remain illiquid for years. That is a brilliant negotiation from the GP. It is a terrible entry for an LP.
Dragoneer is telling you that OpenAI and SpaceX are so scarce that they deserve no discount. Maybe they are. But scarcity is not the same as price stability. Liquidity is the only truth in a fragmented chain. Here, the chain is broken by design. The algorithm executes, but the human decides the discount. That human is the seller. Do not confuse your seat at the table with a seat on the pricing committee.
OpenAI and SpaceX are extraordinary companies. The continuation fund is not the asset. It is the gate. Check the gate. The ledgers do not lie, only the auditors do.