Hyperliquid, OpenAI, and the strange new shape of 'going public'
Two pieces of news landed inside twelve hours of one another on the morning of 9 June 2026, and the financial press is dutifully filing them under different desks. One is a perpetual-futures exchange that, by one research note's reckoning, has accounted for nearly half of every dollar spent on crypto token buybacks in 2025. The other is a confidential S-1 filed by a company most readers will recognise by its products, not its legal name, weighing whether to go public at all. They are not the same story. They are, however, the same shape.
The shape is this: in 2026 the boundary between "listed" and "unlisted" capital has become a marketing choice rather than a regulatory one, and the most consequential venues — whether they are decentralised order books or frontier-model labs — are quietly designing around it.
The venue that decided to be its own market-maker
Citrini Research, in a note circulated overnight, called the $HYPE token "a compelling investment" on the strength of a single, striking figure: Hyperliquid, the on-chain perpetual futures venue, has driven close to half of all crypto token buybacks recorded in 2025, according to the Telegram-distributed copy of the research. The mechanism is straightforward once one stops thinking of "exchanges" the way the 2018 crypto press did. Hyperliquid runs an on-chain order book, charges fees, and routes a share of protocol revenue to systematic buybacks of its native token. The result is a venue that effectively pays its users to be its market-maker — a closed loop that is at once a treasury policy, a user-acquisition tool, and a quiet admission that the trader's relationship with the exchange is now an equity relationship wearing a token's clothes.
What makes the figure unusual is not the existence of buybacks. Public companies have repurchased their own stock for decades. What is unusual is doing it through a perpetual-futures interface, with no prospectus, no 10-Q, and a token whose float is governed by code rather than a transfer agent. The Citrini note frames $HYPE as a buyback story; the more honest framing is that Hyperliquid has built a listed-equity experience for users who never wanted to deal with a regulator. Whether that is innovation or evasion depends on who one asks, and on whether the next downturn tests the assumption that a token, and not a balance sheet, can absorb a panic.
The lab that filed to decide later
Twelve hours earlier, the same Telegram wires carried a single-line confirmation that OpenAI has filed a confidential S-1 for a potential initial public offering, with "no timeline set" and the explicit framing that the company is weighing the tradeoffs of going public against remaining private. Confidential S-1 filings, permitted under the JOBS Act framework for emerging-growth companies, allow issuers to submit draft registration statements to the Securities and Exchange Commission for review without making them public. The substance of OpenAI's filing is unknown. The signal is not.
A confidential S-1 is a company telling its bankers, its largest shareholders, and the regulator that it wants the optionality of a public listing without the obligations of being public yet. For a company that has spent the last two years signing compute commitments measured in the tens of billions of dollars, that posture is less a financial decision than an infrastructure one. The market that would absorb an OpenAI float is the same market that has, in the last quarter, learned to underwrite infrastructure as a multi-decade capex story — the way it once underwrote pipelines and rail.
The connective tissue, in plain prose
Read the two stories side by side and a single pattern emerges. In both cases, the issuer is using a public-market format — a perpetual-futures token, a confidential S-1 — as a way to capture the benefits of a listed venue while minimising the friction. Hyperliquid externalises its capital structure onto a chain that operates twenty-four hours a day, across borders, without a listing venue. OpenAI is buying time inside the SEC's review process so that it can choose its moment, with the public markets functioning as a contingent backstop rather than a permanent commitment. Both are, in effect, optionality plays dressed as financing events.
That is not a critique. It is a description of where the cost of issuance has moved. The traditional IPO is expensive, slow, and forces disclosure on a fixed calendar. A token buyback funded by protocol revenue is fast, continuous, and disclosure-light. A confidential S-1 is the regulator's own concession that the binary "public or private" choice no longer fits the way large tech companies actually raise and deploy capital. The press will frame these as crypto and AI stories. They are both, more accurately, capital-structure stories, and the lesson is that the boundary of the listed market is no longer drawn by the regulator — it is drawn by the issuer.
What this publication finds, and what remains contested
Three things follow. First, the Citrini research note is the strongest external claim yet that a single venue has come to dominate a market-wide flow, and the underlying figure — that Hyperliquid has driven close to half of all crypto token buybacks in 2025 — warrants independent replication. The sources cited here do not specify the methodology by which "all crypto token buybacks" was measured, and competing trackers, including those run by major data providers, may produce different totals depending on what counts as a buyback. Readers should treat the share as directionally correct, not as audited fact.
Second, the OpenAI filing is real but the size, the timing, and the lead underwriters are not in the public record. Confidential S-1 submissions, by design, are not disclosed. Any specific dollar figure attached to a putative OpenAI IPO in the next six months is, at this point, a guess. The honest position is that the filing is evidence the option is being kept warm, not evidence the option is being exercised.
Third, the two stories share a common blind spot. Neither is yet a stress test. Buyback-funded tokens have not been through a Hyperliquid-specific liquidity shock large enough to break the loop. A confidential S-1 has not yet been forced into the public, full-disclosure form that an actual listing requires. Until one of those tests happens, both venues are operating on trust — trust in code, in the SEC, and in the assumption that the cost of capital in 2026 will look roughly like the cost of capital in 2027. That is, historically, the most expensive assumption a balance sheet can make.
This publication is filed under the same wire the desk reads. Where the wires split, Monexus splits with them — and asks which framing the issuer actually wants you to see.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://t.me/s/cointelegraph
- https://t.me/s/cointelegraph
- https://t.me/s/cointelegraph