The market is still pricing HYPE on its unlock schedule. That is the wrong variable to anchor on. Hyperliquid is the largest cash generating business in crypto that does not issue a stablecoin, and it routes nearly all of that cash into buying its own token off the open market, every day, with no discretion and no announcement. Price the cash flow first. The supply schedule is a question of entry, not of thesis.
Every flat to falling crypto tape rewards the same trade, and it is never the L1 with a new consensus model or the latest points farm. It is the exchange. The venue earns its cut whether you are long or short, and it earns more of that cut when volatility spikes and volume runs. Arthur Hayes made this argument in March, pointing back to GMX running to a $90 all time high in April 2023 while the rest of the market was still picking itself off the floor. The pattern holds because the pattern is not a narrative. It is the rake.
HYPE is the cleanest version of that rake this asset class has produced. It made fresh all time highs in mid June, trading in the mid seventies after a January low near $20, and it did it while US spot Bitcoin ETFs bled roughly $5.6bn over the same weeks. That divergence is the whole story compressed into one line. Capital is leaving the asset everyone already understands and finding the one most people still cannot explain, because one of them is a passive store of value waiting on liquidity and the other is a business that prints cash and hands it back.
The flywheel
Start with the mechanism, because everything else hangs off it. Hyperliquid earns fees on perpetuals, on spot, on builder deployed markets, and now on stablecoin collateral. Citrini Research puts the run rate near $1.06bn in annualized fees with more than 90 percent directed to an Assistance Fund. DefiLlama's live tracker is more conservative as activity has cooled, showing close to $1.0bn in annualized fees and roughly $760m of annualized revenue. Either figure makes Hyperliquid the highest revenue protocol in crypto outside the stablecoin issuers.
The Assistance Fund is the engine. Somewhere between 97 and 99 percent of net protocol fees flow into it, and the fund continuously buys HYPE on the open market and holds it off the float. This is a near total earnings payout delivered as a repurchase rather than a dividend. By volume of capital returned, no other project in crypto comes close. Citrini estimates the buyback has absorbed more than $2bn of HYPE since January 2025, close to half of all crypto sector buybacks for the year, and runs at roughly 7 percent of token market cap annually. Set that against the discretionary quarterly burns at the centralized exchanges and the difference is structural, not cosmetic.
The mechanism is also counter cyclical by construction. The fund spends a fixed number of dollars, so it retires more tokens when the price is low and fewer when it is high. The bid gets stronger precisely when the tape gets weaker. That is the feature that lets HYPE behave differently from the rest of the book in a drawdown.
Now the honest part. Revenue is off its peak. Fees topped out in the third quarter of 2025 near $309m and have stepped down for three consecutive quarters as the entire perp DEX sector cooled and incentive funded rivals like Aster and Lighter fragmented volume. Monthly volume peaked above $1.2trn in October 2025 and fell to roughly $629bn by April 2026. DefiLlama currently shows fees down around 45 percent and volume down around 54 percent from the highs. The bull case is not that this never happened. The bull case is that the buyback runs regardless of the cycle, that the decline was sector wide rather than a loss of position, and that the franchise reaccelerates from a base that is still the largest in the category. State the tension plainly, then weigh it.
Real volume
The sharpest analytical point Hayes contributed is that the only honest way to rank these venues is average daily volume over open interest. Volume is trivial to fake through wash trading and reward farming. Open interest requires real capital sitting at risk. On that measure Hyperliquid carries the lowest ratio among the top five perp venues, which means the most real volume per unit of advertised volume. Aster, by contrast, ran volume at roughly eight times its open interest, the signature of a farm. As rival incentive programs sunset, that farmed volume evaporates and real flow gravitates back to the venue with the tightest execution. Hyperliquid is typically the cheapest place to fill a $100k to $10m clip, which is why size shows up there even when a competitor advertises a marginally lower headline fee.
Market share figures vary entirely on the denominator you choose, and any honest writeup should give the range rather than the flattering number. Strict 30 day perp DEX share on DefiLlama sits near 32 percent. Trailing dominance trackers put it between 40 and 56 percent. And against the entire perpetuals market, centralized venues included, Hyperliquid recently set a record near 7.6 percent. On open interest and real volume the lead is wider than the headline volume share implies.
From exchange to platform
The growth engine is HIP-3, live since October 2025, which lets anyone staking 500,000 HYPE deploy a permissionless perpetual market on anything they like. Equities. Commodities. Currencies. Pre IPO names. This is the shift from running an exchange to running infrastructure other people build on, and the token captures value from every trade hosted on top of it. Grayscale's framing is the correct one. Hyperliquid is closer to Amazon Web Services than to a stock exchange.
The traction is real. Builder deployed markets have done roughly $300bn in cumulative volume and peaked above $3.2bn in open interest in June 2026, with stock linked perps alone running near $18.8bn monthly. The first S&P 500 perpetual went live in March 2026 after the index was formally licensed to a HIP-3 deployer. The marquee proof point came on June 12. When SpaceX held the largest IPO in history, three centralized exchanges cancelled their tokenized SpaceX allocations citing share shortages and left traders with nothing, while the SPCX pre IPO perpetual on Hyperliquid traded continuously and did $1.4bn in volume on IPO day, roughly 30 percent of all builder market activity, against a $26m daily average over the prior three weeks. The same pre IPO market had already called Cerebras within about 1.3 percent of its Nasdaq open. This is the disruption thesis stated in numbers rather than adjectives.
Wall Street shows up
The demand side now has a second leg the prior cycle never had. Three US spot ETFs hold the token directly. 21Shares launched THYP on May 12, Bitwise launched BHYP on May 15, and Grayscale launched HYPG on June 3 with a staking yield near 2.2 percent. Combined, they have pulled roughly $172m in cumulative net inflows with no negative weekly outflow since launch, against the roughly $5.6bn that left Bitcoin ETFs over the same window. Bitwise additionally directs 10 percent of its management fee into buying and staking HYPE on its own balance sheet. Its CIO Matt Hougan put the framing bluntly, calling the market one percent penetrated.
The point that matters is the stacking. These flows sit on top of the Assistance Fund, not beside it. Protocol cash flow and ETF management fees are both converting into recurring spot demand for the same token, through two completely separate channels, neither of which depends on the next retail mania.
The stablecoin tax, reclaimed
For most of its life Hyperliquid earned nothing on the roughly $5bn of USDC sitting on it as margin. That collateral generated something like $150m to $220m a year in reserve yield, and all of it went to Circle. The protocol fixed this. After a validator vote in September 2025 handed the USDH ticker to Native Markets, the structure shared reserve yield back to the Assistance Fund. Then in May 2026 Coinbase and Circle effectively capitulated to the model. Under the new arrangement Coinbase became the official USDC treasury deployer, Circle staked 500,000 HYPE toward validator status, and validators activated a system routing roughly 90 percent of USDC reserve yield to HYPE buybacks. Coinbase itself estimated the arrangement could add as much as $200m a year. That is recurring, rate sensitive revenue feeding the same buyback, won from a counterparty that until recently kept all of it.
What it is worth
On fully diluted value, somewhere near $65bn to $70bn at current prices, HYPE trades around 61 times trailing earnings. That is expensive against Coinbase near 38 times, ICE near 22, and CME in the high twenties to forties. On circulating market cap, roughly $15bn to $18bn, the same earnings imply closer to 14 times with a buyback yield near 7 percent, and DefiLlama prices it at a price to fees ratio around 17.5 times.
The bull argument is not that the multiple is cheap on a screen. It is that the earnings are higher quality than the comps. Roughly 83 cents of every fee dollar reaches HYPE holders through the buyback. At Coinbase, well under 20 cents of each revenue dollar lands as net income, and none of it returns to holders with anything like the same completeness. Pair that payout efficiency with a growth rate several times faster than any listed exchange and a premium multiple is the rational outcome, not the speculative one.
The published targets should be read as models, not facts, and attributed as such. Hayes set an August 2026 target of $150, built on a rerating from roughly 12 times earnings toward the 25 times multiple CME carries, and argued that under 4 percent of centralized perp volume migrating on chain gets the franchise back to a $1.4bn run rate it already touched in August 2025. CF Benchmarks ran a probability weighted framework that lands near $106 per token, roughly 90 percent above its early June reference, with a bull scenario above 200 percent and a bear scenario near 30 percent down. Treat all of it as a range. And note that Hayes is a hype man who talks his book and trades it, having sold 247,000 HYPE for over $18m on June 4. His thesis stands on the data, not on his conviction.
The overhang
The real risk is supply, and a bullish piece that buries it is not worth reading. The entire scheduled overhang is a single bucket: roughly 238 million tokens for core contributors, about 24 percent of supply, vesting near 9.9 million per month through late 2027, with the next unlock on July 6. There is no venture cliff and no early investor unlock waiting behind it. At current fees and price the buyback retires only about one sixth of each month's authorized unlock, which means net float is still expanding by roughly 8 million tokens a month. Say it directly. The unlock caps the upside in the near term the same way the buyback caps the downside.
Three things sit on the other side of that. The claims are discretionary and have run far below the authorized maximum, with cumulative claimed supply a small fraction of what has technically vested at the contract level. Contributors are long the franchise and can stake for yield near 2.2 to 2.4 percent plus fee tier discounts, which dampens the incentive to dump. And an unlock is eligibility to sell, not selling. The asymmetry also inverts in late 2027. Once the vesting schedule rolls off against a roughly fixed float, the same buyback that cannot fully offset issuance today turns the token net deflationary.
The rest of the risk ledger is shorter but real. Validator concentration is meaningful, with a small set of validators and the top three pools controlling a sizable share of stake. The March 2025 JELLY delisting and force close set a precedent for intervention that cuts against the decentralization story. The chain saw security incidents and socialized vault losses in 2025. The builder market structure is fragile, with the SPCX perp dropping roughly 45 percent in thirty minutes during one early June session and over 400 users liquidated, and with the overwhelming majority of builder volume still routed through a single deployer. The venue runs without KYC and remains geoblocked in the US. The buyback is protocol policy, not a contractual obligation. And HYPE still trades as a high beta crypto asset. The scenario that overrides the entire structural case is a severe Bitcoin drawdown that crushes volume and price at the same time, because low fees and low token retirement is the worst possible combination for the flywheel.
Positioning
We are long and have been long for a long time.
Base case, roughly 55 to 60 percent. Fees stabilize and grind back toward a $1.2bn to $1.4bn run rate, ETF inflows persist, builder market open interest makes new highs, and the multiple rerates higher on a larger revenue base. That path supports something in the $110 to $140 range over twelve months.
Bull case, roughly 20 to 25 percent. Reacceleration arrives alongside a scaling real world asset franchise, more pre IPO listings, the ETF complex compounding AUM, and the AQAv2 revenue leg ramping. That is CF Benchmarks bull territory and Hayes $150 or beyond, call it $180 to $220 plus.
Bear case, roughly 20 percent. Sector volume keeps bleeding in a flat to up tape, rivals retain volume after their incentives lapse, contributor selling accelerates into the unlocks, and ETF flows stall. The buyback cannot absorb the float and price works back toward the recent base near $45 to $55.
Stay long while perp share holds its range, 30 day fees reaccelerate toward a $1.2bn to $1.4bn annualized run rate, builder market open interest makes new highs above $3.2bn, ETF inflows continue with few or no outflow days, and realized contributor claims stay well below the 9.9 million monthly maximum.
Trim or exit if perp DEX share erodes below 30 percent with rivals keeping their volume after incentives lapse, 30 day fees make lower lows for two or more consecutive months in a flat to up market, the buyback policy is changed, contributor selling steps up materially, or the ETFs flip to sustained net outflows.
The market is still trading the supply. The business is busy retiring it.