Hyperliquid's $240B Month Is Not a Revenue Record: What the HIP-3 Fee Split Changed
Hyperliquid volume hit $240B over 30 days while revenue fell four quarters running. HIP-3 builders keep up to half the fees, and the HYPE buyback shrinks.
$ Stop reading delayed data. Read live order book depth on the 100 most liquid Hyperliquid pairs right now.
Launch Free Terminal →HYPE printed a fresh all time high on September 19 and Hyperliquid's trading volume over the prior 30 days ran to roughly $240 billion, according to Cryptonomist. Over roughly the same stretch the protocol's quarterly revenue has fallen four quarters running. Those two facts are not in tension. They are the direct result of a fee split most traders have never read.
Four straight quarters of falling revenue behind a record volume print
Hyperliquid's gross protocol revenue peaked near $357 million in the third quarter of 2025 and has declined every quarter since, to roughly $295 million, then about $217 million, then approximately $202 million in the second quarter of 2026, CoinDesk reported using DefiLlama data. That is a 43% drop from the top.
The trade count moved the other way over the same period. So did open interest, which reached just above $11 billion on July 13, the platform's high for 2026 at the time, with Hyperliquid settling roughly 9% of all open perp positions worldwide including centralized venues.
A venue can set volume records and collect less money. The mechanism is not subtle.
HIP-3 hands up to half the trading fee to the builder
Since October 2025, anyone who stakes 500,000 HYPE can deploy their own perpetual futures market on Hyperliquid's order books and keep up to half the trading fees, per the HIP-3 specification.
At the start of 2026 these builder deployed markets were about 2% of Hyperliquid's perp volume. By August they were roughly half of it, CoinDesk reported.
The pass through is visible in the accounts. Cost of revenue, meaning the share of fees Hyperliquid hands straight to builders, market makers and its own liquidity vault, was under 6% of gross revenue in the second quarter of 2025. A year later it was 18%.
Builder code fees, charged by front ends for routing an order, arrived as roughly $16 million of revenue in the second quarter of 2026 and left as roughly $16 million of cost in the same quarter. Every dollar of it passes through.
The buyback is a function of revenue, not volume
Roughly 97% of trading fees route into the Assistance Fund, which buys HYPE on the open market and retires it. About 44.5 million HYPE have been removed from supply this way.
The size of that bid tracks earnings directly. The fund bought close to $290 million of HYPE in the third quarter of 2025 and roughly $149 million in the second quarter of 2026, a little over half as much.
This is the part that matters for anyone sizing a HYPE position off volume headlines. A record volume month with a lower fee capture rate produces a smaller structural bid than a quieter month did a year earlier. Volume is the input. Revenue is what actually buys the token.
The same compression reached the HLP vault. HLP fee receipts fell from $7.15 million in the third quarter of 2025 to $1.5 million in the second quarter of 2026, per reporting on the builder fee split, with HLP receiving roughly 1% of perpetual and spot trading revenue. HLP returns now ride strategy PnL rather than a fee subsidy, which changes what the vault's performance tells you about market maker conditions on the venue.
One deployer carries most of the open interest
Trade.xyz accounts for more than 90% of all HIP-3 open interest, CoinDesk reported. Real world asset perps on contracts including crude oil, gold, Nvidia, Tesla and pre IPO names reached a record $3.6 billion in open interest in August and overtook bitcoin as the platform's largest market by that measure.
Between July 13 and July 19, tokenized stocks and commodities did $25 billion in volume, 52% of the weekly total, outpacing crypto perps for the first time. These contracts settle in stablecoins, never expire, and keep trading through the weekend when the New York Stock Exchange is closed.
Concentration like that turns one firm's oracle choices and margin settings into a venue wide risk. In early August a single trade on a thin Korean pre market venue moved Trade.xyz's SK Hynix contract 19% and triggered liquidations the firm agreed to reimburse. The contract behaved correctly relative to its reference price. The reference price was the problem.
For a trader that is a fill quality question, not an abstract governance one. When the underlying venue feeding a contract's oracle is thin, the price you get liquidated at is set by a book you are not watching.
Reading volume as a trader rather than a headline
Volume tells you how much turnover a venue processed. It does not tell you how much of that turnover the protocol kept, and it says nothing about how positioning changed.
Three cleaner reads. Open interest, which measures positions actually held rather than churn. The split of that open interest between core perps and HIP-3 markets, since the two carry different fee economics and different oracle risk. And funding, which prices the cost of holding whatever position the volume created.
Buildix covers Hyperliquid natively, HIP-3 markets included, so the screener shows open interest, funding and orderflow across core and builder deployed contracts in the same table instead of treating them as separate universes.
A record on a chart is a fact about activity. Whether it is also a fact about the money requires knowing who keeps the fee, and since October 2025 the answer on roughly half of Hyperliquid's volume has been: not Hyperliquid.