Hyperliquid Revenue vs Volume: What Record Months Actually Earn the Protocol

Hyperliquid

The headline and why it seems contradictory

In early August 2026 several outlets ran a version of the same story: Hyperliquid was posting record trading volume while protocol revenue fell sharply and HYPE buybacks shrank. Read quickly, that sounds impossible. A perpetuals exchange charges a fee on every fill, so more fills should mean more money.

It is not impossible, and the mechanism is not mysterious. It is just that "revenue" on a perpetuals venue is the product of three separate numbers, and reporting usually collapses them into one. Volume is the first. The effective take rate, meaning fees collected divided by volume traded, is the second. The share of those fees the protocol actually retains, rather than paying out to the people providing liquidity, is the third. Any of the three can move on its own, and over the past year all three have.

This page assembles the series month by month so the divergence is visible rather than asserted. Every figure below comes from a public source named at the point of use, with the date we read it.

The revenue series, month by month

These are monthly totals of daily protocol fees and daily protocol revenue for Hyperliquid, summed from DefiLlama's public fees API, read on August 11, 2026. In DefiLlama's framing, fees are everything collected from traders and revenue is the portion the protocol retains rather than paying out to liquidity providers.

August 2025 was the peak on both lines: $144.97M in fees and $113.86M in revenue. July 2026 produced $38.42M of revenue, which is the lowest complete month anywhere in the window we pulled, narrowly below March 2025's $38.69M, and about 66 percent below the August 2025 peak.

The right-hand column is the part worth staring at. It was 94.2 percent in June and July 2025. It has not been above 79 percent since. In July 2026 it was 69.7 percent. That drift is slow, it never produces a headline on any single day, and it compounds.

The volume series over the same period

Monthly perpetual volume, summed from daily figures in Hyperliquid's own public statistics feed, read on August 11, 2026.

The take rate column is computed, not reported: it is that month's fees divided by that month's volume, expressed in basis points. A basis point is one hundredth of one percent, so 1.59 bps means the protocol collected about $159 in fees per million dollars traded.

One comparison inside this table does most of the explanatory work. December 2025 saw $353.7B of volume and $68.75M of fees. March 2026 saw $435.8B of volume, about 23 percent more, and $69.49M of fees, about 1 percent more. Almost a quarter more trading, essentially the same money. That is the divergence in miniature, three months before it became a headline.

Why they diverged

Three forces, working on different timescales.

Fee-tier mix. Hyperliquid's fee schedule charges less as a trader's rolling volume rises. Base tier is 0.045 percent taker and 0.015 percent maker on perps. The largest accounts pay materially less. When growth comes disproportionately from a handful of very large traders and market makers rather than from many small ones, volume climbs while the blended take rate falls. This is the ordinary maturation path for any venue that wins professional flow, and it is not a problem so much as a repricing.

Maker rebates and the maker-taker balance. Fees are collected from takers and partly paid back to makers. A month where a larger fraction of volume is passive, resting liquidity rather than aggressive crossing produces less net fee income per dollar traded. The take rate rising through late 2025 and then falling through 2026 is consistent with the maker-taker balance shifting back after a period of unusually aggressive flow.

The retained share, which is the big one. The gap between the fees column and the revenue column is money collected from traders that does not land on the protocol's revenue line. It goes to the people supplying liquidity, most visibly through HLP, the protocol's vault, and through the builder share on HIP-3 markets, where a portion of every fee routes to whoever deployed the market. HIP-3 venues such as trade.xyz charge 0.09 percent taker and 0.03 percent maker precisely because a cut goes to the builder rather than the protocol. As HIP-3 markets grew into a real share of activity, the blended retained share fell. That is a deliberate design choice working as intended, and it is also the single largest contributor to revenue falling faster than fees.

What halved buybacks mean for HYPE holders

The retained portion of fees is what funds the Assistance Fund's open-market purchases of HYPE. The revenue line is therefore the ceiling on how much buying the protocol can do in a month. When revenue falls by a third, the capacity to buy falls with it.

We are not publishing a buyback dollar figure. The rec that prompted this analysis asked for one, and we could not verify a monthly buyback total against a primary on-chain source we would be willing to stand behind. Putting an unsourced number next to sourced ones would undermine the rest of the page, so the honest version is this: buyback capacity tracks the revenue column above, and readers who want the exact spend should read the Assistance Fund's transactions directly rather than take our word or anyone else's.

What the revenue series does support saying is narrower and more useful. Buyback pressure on HYPE is not a fixed subsidy. It is a variable that moves with the retained share of fees, and the retained share has been falling steadily for a year for structural reasons rather than cyclical ones. A holder modelling future buyback support from a mid-2025 revenue run rate is using a number that no longer describes the protocol.

What to watch next month

Four things, in rough order of how much they would change the picture.

The retained share is the number to track. If it stabilizes around 70 percent, the revenue decline was a repricing that has finished. If it keeps sliding, the trend has further to run and every forward revenue estimate anchored on 2025 is too high.

Whether Hyperliquid's public volume feed resumes. Its silence since April 2026 is the reason the volume half of this analysis is short, and it is the reason the "record volume" claims are hard to verify independently.

The HIP-3 share of total activity. Builder markets are the clearest structural explanation for the falling retained share, so their growth rate is a direct input to the revenue line.

Whether fees stabilize independently of volume. June 2026 fees of $80.98M against May's $62.18M showed the fee line can still move sharply upward. One month is not a trend, but a second consecutive recovery month would argue the take rate compression has bottomed.

Methodology

Monthly fees and revenue are summed from the daily series in DefiLlama's public fees API for Hyperliquid, at api.llama.fi/summary/fees/hyperliquid, collected on August 11, 2026. Fees are the total collected from traders; revenue is the portion DefiLlama classifies as retained by the protocol rather than paid to liquidity providers. Monthly perpetual volume is summed from the daily series in Hyperliquid's public statistics feed at d2v1fiwobg9w6.cloudfront.net/daily_usd_volume, collected on the same date; that feed's most recent data point is dated April 3, 2026, which is why the volume table ends with March 2026.

The effective take rate column is our own calculation, not a reported figure: monthly fees divided by monthly volume, expressed in basis points. Month boundaries are calendar months in UTC. Figures for the current month are excluded from all tables because they are partial.

Where our arithmetic differs from figures reported elsewhere, we have said so and shown both rather than reconciling silently. No figure on this page is estimated, interpolated, or carried over from a secondary source.

If you want a second opinion on the same period, two independent groups publish work on Hyperliquid's financials. The Hyperliquid Research Collective, co-founded by Four Pillars and GLC Research, publishes annual and quarterly reports in a TradFi format. ASXN maintains live dashboards for buybacks and, since August 2026, cross-venue perps liquidity. Neither is affiliated with Hyperliquid, and neither uses our methodology, so treat differences in their numbers as a reason to check both rather than as an error in either.

This page is updated monthly as new complete months close. You may republish these figures with attribution and a link to https://hyperliquidguide.com/ecosystem/hyperliquid-revenue-vs-volume.