Real Yield in DeFi: Which Protocols Pay From Real Fees
Use this to sort a DeFi yield into real or subsidised: Hyperliquid and Aero pay holders from fees, Curve prints more tokens than it earns, and restaking earned $45,098 in a week.

Introduction
A decentralized finance (DeFi) token ↗ can advertise a double-digit yield while the protocol behind it earns almost nothing. In the 30 days to 15 November 2026, only 8.4% of the $2.15B in fees tracked by DefiLlama reached token holders, and the restaking category earned $45,098 in a week on $7.51B of locked value (DefiLlama, 2026). Real yield names the payouts that user fees fund, as opposed to tokens a protocol prints. This guide defines the term with DefiLlama's fee, revenue and holders-revenue data, compares ten protocols, sets holders revenue against new token supply, and ends with five checks that apply to any yield. Uniswap, Aave and Hyperliquid get separate treatment because their buyback status is dated or disputed. Every figure carries a date and a source.
Key Takeaways
- Only 8.4% of the $2.15B in fees DefiLlama tracked in the 30 days to 15 November 2026 reached token holders; protocols kept 54.6% as revenue (DefiLlama, 2026).
- Hyperliquid passes 77% of fees to token holders, Aero, formed when Aerodrome and Velodrome merged on 21 October 2026, passes 75% and GMX 74%, while Lido, Aave and Ethena record 0% over the same window (DefiLlama, 2026).
- In the 30 days to 8 October 2026, Aerodrome earned $1.19 of holders revenue per dollar of new tokens, Uniswap $0.83 and Curve $0.29, a measured gap between revenue and emissions.
- Uniswap's fee switch, voted on 25 December 2025 and extended to v4 on 27 July 2026, lifted holders revenue from $2.9M in January to $15.5M in September, before $8.5M in October.
- Restaking earned $45,098 in fees over seven days on $7.51B locked, an annualised 0.03%, against 3.07% for liquid staking.
What Is Real Yield in DeFi, and How Does It Differ From Token Emissions?
Real yield is a payout funded by fees that users pay, rather than by tokens a protocol prints. DefiLlama draws the same line in its definitions: block rewards and token emissions are incentives, a cost to the protocol, and are never counted as fees or revenue (DefiLlama, 2026).
Revenue Paid by Users
A trader pays a swap fee, a borrower pays interest, a perpetual-futures trader pays a trading fee. Part of that money goes to liquidity providers or lenders, and the remainder is protocol revenue. Real yield is the slice of revenue that reaches a token or its lockers. Hyperliquid, a derivatives exchange with its own blockchain ↗, collected $91.7M in fees in the 30 days to 15 November 2026, and DefiLlama attributes 77% of that to holders of its HYPE token (DefiLlama, 2026). Customers funded every dollar of it.
Emissions Paid by Dilution
Emissions are new tokens that a protocol mints and hands to depositors. Dilution funds them, so the cost falls on every holder of the token. The companion guide to yield farming ↗ covers the mechanics. One measured farm shows the pattern: the reward rate on Aerodrome's USDC-AERO pool stayed between 13.9% and 40.6% for a year, while the pool shrank 72% and its daily dollar emissions fell 77% (DefiLlama, 2026). The percentage held. The capital and the token price absorbed the decline. Aerodrome splits the two payers cleanly: staked liquidity providers earn AERO emissions, while the swap fees those pools generate go to voters who lock AERO (Aerodrome docs, 2026). One protocol therefore pays both a subsidy and a fee-funded yield, to different people.
Fees, Revenue and Holders Revenue
DefiLlama models the flow as an income statement. Fees are everything users pay. Revenue equals fees minus the supply-side share paid to liquidity providers, lenders and stakers. Token Holder Revenue is the part of revenue that reaches tokenholders through buybacks and burns, fee burns or direct distribution (DefiLlama, 2026). Treasury-funded buybacks do not count, and neither do mint-and-burn schemes. DefiLlama also defines Incentives, the value of tokens emitted as rewards, and Earnings, revenue minus incentives. "Real yield" is community shorthand, and holders revenue is the measurable version. Lido shows why the ladder matters: its fees of $54.7M are all staking rewards on staked ETH, its revenue is $3.1M, and holders revenue is zero in this window, after $2.3M in September (DefiLlama, 2026).
Fee data shows which protocols pay holders and which collect fees for themselves.
How Much of DeFi's Fee Revenue Reaches Token Holders, Protocol by Protocol?
Of $2.15B in fees recorded across DefiLlama's fee listings in the 30 days to 15 November 2026, $181M, or 8.4%, reached token holders (DefiLlama, 2026). Among ten protocols, the share runs from 0% to 77%, and the order has little to do with size.
Ten Percent Reaches Holders
Protocols kept $1.17B of those fees as revenue, 54.6% of the total, but two stablecoin issuers account for $716M of it: Tether $511.6M and Circle's USDC $204.0M (DefiLlama, 2026). Neither has a governance token to receive the money, so their holders revenue is zero. Lending shows the same gap. Morpho collected $34.1M in fees and DefiLlama records no revenue, because borrowers' interest passes entirely to lenders ↗. Ethena's USDe collected $18.3M against $48,700 of revenue. A high fee total says how busy a protocol is, and says little about what a token receives.
Who Receives the Fee
Four recipients take the money: liquidity providers, lockers of a governance token, the treasury, and the token itself through buybacks and burns. Sky (formerly MakerDAO) routes holders revenue to SKY buybacks and staking rewards; the DAI-to-USDS rebrand ↗ explains how the token changed. Aero pays voters who lock AERO. Curve sends 90% of its share of pool fees to lockers of CRV, plus any bribes that other protocols pay for gauge votes (DefiLlama, 2026). Jupiter has used 50% of platform revenue for JUP buybacks since 17 February 2025 (DefiLlama, 2026). Aave's treasury kept $5.6M of revenue, and the buyback question is taken up below. The recipient decides how permanent the payout is, because a treasury can redirect its share and a locker cannot.
Data current as of November 2026.
Revenue that reaches a token still has to be weighed against the tokens the same protocol issues.
Do Protocols Pay Holders More Than They Print in New Tokens?
Holders revenue per dollar of new tokens ran from $1.19 at Aerodrome to $0.29 at Curve in the 30 days to 8 October 2026 (DefiLlama and CoinPaprika, 2026). DefiLlama defines the gap as Earnings, revenue minus incentives, but publishes emissions only through a paid API, so the figures here use a proxy.
What the Token Supply Shows
The proxy is the change in circulating supply, market capitalisation divided by price, between 9 September and 8 October 2026 (CoinPaprika, 2026), valued at the 30-day average price. Aerodrome added 18.3M AERO worth $13.0M against $15.5M of holders revenue. Uniswap added 1.73M UNI worth $14.2M against $11.8M, a release consistent with the 20M-UNI annual growth budget in its governance proposal (Uniswap Governance, 2025). Curve added 16.7M CRV worth $5.9M against $1.7M. Pendle added 0.30M PENDLE worth $0.7M against $0.3M.
Market Cap Against Holders Revenue
Dividing market capitalisation by annualised holders revenue gives a multiple that shows what a token's owners receive. On 16 November 2026, using holders revenue for the 30 days to 15 November, Curve stood at 10.7 times, Sky at 17.9, Hyperliquid at 18.6, Jupiter at 19.6 and Uniswap at 31.7 (calculated from DefiLlama and CoinPaprika data, 2026). Aero's 1.3 rests on a CoinPaprika supply figure taken mid-migration and is the least reliable of the set. The common price-to-fees ratio misleads where fees pass through to others: Lido's market cap is about 0.34 times its annualised fees, because those fees are staking rewards paid to stETH holders. The multiples describe payouts and offer no verdict on price. Over 12 months, Hyperliquid's holders revenue was $680.7M and Aero's was $85.3M, so the 30-day figures sit inside a longer record (DefiLlama, 2026).
Where the Proxy Fails
CoinPaprika's supply series can lag unlock events. Hyperliquid, Aave, Sky, Jupiter, Lido and GMX each show a change under 0.6%, and for Hyperliquid the trackers disagree on unlock size: one reports 9.92M HYPE a month for core contributors, another a 3.75M-token cliff on 7 October 2026 (KuCoin, 2026; AlphaPilot, 2026). Neither is used. DefiLlama's own unlock and emission schedules sit on a separate dashboard that the free API does not expose (DefiLlama, 2026). Ethena's supply jumped 33% in the window, which looks like a data revision, and is excluded. A re-run for 16 October to 15 November 2026 could not be read: Aero's token migration cut CoinPaprika's AERO supply series by more than 70%, and UNI's fell by 17.4M tokens. A free-data comparison therefore covers four protocols in the September window, and the rest stay unmeasured.
Data current as of November 2026.
Reaching a token is one step; the route it takes decides how permanent the payout is.
How Do Fee Switches, Buybacks and Burns Move Revenue to a Token?
Revenue reaches a token by four routes, and three of the best-known programs have dates and disputes attached. A fee switch is a contract setting that diverts part of the fees paid to liquidity providers toward the token. The table lists the routes; the three cases below show how each is dated and where the data conflicts.
Uniswap: A Vote, Then a Burn
Uniswap governance passed the UNIfication proposal on 25 December 2025, with more than 125 million UNI voting for and 742 against (The Block, 2025). The proposal turned on protocol fees, so a Uniswap v2 pool's 0.30% fee splits into 0.25% for liquidity providers and 0.05% for the protocol (Uniswap Labs, 2025). Fees collect in a contract that releases them only when UNI is burned (Uniswap Labs, 2025). Proposal 100 extended the switch to Uniswap v4 on seven chains on 27 July 2026 (Uniswap Governance, 2026). DefiLlama's holders revenue for Uniswap rose from $2.9M in January 2026 to $15.5M in September, then fell back to $8.5M in October.
Aave: Buybacks in Dispute
Aave's decentralized autonomous organization (DAO) began buying AAVE on 9 April 2025 and had spent about $42M on 205,000 tokens by February 2026 (Aave Governance, 2026). It paused purchases from 19 April 2026 after the 18 April 2026 rsETH bridge exploit at Kelp DAO (Aave Governance, 2026). The Defiant reported automated buybacks live from 27 June 2026, while a 22 September report called them still suspended (The Defiant, 2026; Woofun, 2026). DefiLlama records $0 of holders revenue from May 2026 through 15 November, and on 22 October founder Stani Kulechov said buybacks would return in an automated form under Aavenomics 3.0, without a size, a start date or a governance proposal (DefiLlama, 2026; Bullrunners, 2026). On the data, the pause still holds, and the purchased tokens sit in a DAO reserve rather than being burned.
Hyperliquid: Held or Burned
Hyperliquid routes 99% of perpetual-futures fees to an Assistance Fund that converts them into HYPE automatically, and its documentation says the HYPE in the fund is burned (DefiLlama, 2026; Hyperliquid docs, 2026). Independent trackers describe a fund holding roughly 46 to 48 million HYPE, with only a small part destroyed (netsupply.org, 2026; DEXTools via BTCtiming, 2026). The two accounts disagree on whether holders gain from a burn or from a large standing balance. Either way, DefiLlama's holders revenue for Hyperliquid reached $70.4M in the 30 days to 15 November and $680.7M in 12 months.
Data current as of November 2026.
A yield that depends on one switch, one vote or one wallet is as durable as that switch, and each of the three can change in a single governance cycle.
What Did the Restaking Fee Collapse Reveal About Unsustainable Yield?
DefiLlama's restaking category earned $45,098 in fees in the seven days to 15 November 2026, an annualised 0.03% of its $7.51B in locked value (DefiLlama, 2026). Liquid staking earned $24.58M on $41.72B, an annualised 3.07%. The guide to EigenLayer restaking ↗ covers the mechanics; the fee data covers what the second layer of yield paid.
A Category Earning $45,098 a Week
CoinDesk reported on 28 September 2026 that the category produced $99,977 in the week to 8 September on $10.02B, against $27.35M for liquid staking on $51.87B, about 53 times more per dollar (CoinDesk, 2026). The same calculation on 15 November gives 98 times. Most of what DefiLlama counts as liquid staking fees ↗ is Ethereum's own staking reward, so restaking's 0.03% is the premium on top (Blockchain Dose, 2026). CoinDesk attributes the shortfall to ended points programs and to slashing, live since April 2025, which added a priced downside with no extra yield. The five largest liquid restaking tokens earned $953,350 in gross profit in the second quarter of 2026, against $2.18M three quarters earlier (CoinDesk, 2026).
EigenCloud's Twelve-Month Slide
EigenCloud (formerly EigenLayer) recorded $8.89M in monthly fees in September 2025 and $0.20M in September 2026, a fall of 98%. October 2026 brought $0.48M, still 95% below the September 2025 level (DefiLlama, 2026). DefiLlama records no revenue for it: all rewards go to suppliers. Capital left far more slowly than fees did. EigenCloud still shows $4.63B locked (DefiLlama, 2026), and ether.fi is removing its last tie to the platform this quarter (CoinDesk, 2026). A yield that pays 0.03% on top of base staking while adding slashing risk has no revenue underneath it.
Data current as of November 2026.
Fees that vanish while deposits stay show how a yield outlives its revenue.
Which Five Checks Separate Real Yield From Subsidised Yield?
Five checks, each answerable from free data, separate fee-funded yield from subsidised yield: who pays, who receives, how the payout compares with new tokens, whether the mechanism is switched on, and what the token costs. None requires trusting a protocol's own dashboard, since DefiLlama's fees pages and its public fees endpoints publish fees, revenue and holders revenue by protocol.
Five Checks for Any Yield
First, identify the payer: user fees fund real yield, and token emissions fund the rest. Second, find the recipient by comparing revenue with holders revenue, since a protocol like Morpho collects fees and passes all of them to lenders. Third, set holders revenue against the dollar value of new tokens, as in the table above. Fourth, confirm the mechanism is switched on today, because Aave's holders revenue was $4.0M in February 2026 and $0 from May through October. Fifth, divide market capitalisation by annualised holders revenue, which turns a payout into a price. A multiple of 11 and one of 32 describe very different claims on the same dollar of fees. Repeating the five checks each quarter takes minutes with the same endpoints and exposes drift early.
Where Real Yield Still Fails
Revenue falls with activity. A fee-funded payout shrinks when trading volume does, and no emission schedule cushions it. Real yield also carries contract, governance and data risks: DefiLlama's 30-day window changes daily, and a vote can pause a buyback in a day. A protocol that pays a high share of fees today can also be a small business: Pendle paid 69% of its fees to holders, but those fees totalled $0.53M in the 30 days to 15 November. Share and scale have to be read together. The comparison of RWA yield and DeFi yield ↗, the survey of on-chain yield from tokenized Treasuries ↗ and the guide to yield-bearing stablecoins ↗ apply the same payer test to other sources.
Conclusion
Real yield is a test of origin more than a rate. Fees paid by users fund it and emissions do not, so one percentage can describe a business or a subsidy. The ten protocols measured span the range: Hyperliquid, GMX and Aero pass most fees to token holders, Uniswap passes 11.2% and raised it by governance vote, and Ethena and Morpho collect fees that never reach a token (DefiLlama, 2026).
The good case and the bad case look alike until revenue falls. A fee-funded payout shrinks with trading volume but stays honest, while an emission-funded payout holds its percentage and loses capital, as Aerodrome's farm and EigenCloud's slide from $8.89M to $0.20M in monthly fees show. The difference surfaces after the token price has already moved, and buyback programs that a vote can pause add a second blind spot.
Before relying on a yield, look up holders revenue on DefiLlama, set it against the value of new tokens, and confirm, from a dated source, that the buyback or fee switch is running. Treat a single month as one point on a longer record. A payout that passes all three checks can still be small, so compare its scale as well as its share. Pendle paid 69% of its fees ↗ to holders on $0.53M of total fees, a share that means little without the scale.
Quick Stats
- 8.4% — share of tracked fees that reached token holders, 30 days to 15 November 2026
- $181M — holders revenue across DefiLlama's fee listings in that window
- 77% — Hyperliquid's share of fees paid to holders, highest among the ten protocols measured (DefiLlama, 2026)
- $680.7M — Hyperliquid's holders revenue over 12 months
- $45,098 — restaking category fees in the seven days to 15 November 2026
- 98% — fall in EigenCloud's monthly fees, September 2025 to September 2026
Data current as of November 2026.
FAQ
?Is "real yield" an official DefiLlama metric?
No. DefiLlama's data definitions page uses Fees, Revenue, Token Holder Revenue, Incentives and Earnings, and defines Token Holder Revenue as revenue distributed through buybacks and burns, fee burns or direct distribution (DefiLlama, 2026). "Real yield" is community shorthand for that last category. Treasury-funded buybacks and mint-and-burn schemes are excluded from the DefiLlama figure.
?Where can a reader check a protocol's fees and revenue directly?
DefiLlama's fees pages list every protocol, and its public endpoints return the same data. The summary endpoint takes a protocol name and a data type, such as dailyFees, dailyRevenue or dailyHoldersRevenue, and returns a daily series. The overview endpoint returns all protocols at once. Checking the 30-day and 12-month totals side by side shows whether a month is typical.
?Can a reader earn real yield without holding the governance token?
Yes. Swap fees paid to liquidity providers and interest paid to lenders are user-funded and need no governance token. On 16 November 2026, the Uniswap V3 USDC-WETH 0.05% pool showed a base rate of 21.19% over 24 hours, and Aave V3's USDC market on Ethereum paid suppliers 4.21% (DefiLlama, 2026). Both figures move daily, and neither is a guarantee.
?What happens to real yield when trading volume falls?
It falls with the volume, because fees are a share of activity. The Uniswap V3 USDC-WETH 0.05% pool's base rate read 21.19% over 24 hours on 16 November 2026, 23.76% as a 30-day mean and 54.36% a week earlier (DefiLlama, 2026). Fee-funded yield cannot be held steady by a schedule, which is the property that separates it from emissions.
?Is real yield taxable?
Treatment depends on the country and the structure, and this is general information rather than tax advice. Rewards paid out to lockers, such as those on Aerodrome or Curve, are distributions, and whether they count as income varies by jurisdiction. Buybacks and burns make no direct payment to holders, so no receipt arises, only a change in supply. A local tax adviser can confirm the rules that apply.
?Is a buyback the same as a dividend?
No. A dividend pays cash to each holder, while a buyback purchases tokens on the open market and pays nobody directly. A holder benefits only if the purchased tokens are burned or stay out of circulation, and Aave's purchased tokens sit in a reserve of the decentralized autonomous organization (DAO) that governance can spend again (Aave Governance, 2026). DefiLlama counts buybacks and burns funded by revenue as holders revenue, but the economic effect differs by design.
?Does DefiLlama publish token emissions for free?
Not through its public API. The emissions endpoint returned a paid-plan notice on 16 November 2026, while DefiLlama's definitions describe Incentives and Earnings as tracked metrics (DefiLlama, 2026). A free comparison therefore relies on proxies such as the change in circulating supply, which can lag unlock events and covers only some protocols.
References / Sources
Platform & Company Data
Datasets, definitions and official documentation behind the fee, revenue and token figures.
- DefiLlama: Fees and Revenue Overview and Summary Data (api.llama.fi, Nov 2026)
- DefiLlama: Data Definitions for Fees and Revenue (docs.llama.fi, Oct 2026)
- CoinPaprika: Market Cap and Supply History (api.coinpaprika.com, Nov 2026)
- Aerodrome: Emissions and Gauge Documentation (aerodrome.finance, 2026)
- Hyperliquid: Assistance Fund Documentation (hyperliquid.gitbook.io, 2026)
- Aave: Buyback Pause ARFC (governance.aave.com, 2026)
Governance & News Sources
Primary governance records and contemporaneous reporting on fee switches, buybacks and restaking.
- Uniswap Labs: UNIfication Proposal (blog.uniswap.org, Nov 2025)
- Uniswap Foundation: Activate v4 Protocol Fees (vote.uniswapfoundation.org, Jul 2026)
- The Block: Uniswap Passes UNIfication (theblock.co, Dec 2025)
- CoinDesk: Restaking Gold Rush Is Over (coindesk.com, Sep 2026)
- The Defiant: Aavenomics 3.0 Goes Live (thedefiant.io, Jun 2026)
- Bullrunners: Aave Buybacks Under Aavenomics 3.0 (bullrunners.com, Nov 2026)
- Netsupply: HYPE Burn and Unlock Ledger (netsupply.org, 2026)
Original CoinPaprika guide.