Crypto protocols spent more than $1.4 billion repurchasing their own tokens in 2025, according to [DWF Labs](https://www.dwf-labs.com/research/547-token-buybacks-in-web3). The results are mixed. Hyperliquid's Assistance Fund accumulated $1.04 billion in HYPE at an average cost of $24.3, generatin...
"If you own AAVE, you own not just the economic rights of the protocol, but the brand, the users, and the integrations." — Stani Kulechov, Founder, Aave
Crypto protocols spent more than $1.4 billion repurchasing their own tokens in 2025, according to DWF Labs. The results are mixed. Hyperliquid's Assistance Fund accumulated $1.04 billion in HYPE at an average cost of $24.3, generating unrealized gains as the token traded near $42 — a 500% surge correlated with its fee-funded buyback model. Jupiter spent $70 million on JUP repurchases and watched the token fall 89% from its peak, overwhelmed by 53 million tokens unlocking monthly through June 2026. The divergence raises a structural question: when do buybacks return value to token holders, and when do they subsidize insider exits?
The week ending April 14, 2026 brought two events that sharpen this question. Aave DAO passed the "Aave Will Win" proposal, directing 100% of product revenue — estimated at $150–160 million annually — to the DAO treasury and, by extension, AAVE token holders. Separately, Sky Protocol (formerly MakerDAO) cut its daily SKY buyback by 87%, from $300,000 to $37,600, to shore up stablecoin reserves. These opposing moves illustrate the tension at the center of DeFi tokenomics: growth investment versus holder distributions.
This report maps the buyback landscape across nine protocols, compares execution models, and identifies where token holder value accrual is genuine versus performative.
Development activity on buyback and treasury infrastructure is accelerating. A GitHub search for "token buyback" repos sorted by recency surfaces several notable projects:
Treasury-Tax-Harvester (updated March 27, 2026) — A Solidity module integrating with Uniswap V3 to convert protocol fees into native tokens automatically. The repo contains a TaxHarvester.sol contract and swap parameter calculator, signaling that buyback infrastructure is being commoditized. Teams no longer need to build custom smart contracts; off-the-shelf modules are emerging.
Autonomous_Buyback by loothero — A Cairo library for autonomous token buybacks via Ekubo's TWAMM (Time-Weighted Average Market Maker) on Starknet, updated January 2026. The use of TWAMM for buyback execution mirrors institutional VWAP strategies, reducing frontrunning risk.
M0 Foundation's TTG Frontend (updated April 9, 2026) — The Two Token Governance system remains under active development. Recent commits include new proposal card UI components and password protection for proposal creation. M0 raised $100 million across Series A and B rounds led by Bain Capital Crypto and Polychain, and its governance model — splitting voting power (POWER token) from veto/guardian rights (ZERO token) — represents a structural departure from single-token DAO governance. Per The Defiant, M0's network supply grew 215% since January and passed $300 million.
GnosisDAO Treasury Tracker (updated daily, most recently April 14, 2026) — An automated repo that commits treasury snapshots daily, providing on-chain transparency into one of the largest DAO treasuries. The consistent update cadence indicates active treasury monitoring by external researchers.
BagsAI Agent Forge (updated April 5, 2026) — A Solana-based platform enabling AI agent tokenization with built-in buyback mechanics. Each deployed agent gets its own token with automatic fee splitting and token buybacks. This represents the convergence of AI agent infrastructure and buyback tokenomics, though the repo has zero stars, suggesting the concept is early-stage.
Buyback programs across DeFi fall into four structural categories, each with different implications for token holders:
1. Revenue-funded buy-and-burn (Hyperliquid, Sky, Uniswap) Protocol fees are used to purchase tokens on the open market and permanently destroy them. Reduces circulating supply. Effectiveness depends on buyback volume relative to new issuance and unlocks.
2. Revenue-funded buy-and-hold (Lido, Treehouse) Tokens are repurchased but retained in the DAO treasury. Does not reduce supply; instead concentrates governance power. Lido's proposed $20 million LDO buyback would acquire approximately 65 million tokens (8% of circulating supply) but all tokens return to the treasury, per CoinDesk.
3. Revenue-funded buy-and-distribute (Aave, GMX, Maple) Tokens are repurchased and distributed to stakers or governance participants. Creates direct cash flow to holders. Aave's expanded program now directs 100% of product revenue to the DAO, which funds weekly AAVE buybacks of $250,000 to $1.75 million, per The Block.
4. Scheduled programmatic buybacks (Spark) Predetermined amounts at predetermined intervals. Spark completed its first month of SPK buyback (26.66 million SPK) and commenced its second with $414,000 USDS on April 7, 2026, per BitcoinWorld. Transparency is high but market impact is muted.
Hyperliquid allocates 97% of trading fees to its Assistance Fund, which buys HYPE on the open market. As of April 2026, the fund holds 42.99 million HYPE purchased for $1.04 billion at an average price of $24.3, according to DL News. The token trades near $37.8, generating substantial unrealized gains.
The model works for Hyperliquid because of three structural factors: (a) the protocol processes roughly $8 billion in daily trading volume, generating consistent fee revenue; (b) the 97% allocation rate is exceptionally aggressive; and (c) the Assistance Fund acts as a permanent bid, creating a price floor effect.
However, 9.92 million HYPE tokens ($375 million) are set to unlock for core contributors in April 2026, per Ainvest. The corporate structure question is paramount: the Assistance Fund benefits all holders, but the unlock disproportionately benefits the founding team. Buybacks funded by user fees effectively provide exit liquidity for insiders.
Messari researcher Sunny Shi argues buybacks represent poor capital allocation, noting that "as protocols gain traction and earn more fees, they end up acquiring tokens at significantly higher prices than initial purchases."
Jupiter's buyback program represents the cautionary tale. The Solana DEX aggregator spent over $70 million — roughly half of protocol fee revenue — on JUP repurchases in 2025, per Crypto.News. The token traded near $0.20–$0.22 by early January 2026, down approximately 89% from its peak.
The structural problem: JUP's circulating supply increased by roughly 150% since launch, with 53 million tokens unlocking monthly through June 2026. The buyback program offset only a fraction of new issuance. Solana co-founder Anatoly Yakovenko publicly addressed the failure, suggesting that storing profits as claimable assets with one-year staking lock-ups would better align incentives, per BeInCrypto.
Jupiter has since cut its planned 2026 airdrop from 700 million to 200 million JUP, acknowledging the dilution problem.
Aave's "Aave Will Win" proposal, passed April 12–13, 2026 with approximately 75% support, restructures the protocol's entire economic model, per CoinDesk. The vote resolved a months-long dispute triggered when swap fees from CoWSwap integration were redirected away from the DAO treasury to an external recipient.
Under the new framework: 100% of gross revenue from all Aave-branded products (Aave Pro, Aave App, Horizon, Aave Kit) flows to the DAO treasury. Protocol revenue reached $140 million in 2025, with an additional $10–20 million from swap activity on Aave.com and Aave Pro. Aave Labs receives $25 million in stablecoins and 75,000 AAVE tokens (~$6.8 million at current prices) over 48 months.
The corporate structure angle: Aave Labs operates as the protocol's development company. The dispute centered on whether Labs or the DAO controlled revenue from user-facing products. The resolution — 100% to the DAO — represents one of the strongest pro-token-holder outcomes in DeFi governance history. Aave's TVL stands at approximately $25 billion across chains.
Treehouse, a DeFi protocol with $294 million in deposits, launched a TREE buyback program committing 50% of all Market Efficiency Yield (MEY) fees from its flagship product tETH to open-market token purchases, per DL News. The proposal passed with 99.59% community support.
Buybacks execute weekly on Ethereum at irregular intervals to mitigate frontrunning. Purchased tokens are held in a DAO-controlled multisig as strategic reserves — buy-and-hold, not burn. The token doubled following the announcement but remains 35% below its August peak of $0.40.
Spark Protocol operates within the Sky (formerly MakerDAO) ecosystem. Its buyback program follows a time-scheduled approach: the protocol completed month one by acquiring 26.66 million SPK, and commenced month two with a $414,000 USDS transfer on April 7, 2026, per Blockster.
The corporate structure is notable: Spark operates under the Spark Foundation, which receives grants from the broader Sky DAO. Buyback funding flows from Sky DAO → Spark Foundation → open market SPK purchases. This multi-entity structure means token holders rely on continued inter-organizational grant approvals.
Maple Finance shifted from staking rewards to a buyback model, allocating 25% of protocol revenue to SYRUP token repurchases, per Crypto.News. Q4 2025 buybacks totaled $615,000; Q1 2026 rose to $827,000 — a 34% quarter-over-quarter increase.
Maple targets $100 million in Annual Recurring Revenue by end of 2026 through its institutional lending business. Once final token issuance completes in late 2026, the 25% buyback allocation is projected to outpace new issuance, creating net deflationary pressure. The protocol is also introducing "Builder Codes" for autonomous partner integrations of syrupUSDC and syrupUSDT, expanding the revenue base.
Parallel to buyback programs, several protocols activated "fee switches" — mechanisms that route protocol revenue directly to token holders through staking or burning rather than accumulating in treasuries.
Uniswap DAO approved the "UNIfication" proposal on December 25, 2025, activating the long-awaited protocol fee switch, per Blockworks. Revenue from v2 and v3 pools on Ethereum mainnet flows into a "TokenJar," where UNI holders can burn tokens to withdraw equivalent value. A retroactive burn of 100 million UNI from the treasury compensates for years of zero value accrual.
Per Coin Metrics, early data implies roughly $26 million in annualized protocol fees against a $5.4 billion market cap — a 207x revenue multiple. The annualized burn rate of approximately 4.4 million UNI equals only 0.4% of supply per year.
A follow-up proposal to expand the fee switch to eight Layer 2 chains (Base, Arbitrum, OP Mainnet, and others) is under vote. If approved, at least one-sixth of LP fees on those chains would flow to the TokenJar, per WEEX. Governance also created a 20 million UNI annual growth budget starting January 1, 2026.
Ethena activated its fee switch in late 2025, directing protocol revenue to sENA (staked ENA) holders, per Blockworks. The activation coincided with an $890 million token buyback program (DAT) launched in late 2025.
However, Q1 2026 saw Gross Protocol Revenue fall 32% quarter-over-quarter to $65.06 million, with TVL dropping approximately $130 million since early March, per FX Leaders. Additionally, 300 million ENA tokens (2% of total supply, ~$27.6 million) unlocked on April 2, 2026. The fee switch is live, but the revenue base is contracting.
Pendle executed a structural overhaul of its value accrual model in January 2026, replacing the vote-escrowed vePENDLE system with liquid sPENDLE staking, per BanklessTimes. Under the new model, up to 80% of protocol revenue is used to buy back PENDLE and distribute it to active sPENDLE holders.
The transition eliminated multi-year lock-ups, replacing them with a participation-based system where holders vote on "critical" Pendle Protocol Proposals to maintain eligibility. A snapshot on January 29 captured remaining vePENDLE lock durations, granting legacy holders a "virtual" sPENDLE boost of up to 4x that decays linearly over two years. This is among the most aggressive revenue-sharing models in DeFi — 80% of fees to stakers versus typical allocations of 25–50%.
The central question: where does the money go?
| Protocol | Mechanism | % Revenue to Holders | Corporate Entity | Holder Benefit Rating | |----------|-----------|---------------------|------------------|----------------------| | Hyperliquid | Buy-and-burn | 97% of fees | Hyper Foundation | High (but unlock dilution) | | Aave | Buy-and-distribute | 100% of product revenue | Aave Labs (funded separately) | High | | Pendle | Buy-and-distribute via sPENDLE | Up to 80% | Pendle team (undisclosed equity) | High | | Uniswap | Burn via TokenJar | Variable (pool-level) | Uniswap Labs / Foundation | Medium | | Maple | Buy from revenue | 25% | Maple Labs Pty Ltd | Medium | | Treehouse | Buy-and-hold | 50% of MEY fees | Treehouse Foundation | Medium | | Sky/Spark | Buy-and-burn / scheduled | Variable (recently cut 87%) | Sky Foundation / Spark Foundation | Medium-Low (reduced) | | Lido | Proposed buy-and-hold | One-time $20M | Lido DAO (no equity entity) | Low (treasury retention) | | Jupiter | Buy from fees | ~50% of fees | Jupiter Labs | Low (overwhelmed by unlocks) | | Ethena | Fee switch + DAT | Variable | Ethena Foundation | Low (declining revenue) | | GMX | Paused; conditional restart | 0% currently (paused at $90 target) | GMX DAO | Suspended |
The protocols with the strongest token holder alignment — Hyperliquid, Aave, and Pendle — share a common trait: they route the majority of revenue to holders through transparent, automated mechanisms. The weakest alignment occurs where buybacks are undersized relative to token unlocks (Jupiter), recently reduced (Sky), or conditional on arbitrary price targets (GMX, which paused staking rewards until the token reaches $90, per GMX Governance).
Critically, even "100% to the DAO" does not mean 100% to token holders. Aave's DAO approved a $25 million payment to Aave Labs alongside the revenue redirect. Sky's foundation structure routes funds through multiple entities before reaching buyback execution. The corporate layer between protocol revenue and token holder wallets remains non-trivial.
The DeFi buyback experiment has produced a clear hierarchy. Protocols that combine high revenue, aggressive allocation rates, and controlled token supply — Hyperliquid (97% of fees, $8B daily volume), Aave (100% of product revenue, $140M+ annually), and Pendle (80% to stakers) — deliver measurable value to token holders. Protocols where buybacks are undersized relative to dilution — Jupiter, Ethena — have spent tens of millions with negligible price impact.
The structural lesson from 2025-2026 is that buybacks are necessary but insufficient. They work only when paired with disciplined supply management. Jupiter's failure and Hyperliquid's success represent the same mechanism producing opposite outcomes, separated by the ratio of buyback volume to token emissions.
Aave's April 2026 vote marks an inflection point in the relationship between protocol development companies and token holder DAOs. The question of who controls revenue — Labs or the DAO — has been answered definitively in Aave's case. Whether other protocols follow that precedent or continue extracting value through opaque corporate structures will define the next phase of DeFi governance.
The buyback is not the innovation. The governance architecture that determines who benefits from the buyback is.