DeFi protocols deployed over $1.4 billion in token buybacks in 2025, and the pace is accelerating in 2026. Hyperliquid leads with $644 million in cumulative repurchases and a near-$1 billion annualized run rate. Sky Protocol (formerly MakerDAO) has spent $114.5 million buying back 1.83 billion SK...
"The current DAO structure isn't working as intended. We hear the complaints. We see the breakdown in trust." — Kash Dhanda, Jupiter Team
DeFi protocols deployed over $1.4 billion in token buybacks in 2025, and the pace is accelerating in 2026. Hyperliquid leads with $644 million in cumulative repurchases and a near-$1 billion annualized run rate. Sky Protocol (formerly MakerDAO) has spent $114.5 million buying back 1.83 billion SKY tokens. Optimism launched a 12-month pilot committing 50% of sequencer revenue to OP buybacks. Pyth Network deploys 33% of its DAO treasury monthly. Aave made its $50 million annual buyback permanent, though it may reduce to $30 million in 2026.
The shift from revenue sharing to buyback-and-burn is structural, not cyclical. Four Pillars Research published a widely-cited thesis: direct revenue sharing gives insiders easy liquidity to exit without market impact, while buyback-and-burn raises token value proportionally for all holders. But the data is mixed — per CoinGecko, 10 projects accounted for 92% of all buyback spending in 2025, and many tokens with active buyback programs remain down 40-90% from highs. Heavy unlock schedules and ongoing emissions frequently overwhelm repurchase demand.
This report maps the buyback landscape across eight protocols, examines who actually benefits from these programs, and identifies the corporate structures that determine whether value flows to token holders, shareholders, or neither.
Development activity provides a ground-truth signal beneath the governance announcements. Three patterns stand out this week:
Pendle Finance shows sustained, substantive GitHub activity. The pendle-finance/pendle-core-v2-public repo pushed code as recently as March 4, including a commit to "Move vePendle related addresses to deprecated," confirming the on-chain transition from vePENDLE to sPENDLE. The boros-core-public repo pushed v1.0.6 on March 2 with OTCModule additions to the Router — indicating active expansion of the Boros interest-rate swap product. An sPENDLE security audit report (WatchPug) was added February 10. This is real engineering, not announcement-driven governance theater.
Autonomous Buyback Infrastructure is emerging on Starknet. The loothero/autonomous_buyback repo — a Cairo library for autonomous token buybacks via Ekubo TWAMM — pushed a v2 buyback component with enhanced features on January 23, 2026, including ERC20 burn support. While small (zero stars), it signals that buyback infrastructure is being productized and open-sourced at the smart contract level, not just executed ad hoc by foundations.
M0 Foundation's Two Token Governance (TTG) continues to iterate on a dual-token governance mechanism for maintaining lists and managing communal property — 11 stars, 1 fork, last updated December 2025. The frontend repo was updated as recently as February 15, 2026. This governance-minimized design offers an alternative to the delegate-heavy models now fracturing at Aave.
Hyperliquid maintains an active Python SDK (hyperliquid-dex/hyperliquid-python-sdk) with v0.22.0 released February 4, 2026. The broader Hyperliquid ecosystem on GitHub is dominated by third-party trading bots (OctoBot, Passivbot, HammerGPT), suggesting substantial programmatic trading volume — which directly feeds the protocol's buyback engine.
The following table represents the most significant token buyback programs active in Q1 2026, ranked by cumulative spend:
| Protocol | Cumulative Spend | Tokens Repurchased | % of Supply | Revenue Source | Destination | |---|---|---|---|---|---| | Hyperliquid | ~$1B+ | 40.5M HYPE | ~13%+ | 97% of trading fees | Burned | | Sky Protocol | $114.5M | 1.83B SKY | ~2% | Treasury surplus | Burned/reserves | | Jupiter | $70M+ | 95M JUP | 1.37% | 50% protocol revenue | Burned | | Aave | ~$50M+ | 205K AAVE | 1.28% | Protocol revenue | Stakers | | Ether.fi | $13.18M | ~987K ETHFI | <1% | Withdrawal fees + protocol rev | sETHFI holders | | Metaplex | ~$13.7M (H1 '25) | 73.8M MPLX | 7.4% | 50% protocol fees | DAO treasury | | Pyth | ~$500K/mo | TBD | TBD | 33% DAO treasury | Locked reserve | | Optimism | ~$7.5M/yr est. | TBD | TBD | 50% sequencer rev | Collective treasury |
Per CoinGecko's 2025 report, 28 notable projects executed buybacks in 2025, but the top 10 accounted for 92% of total spend. Hyperliquid alone represented 46% of all token buyback volume, according to CryptoPatato.
The headline "protocol does buybacks" obscures critical design differences that determine actual value flow to token holders.
Burn Model (Hyperliquid, Jupiter, Sky): Repurchased tokens are permanently destroyed. This mechanically reduces total supply, increasing each remaining holder's proportional claim. Hyperliquid's Assistance Fund burned 37.5 million tokens ($912M) in a single governance vote, per The Defiant. Sky burns at a rate of ~3.6 million SKY per day, funded by $250K/day in protocol revenue, according to Sky's dashboard. The Sky Frontier Foundation projects $611.5 million in gross protocol revenue for 2026, an 81% YoY increase, per CoinDesk.
Redistribute Model (Aave, Ether.fi): Repurchased tokens flow to stakers. Aave's buyback-derived AAVE goes to stkAAVE holders via the new "Anti-GHO" mechanism, per Mint Ventures. Ether.fi sends all buyback-derived ETHFI to sETHFI holders, per its governance documentation. This model rewards active participants but does not reduce supply — tokens re-enter circulation when stakers unstake and sell.
Treasury Accumulation Model (Metaplex, Pyth, Optimism): Repurchased tokens go to a DAO treasury or reserve. Metaplex's DAO now holds 231 million MPLX (the largest holder) after buying back 73.8 million tokens, per Metaplex's H1 2025 recap. Pyth's reserve requires a 67% supermajority vote to unlock, per The Block. Optimism's repurchased OP enters a "collective treasury" where governance decides its fate, per The Block. This model centralizes value in the DAO itself — beneficial if governance is functional, risky if captured.
The Effectiveness Question: Per The Block, Keyrock researcher Amir Hajian noted that heavy unlock schedules and ongoing emissions can easily outweigh buyback demand. Boris Revsin of Tribe Capital observed that many programs "bought tokens when prices and revenues were already high rather than during downturns." Helium paused its buyback after seeing no market impact. Jupiter is "rethinking its approach" after spending $70M with JUP still trading well below highs, per CoinGecko data.
The largest DeFi protocol by TVL ($36.4 billion) is experiencing a governance crisis that directly intersects with its buyback program.
On March 3, 2026, the Aave Chan Initiative (ACI) announced it would shut down operations after a dispute with Aave Labs over the "Aave Will Win" proposal — a $51 million budget request for V4 development, per CoinDesk. ACI founder Marc Zeller stated that addresses linked to Aave Labs voted on their own funding proposal, tipping the outcome. The Temp Check passed 52.58% in favor, 42% against, 5.42% abstaining.
ACI's departure is significant: the group drove 61% of governance actions over three years, helped grow GHO from $35M to $527M supply, and cost the DAO $4.6M total, per The Defiant. BGD Labs, the team that built Aave V3, previously announced its own departure. AAVE fell 11% in 24 hours following ACI's exit, per DL News.
The corporate structure angle matters here. Aave Labs is a private company (Avara) that builds for the DAO but is not governed by it. When Avara-linked addresses vote on Avara's own budget requests, the conflict of interest is structural. The DAO's $50M annual buyback program (now potentially reduced to $30M, per Aave governance forums) effectively uses token holder capital to support a token that Avara also holds. Whether this represents value alignment or value extraction depends on governance oversight — which just lost its two most active contributors.
Pendle: From vePENDLE to sPENDLE
Pendle completed its transition from vePENDLE (vote-escrowed, long lock-ups) to sPENDLE (liquid staking) in January 2026. The rationale, per Startup News: long lock-ups, complexity, and lack of interoperability had become "significant barriers." Despite generating over $37M in 2025, rewards concentrated among a small fraction of sophisticated users.
Under sPENDLE: stakers can withdraw after a 14-day unwinding period (or instantly for 5% fee); Pendle conducts token buybacks using up to 80% of protocol revenue; manual governance voting is replaced by algorithmic emission allocation based on KPIs, cutting total emissions by ~30%. Pendle's TVL stands at ~$3.5 billion, per DeFi Llama. GitHub activity confirms this is active — the vePENDLE deprecation commit and sPENDLE security audit are both visible in the pendle-core-v2-public repo.
This is notable: Pendle is the first major protocol to explicitly abandon the ve-model in favor of liquid staking with revenue-funded buybacks. The shift acknowledges that locking mechanisms, while good for supply reduction, create governance plutocracies.
Pyth Network: Treasury-Funded Programmatic Buying
Pyth launched its buyback program in December 2025, deploying 33% of DAO treasury each month. Per Pyth's February 2026 report, the Pythian Council executes purchases. The fixed 33% allocation removes ongoing governance decisions. Revenue comes from four products: Pyth Pro ($1M ARR in first month, targeting $50M ARR), Pyth Core, Entropy, and Express Relay.
The structural design is notable: buyback size scales automatically with treasury growth, purchased tokens require a 67% supermajority to unlock, and execution is delegated to a council rather than ad hoc proposals. This is governance-minimized capital allocation.
Metaplex: Protocol-Fee-to-DAO Pipeline
Metaplex earns fees on nearly every token and NFT created on Solana. 50% of these fees fund monthly MPLX buybacks for the DAO treasury. Through H1 2025, the protocol generated $13.7M in fees and repurchased 38.5 million tokens (3.9% of supply), with no future token unlocks remaining, per Metaplex's blog. Cumulative buybacks have reached 73.8 million MPLX (7.4% of total supply).
Key distinction: Metaplex has zero remaining token unlocks. Combined with revenue-funded buybacks, this creates a structurally deflationary token — unlike protocols where buybacks merely offset unlock dilution.
The central question: who captures the economic surplus these protocols generate?
Token Holders Win: Hyperliquid's model is the clearest case. 97% of fees go to buyback-and-burn. No foundation, no equity investors, no token unlocks (beyond the team's 1.2M/month vesting started January 2026). The Hyper Foundation proposed burning the entire Assistance Fund permanently — $920M in tokens. If passed, this is among the largest voluntary destructions of crypto protocol value in history.
Stakers Win (Maybe): Aave and Ether.fi distribute buyback-derived tokens to stakers. This rewards active governance participants but creates a two-tier system: passive holders subsidize staker rewards. The value proposition is clear only if staking yields exceed the opportunity cost and the token itself holds value.
DAOs Win (Ambiguously): Metaplex, Pyth, and Optimism accumulate tokens in DAO treasuries. Token holders benefit only if governance deploys this capital effectively. Metaplex's DAO is now the largest MPLX holder — a form of protocol self-ownership. Whether this resembles a well-managed corporate treasury or an idle capital pile depends entirely on governance quality.
Corporate Entities Win (Quietly): Aave Labs (Avara) requested $51M from the DAO while holding AAVE tokens and voting on its own proposal. The Ethena Foundation controls fee switch parameters for ENA, which has generated $666M in cumulative fees but has not yet finalized what percentage flows to stakers, per Blockworks. Sky Frontier Foundation projects revenue but executes buybacks through DAO governance that it heavily influences.
Nobody Wins (Yet): Optimism launched buybacks in February 2026, but OP has fallen 42.6% in the past month, per CoinMarketCap data. Coinbase's Base departed the Superchain, removing a key revenue source. The 50% sequencer revenue commitment amounts to an estimated $7.5M/year — insufficient to offset the narrative and financial damage of Base's exit. Ethena's ENA fee switch remains in "governance discussion" since September 2025, while $171.8M in tokens unlocked on March 5, per FX Leaders.
The DeFi buyback wave of 2025-2026 represents a genuine maturation of token economics — protocols are generating real revenue and returning it to token holders through mechanisms borrowed from traditional corporate finance. Hyperliquid's $1B+ annualized buyback run rate, funded by 97% of trading fees with permanent burns, is the current gold standard.
But the aggregate data tells a more cautionary story. Ten projects control 92% of buyback volume. Prices for most tokens with active buyback programs continue to decline. And the governance infrastructure required to oversee these programs is fracturing — Aave's loss of both ACI and BGD Labs, its two most critical contributors, is a stress test that other DAOs will eventually face.
The protocols best positioned are those with: (1) real, diversified revenue streams; (2) completed or minimal token unlock schedules; (3) permanent burn mechanisms rather than redistribution; and (4) governance structures that constrain foundation/labs self-dealing. Hyperliquid and Metaplex currently meet all four criteria. Most others meet one or two.
The thesis is straightforward: buybacks work when they reduce net supply against a backdrop of growing demand. When emissions outpace buybacks, or when the corporate entity behind the protocol captures more value than it returns, the mechanism fails regardless of headline spend.