DeFi Week 27: TVL slides to $70.3B

DeFi TVL fell 4.78% to $70.34B during the Week 27 window from June 22 at 10:29 a.m. to June 29 at 10:00 a.m. GMT-5, the sharpest decline in this channel's recent tracking. The damage was concentrated in staking and restaking: SSV Network lost $5.26B of TVL, Lido lost $1.60B, and 134 of 167 tracked $100M+ protocols finished the week lower. 1 2

Exploit losses slowed from last week's $17.3M pace to about $5.78M across four new incidents, but the attack surface shifted toward wallets, frontends, and stale contracts. The largest new hit was Polymarket's $3.1M frontend supply-chain attack, where malicious code reached users through a compromised third-party dependency rather than a smart-contract failure. 3

The aggregate TVL move was not a normal drift lower. DeFi fell from $73.874B to $70.344B, a $3.53B decline, while more than four times as many tracked $100M+ protocols declined as gained. 1 2 For traders, the relevant point is composition: the selloff was led by ETH-linked staking infrastructure, not by a single exploit-driven protocol exit.

SSV Network was the week's outlier. Its TVL dropped from $12.97B to $7.74B, a 40.47% decline and a $5.26B absolute loss. Phemex reported a 37.20% 24-hour drop on June 23, which triggered its volatility threshold for top-20 protocols. 4 SSV Network is Ethereum staking-pool infrastructure, so the move sits in the same risk bucket as the broader liquid staking and restaking drawdown.

That bucket was weak almost across the board. Lido remained the largest DeFi protocol at $14.33B, but it lost $1.60B, or 10.07%, during the week. EigenCloud fell 11.38%, ether.fi Stake fell 11.31%, Rocket Pool fell 11.31%, Binance staked ETH fell 10.77%, and Kelp fell 12.97%. 2 The common reading is straightforward: liquid staking and restaking collateral was the pressure zone, and yield farmers using those assets as collateral should treat LTV and liquidation buffers as live variables rather than passive settings.

Chain data showed the same split. Ethereum held $37.43B, or 53.1% of DeFi TVL, but lost $2.00B and 5.07% week over week. BSC lost 6.69% to $4.85B, Tron lost 5.02% to $4.43B, and Base lost 3.62% to $4.11B. 6 Solana was the exception: it rose 0.59% to $4.99B, small in percentage terms but important because it was the only major chain with a W26 baseline to finish positive. 6

The positive protocol exception was Maple Finance. Maple TVL rose from about $2.095B to $2.426B, a 15.81% gain and the strongest top-30 performance. BrokerChooser tied the move to Kraken's partnership with Maple to deliver on-chain asset-backed financing, and it reported that SYRUP rose roughly 20% on the announcement. 5 This was not a broad lending rebound. Aave V3 still fell 6.34% to $11.85B, even as Aave remained one of the largest venues in DeFi. 2

A top-30 snapshot confirms the RWA and risk-curation pockets held up better than staking. Circle USYC rose 1.13%, BlackRock BUIDL rose 0.90%, Steakhouse Financial rose 1.62%, and Figure Markets Exchange rose 2.42% while the main staking and DEX names were red. 2 That dispersion matters: this was a week to separate collateral quality and fee source, not to apply one risk setting to every DeFi position.

Week 27 produced roughly $5.78M in new exploit losses, led by Polymarket, SecondFi, Royal.io, and Lixir Finance. 12 The total is lower than Week 26, but the attack vectors are uncomfortable for active users. Smart contracts were not the only failure point; web frontends, wallet-generation software, and legacy reward contracts were also in scope.

Polymarket lost about $3.1M on June 25 after a compromised third-party frontend dependency injected malicious code into the site for some users. Protos reported that 11 user wallets were drained of pUSD, the funds were swapped into roughly 1,893 ETH, and the ETH was bridged from Polygon to Ethereum mainnet. 3 Polymarket said the affected smart contracts were not exploited and pledged full reimbursement to affected pUSD holders, but the operational lesson is harsher than a contract bug: hardware wallets and contract audits do not protect users who sign a malicious frontend transaction. 13

SecondFi, the Cardano light wallet formerly known as Yoroi, suffered a predictable-private-key exploit in its wallet generator software on June 23. TheNewsCrypto reported 178 compromised wallets, about $2.4M in confirmed losses, and SlowMist's estimate of up to 129M ADA exposed, with potential losses above $20M. 7 SecondFi suspended services and told users to move funds to new wallets, but the summary found no detailed audit report or reimbursement plan during the window. 7

Royal.io was smaller but technically revealing. A legacy Polygon royalties contract lost about $263K after an attacker used 100 zero-value transfers and a flash loan to inflate reward accounting in Royal1155LD.beforeLdaTransfer(). CertiK described the result as a 100X reward claim, and TenArmorAlert detected the attack on June 23. 14 The exploit belongs in the same mental bucket as other dormant-contract incidents: old contracts with live balances are still attack targets.

Lixir Finance lost $12,300 on Ethereum through a broken signature-verification vulnerability, according to the DeFiLlama hacks database. 12 The amount is small enough that it should not dominate risk decisions, but it reinforces the same point as Royal.io: small protocols and old deployments can still create approval and wallet hygiene risk.

The follow-up list matters too. Taiko Bridge moved from last week's halt into a four-stage restart plan: fixes and security-committee verification, bridge replenishment to 1:1 backing, L2 transfers and swaps, then full bridge reactivation with conservative withdrawal limits. Xangle reported that the attack vector was blocked and the security fixes received independent expert review. 15 Secret Network's Axelar bridge remained unresolved in the tracked window: Secret and Secret-SNIP IBC connections stayed disabled, no user-compensation plan appeared, and about $642K remained in the attacker's Axelar wallet after Axelar declined a freeze request. 16

Quarterly context is ugly. BingX, citing DeFiLlama data, reported 83 crypto hacks and $755.3M in Q2 losses, with KelpDAO and Drift Protocol accounting for more than 75% of the quarterly total and bridges accounting for about $351M. 17 Week 27 was not the biggest dollar week, but it was a useful reminder that user-layer attacks can scale without touching audited protocol logic.

Arbitrum Foundation's 2027 budget vote passed and executed on-chain after carrying over from Week 26. The final result was 206.39M ARB For, 8.67M ARB Against, and 24.97M ARB Abstain, equal to 86.0% For, 3.6% Against, and 10.4% Abstain across 1,241 addresses. 8 The proposal funds the Foundation with $16M in RWA, 1,700 ETH, and 230M ARB for 2027 operations. 8

Arbitrum immediately opened another treasury decision. The AGV Wind-Down proposal would stop Arbitrum Gaming Ventures from making new investments, return 143.7M ARB to the DAO treasury, and reduce AGV to managing 15 existing investments with $10M already deployed. 18 For ARB holders, the budget vote settled one fight and the AGV proposal opened another: the DAO is still deciding how much capital should sit in semi-autonomous growth vehicles.

Aave DAO continued to harden risk governance after the KelpDAO exploit. Three ARFCs closed and passed during the window: stcUSD onboarding to Aave V3 MegaETH passed with 420,995 AAVE For and 34 AAVE Against; Risk Stewards cooldown reduction passed with 387,360 AAVE For; and the Umbrella parameter update passed with 387,358 AAVE For and 0.4 AAVE Against. 19 A fourth vote, the PT Risk Oracle migration to protocol-owned infrastructure on Chainlink Runtime Environment, remained active at 336,319 AAVE For. 19

Lido DAO closed six carryover votes on June 22, including LIP-33, LIP-35, Nemo's appointment as a Lido Labs director, Simple DVT regular-cluster wind-down, selected wstETH bridge endpoint revocations, and a Galaxy node-operator infrastructure update. Each passed with roughly 55M-57M LDO support and near-unanimous approval. 20 The scope is more important than the vote margin: Lido changed staking architecture, bridge status, and operator infrastructure in one governance batch.

ENS DAO showed the clearest governance stress. The Endowment Investment Policy update passed with 1.01M ENS For and 183K ENS Abstain, while the Security Council renewal passed with only 26.9% For and 71.3% Abstain. 9 Crypto Briefing also reported a separate temperature check from ENS COO Katherine Wu that would delegate day-to-day management of more than $350M in treasury assets to an expanded ENS Foundation board; Brantly Millegan opposed the move, while ENS founder Nick Johnson planned to delegate tokens in support. 21

GnosisDAO passed GIP-151, a one-time pro-rata treasury redemption proposal, with 157,749 GNO For, 2,500 GNO Against, and 1,492 GNO Abstain, reaching 215% of the required quorum. 22 That makes treasury redemption, not protocol growth, the live governance question for GNO holders.

The DeFiLlama yield scan found 64 pools with more than 20 percentage points of 7-day APY movement and more than $1M in TVL. Solana accounted for 20 of those pools, Base for 17, and Ethereum for 13. 10 The useful distinction is incentive versus fee income: the summary's analysis classified 94.9% of the regime as base-rate driven, up from 58% in Week 26. 10

Base still had the most ridiculous headline number. Aerodrome's USDC-VELVET pool showed 653,013.40% APY at $9.45M TVL, but almost all of it came from VELVET token rewards rather than trading fees. 10 That is farmable only as an emissions and exit-liquidity problem. It should not be modeled like organic fee yield.

Solana's DEX pools looked different. Raydium AMM WSOL-ACT showed 300.20% APY, gmtrade SOL-USDC showed 211.80%, and Orca SOL-USDC showed 130.76% at $24.3M TVL, with the summary classifying all three as purely fee-driven and zero-incentive. 10 That does not make them low-risk. It does mean the yield source was trading activity rather than a token-reward schedule that can vanish at epoch end.

The large-pool downside was quieter. The scan found no $10M+ TVL pools with APY collapses greater than 30 percentage points over seven days. 10 TON's TSTON-USD₮ pool cooled from last week's 164% area to 104.23% APY at $7.7M TVL, while Curve's IDAI-IUSDC-IUSDT pool stabilized at 123.01% APY near its 30-day mean. 10

The unstable-asset watch list got worse. apxUSD fell to $0.794, a 20.6% discount to the $1 peg, after closing Week 26 at $0.906. 11 Apyx V2.0 launched on June 16, but the peg had not stabilized after 13 days in the tracked data. 11 Any market using apxUSD as collateral deserves a fresh liquidation-path check.

Two institutional and base-layer items also affect yield positioning. BlackRock will integrate Ethena's USDe into Aladdin and create a $100M liquidity facility for BUIDL with Ethena; CoinDesk reported ENA rose 8% on the announcement. 23 On Solana, Anza CEO Brennan Watt said SIMD-123, SIMD-550, and SIMD-553 are all planned for 2026; SIMD-550 would double the disinflation rate from 15% to 30% annually, and SIMD-553 would introduce resource-based fees with 100% burn. 24 For SOL stakers, the direction is lower emissions and more fee sensitivity, though the governance votes for SIMD-550 and SIMD-553 were not yet scheduled in the summary. 24

For yield farmers, Week 27 argues for tighter collateral review on liquid staking and restaking positions. SSV Network's 40.47% TVL fall and Lido's 10.07% decline do not automatically mean protocol failure, but they do change liquidation and exit-liquidity assumptions for positions built on staking derivatives. 2

For active traders, frontend and wallet hygiene need the same attention as contract risk. Polymarket's contracts were not the failure point, and SecondFi's issue was key generation rather than a DeFi lending pool. 3 7 Separate signing wallets, revoke stale approvals, and avoid signing transactions from newly changed frontends until the protocol has explained the dependency chain.

For APY hunters, the distinction between emissions and fees was the week’s most useful filter. Aerodrome's VELVET pool was an incentive trade; Solana's Raydium, gmtrade, and Orca pools were fee trades in the source data. 10 Those are different risk models, even if the headline APYs sit in the same table.

For governance watchers, Arbitrum, ENS, Gnosis, and Aave all moved money or risk authority. The practical checklist is simple: ARB holders should follow the AGV wind-down, ENS holders should follow the treasury-delegation temperature check, GNO holders should evaluate redemption mechanics, and Aave users should monitor the PT oracle migration before adding structured collateral exposure. 18 21 22 19