DeFi Week 35: TVL jumps 25.1% as Gnosis sets an L1-to-rollup path (2026)

DeFi's aggregate TVL rose from $74.730B to $93.461B across the latest weekly snapshots, a gain of $18.732B or 25.1%. The increase was concentrated late in the window: the daily series moved from $82.027B on August 20 to $91.802B on August 23. 1

The headline needs a flow-versus-valuation check. Lido, SSV Network, and Binance staked ETH all posted large increases in the protocol feed, yet ETH-linked TVL can rise when the underlying asset is marked higher. Uniswap V4's 732.7% jump is large enough to keep coverage or methodology changes in the live explanation set. The feed supplies the before-and-after values; it leaves the primary causes of the largest moves unresolved. 2

The cutoff is 10:00 UTC-05:00 on August 24, 2026. The August 17 and August 24 total-TVL readings are the nearest daily UTC snapshots around the cutoff, rather than exact hour-level readings. The protocol, chain, hack, and yield feeds were read for the August 17-24 window.

Quick scan

Question	Verified reading	Why it matters
Total DeFi TVL	$74.730B to $93.461B; +$18.732B / +25.1%	The aggregate balance rose sharply, with most of the move arriving after August 20. 1
Largest dollar protocol gainer	Lido: +$6.012B	The largest move sits in an ETH-linked staking record, so deposit flow and asset valuation need separate checks. 2
Largest percentage protocol gainer	Uniswap V4: +732.7%	The size of the jump leaves a coverage or methodology change as a live possibility. 2
In-window security records	4 DeFiLlama rows totaling $10.3981M on the feed's amounts	Term Finance accounts for most of the listed amount; Allbridge adds a confirmed flash-loan-assisted bridge case. 3
Verified governance outcome	GnosisDAO GIP-153 passed with 123,158 for, 115 against, 151 abstaining	The vote gives Gnosis a design mandate for an L1-to-rollup transition; migration execution remains ahead. 4
Largest displayed yield	Aerodrome USDC-PROS: 53,600.92067% APY	The feed labels the entire displayed rate as rewards; liquidity, emissions, and exit conditions need verification. 5

The TVL path

The daily series shows a steady climb, followed by a faster four-day expansion. The values below are the exact figures used for the week-over-week calculation; the dates are daily UTC snapshots.

Loading chart…

The daily changes were +2.0037% on August 18, +0.6821% on August 19, +7.6084% on August 20, +2.9015% on August 21, +4.1622% on August 22, +4.4153% on August 23, and +1.8076% on August 24. 1

That timing matters for interpretation. A rising aggregate can reflect new deposits, token appreciation, changes in what a protocol counts, or several of those effects together. The weekly total is therefore a starting point for the screen, rather than a direct measure of fresh risk capital.

Where the balance sits by chain

The current chain endpoint totals $93.651B. Ethereum holds $49.368B, or 52.7150% of that endpoint total. BSC, Solana, Base, and Tron are the next largest balances. The endpoint supplies current levels and no comparable weekly change field, so the table describes current composition rather than weekly chain flows. 6

Chain	Current TVL	Share of endpoint total
Ethereum	$49.368B	52.7150% 6
BSC	$10.980B	11.7246% 6
Solana	$5.580B	5.9579% 6
Base	$5.531B	5.9061% 6
Tron	$5.230B	5.5843% 6
Bitcoin	$4.123B	4.4021% 6
Hyperliquid L1	$1.543B	1.6480% 6
Arbitrum	$1.441B	1.5389% 6
Monad	$0.937B	1.0007% 6

Ethereum's 52.7150% share identifies the location of the balance. It does not identify the source of the week's $18.732B increase. A flow attribution requires a comparable historical series at the chain level.

Protocol ranking: the large moves

The ranking screen uses protocols with at least $50M in TVL, excludes CEX records, and compares each current value with a back-calculated prior value. For each row, prior TVL equals current TVL divided by 1 + change_7d / 100. Displayed figures are rounded to the nearest $0.1M or $0.0001B; the dollar moves use the underlying feed values. 2

Largest dollar gainers

Protocol	TVL before	Current TVL	7-day change	Dollar move	Primary reason
Lido	$18.0924B	$24.1046B	+33.2306%	+$6.0122B	Unresolved in the TVL feed. ETH-linked valuation is a live factor to check. 2
Uniswap V4	$0.7549B	$6.2859B	+732.6668%	+$5.5310B	Unresolved; the scale keeps coverage or methodology change in the investigation set. 2
Aave V3	$14.2377B	$17.3667B	+21.9768%	+$3.1290B	Unresolved in the TVL feed. 2
SSV Network	$9.5943B	$12.7064B	+32.4376%	+$3.1121B	Unresolved in the TVL feed; the record is ETH-linked. 2
Binance staked ETH	$7.0679B	$9.3528B	+32.3274%	+$2.2849B	Unresolved in the TVL feed; the record is ETH-linked. 2

The largest gainer rows should therefore be read as reported TVL changes, not as a clean deposit leaderboard. Lido's dollar increase is the largest in the screen, while Uniswap V4's percentage increase is the most exposed to a possible measurement break. No first-party causal announcement was verified for these leading moves during the window.

Largest dollar losers

Protocol	TVL before	Current TVL	7-day change	Dollar move	Primary reason
Steakhouse Financial	$3.1808B	$2.9445B	-7.4286%	-$236.3M	Unresolved in the TVL feed. 2
BlackRock BUIDL	$3.5464B	$3.3939B	-4.2999%	-$152.5M	Unresolved in the TVL feed. 2
Invesco USTB	$0.7814B	$0.6434B	-17.6649%	-$138.0M	Unresolved in the TVL feed. 2
VanEck Treasury Fund	$0.1933B	$0.0578B	-70.0754%	-$135.4M	Unresolved in the TVL feed. 2
Solstice	$0.5058B	$0.4041B	-20.1128%	-$101.7M	Unresolved in the TVL feed. 2

The loser screen contains several tokenized Treasury records. A reported TVL decline in those rows can reflect withdrawals, redemptions, asset-price effects, or accounting changes; the protocol endpoint alone leaves those alternatives open.

Security events

DeFiLlama returned four hack records with event dates inside the window. Their feed amounts sum to $10.3981M. The table keeps the feed's classifications and adds verified incident detail where a retrieved report supplied it. 3

Protocol	Chain	Amount	Attack vector	Current status
TermFinance Vaults	Ethereum	$8.5M on DeFiLlama; approximately 2,843 ETH and 1.68M USDC, or about $6.9M, in the incident report	Governance exploit: an attacker acquired controlling voting power in a thinly distributed governance token, then passed proposals that redirected vault funds. 7	Term Labs shut down all Term Meta Vaults, revoked DAO governance roles to block new deposits, left withdrawals open, and coordinated remediation with external security teams. Confirmed recovery: unavailable. 7
Maya Protocol	Mayachain	$1.7M	Protocol logic; withdrawal logic flaw; bridge hack classification in the DeFiLlama record	Remediation and recovery status: unavailable. 3
Allbridge	Base	$191,000 on DeFiLlama; approximately $190,000 net loss in the SlowMist account	Flash-loan-assisted bridge exploit: the attacker forged a CCTP-style message, then used an approximately $809,000 Aave flash loan after the Base Router held about $191,000 USDC. Allbridge accepted the forged message without sufficient sender, recipient, minting, or balance-change checks. 8	Patch and recovery status: unavailable in the retrieved incident account. The source difference is $190,000 in the SlowMist report versus $191,000 in DeFiLlama. 3
Arrakis V1	Ethereum	$7,100	Oracle manipulation; spot-price manipulation	Remediation and recovery status: unavailable. 3

Term Finance is the largest listed incident by the feed's amount, and its governance failure is a separate risk category from a contract-level flash-loan bug. Allbridge supplies the week's clearest flash-loan-linked attack path. Maya Protocol and Arrakis V1 remain limited to the fields returned by the incident feed.

Governance

GnosisDAO approved GIP-153 during the window. The official Gnosis announcement records 123,158 votes for, 115 against, and 151 abstaining, from 54 voters. Total turnout was 123,425, against a 75,000 GNO quorum. 4

Proposal	Result	Vote and quorum	Execution impact
GIP-153: move from a standalone L1 toward a ZK-proven Ethereum Economic Zone rollup	Passed	123,158 for; 115 against; 151 abstaining; 54 voters; 123,425 total turnout; 75,000 GNO quorum. 4	Engineering received a design mandate. Gnosis described synchronous composability with Ethereum mainnet and targeted genesis around the turn of the year, subject to technical milestones. The vote itself created the mandate; it did not immediately execute the migration. 4

Daily Crypto Briefs attributed a future implementation consequence to the proposal: the existing validator set would eventually be sunset, and approximately 350,000 staked GNO, described as about 27% of circulating supply, could be unlocked in the future. Those effects remain future implementation context rather than an immediate token distribution. 9

GIP-153 is the only in-window governance outcome in this edition with verified result, quorum, and execution-impact details. The vote changes Gnosis's engineering direction; readers still need the later technical milestones and any implementation transaction before treating the transition as an executed migration.

Yield anomalies

The yield feed contains several extreme Base and Hyperliquid L1 readings. APY, base APY, reward APY, the seven-day APY change, and the 30-day mean below are feed fields. A missing field is shown as unavailable. 5

Pool	Chain	TVL	APY	Base APY	Reward APY	7-day APY change	30-day mean	Reading
Aerodrome Slipstream USDC-PROS	Base	$723,892	53,600.92067%	unavailable	53,600.92067%	+27,438.55431 pts	71,021.67191%	Entirely reward-labelled in the feed. 5
Aerodrome Slipstream USDC-VELVET	Base	$4,396,209	47,060.44941%	1.56564%	47,058.88377%	+7,266.08059 pts	53,266.53214%	99.9967% reward-labelled in the feed. 5
Aerodrome Slipstream CHECK-SAND	Base	$1,435,099,528	21,096.55832%	21,096.55722%	0.0011%	unavailable	2,838.83182%	Almost entirely base-labelled despite unusually large TVL; validate the rate before treating it as an opportunity. 5
Aerodrome Slipstream WETH-CBBTC	Base	$706,751	3,074.85346%	unavailable	3,074.85346%	+1,685.80209 pts	3,082.67824%	Reward-labelled in the feed, with WETH-CBBTC asset exposure. 5
Growihf USDC vault	Hyperliquid L1	$9,406,074	1,370.48053%	unavailable	unavailable	+1,370.48053%	179.97336%	The current APY is several times the 30-day mean; the feed does not identify the base/reward split. 5

These APYs are investigation triggers, not promised returns. Before sizing a position, verify reward-token liquidity, emission schedules, pool depth, token price exposure, and exit slippage. CHECK-SAND deserves a separate validation pass because its displayed rate is almost entirely base-labelled while its TVL is $1.435B. USDC-PROS, USDC-VELVET, and WETH-CBBTC require the opposite check: the feed attributes nearly all or all of their displayed rates to rewards.

Position screen

The week's data leaves four practical checks for an enter, exit, or hold decision:

Separate balance from flow. Treat the $25.1% aggregate TVL increase and the Lido, SSV Network, and Binance staked ETH gains as reported balances first. Check token prices, deposit records, and protocol accounting before reading the moves as fresh capital.
Investigate the measurement break. Uniswap V4's +732.6668% change is the largest percentage move in the screen. Confirm pool coverage and methodology continuity before comparing it with ordinary week-over-week inflows.
Price governance risk alongside contract risk. Term Finance shows how concentrated voting power can redirect vault assets. Allbridge shows how a forged bridge message and a flash loan can combine. Gnosis's passed proposal changes a future engineering path, while its execution remains ahead.
Demand an exit path from extreme APY. Reward-labelled rates depend on emissions and liquid reward markets. Base-labelled rates still require checks on the underlying strategy, pool depth, and pricing mechanism. A quoted APY without those checks is an incomplete risk screen.

The week delivered a large rise in reported DeFi TVL, a governance decision that points Gnosis toward an Ethereum Economic Zone rollup, four recorded security incidents, and several rates that need verification before they can support a position. The largest protocol causes and three incident remediation statuses remain unresolved in the available evidence.