DeFi Week 32: TVL falls 4.1% while vault curation gains $577M
Week 32 DeFi TVL fell $3.14B to $73.73B; Steakhouse Financial gained $577M while staking-heavy protocols lost, and the risk screen found $8.79M in quantified hacks plus reward-driven APY extremes.
DeFi TVL fell from $76.870B on the July 27 daily snapshot to $73.732B on August 3, a loss of $3.139B (-4.08%). The largest dollar gainer was Steakhouse Financial, up $577.2M, while Lido lost $827.1M and Babylon Protocol lost $729.3M. The risk screen also found $8.788M in three quantified hack records, a RARI DAO decision to return Rarible to Solana, and yield prints whose headline APYs were almost entirely reward-driven. 12345
Data cutoff: August 3, 2026, 10:00 a.m. GMT-5. DeFiLlama's historical series is daily and timestamped at 00:00 UTC, so the total compares the daily points dated July 27 and August 3 rather than an hour-level balance. The protocol ranking uses the current feed's
change_7d field; dollar changes are back-calculated and do not distinguish deposits, withdrawals, token-price effects, or accounting changes. 12Week 32 quick scan
| Signal | Result | Trading read |
|---|---|---|
| Total DeFi TVL | $73.732B, down 4.08% from $76.870B | The weekly aggregate gave back more than $3B; protocol-level flows were uneven. 1 |
| Largest protocol dollar gain | Steakhouse Financial: +$577.2M, +23.45%, to $3.038B | A vault-curation inflow deserves a look at the underlying markets, not just the curator label. 2 |
| Largest protocol dollar loss | Lido: -$827.1M, -4.53%, to $17.415B | The feed does not show whether the decline was price-led, withdrawal-led, or both. 2 |
| Quantified security losses | $8.788M across Crypto DAO, LULA, and Set Protocol | Two of the three records were on BSC; no recovery status was returned for LULA or Set Protocol. 3 |
| Governance | RARI DAO backed Rarible's return to Solana | Vote margin and quorum were not disclosed; development, integrations, and audits started with a four-week target. 46 |
| Yield anomaly | Aerodrome USDC-VELVET printed 14,839.31% APY | The pool's reward component was 14,838.74%; this is an emissions and exit-liquidity check, not a base return. 5 |
The daily path was not a straight line. TVL fell sharply on July 28, recovered part of that move on July 29, then slipped through August 2 before a small August 3 rebound.
Loading chart…
Chain TVL: Ethereum still supplies the base, but smaller ecosystems are moving faster
The chain endpoint summed to $73.926B across its returned records, close to but not identical with the historical aggregate above. The difference is a source-timing and endpoint-definition issue, not a weekly chain-flow figure. The endpoint returned current levels but no comparable seven-day change for each chain, so the table shows composition only. 7
| Chain | Current TVL | Share of chain-feed sum | What can be said from the snapshot |
|---|---|---|---|
| Ethereum | $40.509B | 54.80% | Still the main liquidity base. |
| BSC | $4.850B | 6.56% | Larger than Solana in this level snapshot. |
| Tron | $4.841B | 6.55% | Similar in size to BSC; the level alone says nothing about weekly flow. |
| Solana | $4.718B | 6.38% | One of the largest non-Ethereum ecosystems in the returned feed. |
| Base | $4.519B | 6.11% | Remains the largest listed newer L2 by current TVL. |
| Bitcoin | $3.489B | 4.72% | A sizable non-EVM TVL bucket in the endpoint. |
| Hyperliquid L1 | $1.208B | 1.63% | Large enough for protocol-specific moves to affect the chain view. |
| Arbitrum | $1.173B | 1.59% | The chain level does not identify which applications drove it. |
| Polygon | $796.9M | 1.08% | Smaller than the main L2 group. |
| Monad | $751.2M | 1.02% | Still a small base relative to Ethereum. |
Plasma was at $709.3M and Robinhood Chain at $391.5M. Those smaller bases can produce large percentage changes from one protocol launch, migration, or accounting adjustment. Neither should be read as a weekly inflow without a chain-level time series. 7
Protocol ranking: vault curation rose into the top spot while staking unwound
For a comparable screen, I kept protocols with at least $50M of current TVL, excluded records categorized as CEX, and calculated the comparison value as
current TVL / (1 + change_7d / 100). A positive rank delta means the protocol moved up in the current screen; ranks are only within this screen, not the full DeFiLlama universe. 2Largest dollar gainers
| Protocol | TVL before | Current TVL | 7-day change | Dollar move | Rank delta |
|---|---|---|---|---|---|
| Steakhouse Financial | $2.461B | $3.038B | +23.45% | +$577.2M | +6 |
| Figure Markets Democratized Prime | $47.2M | $185.8M | +293.47% | +$138.6M | +157 |
| CCIP | $1.607B | $1.705B | +6.06% | +$97.4M | +3 |
| USDT0 | $3.385B | $3.463B | +2.32% | +$78.4M | 0 |
| Figure Markets Exchange | $1.501B | $1.559B | +3.87% | +$58.2M | +1 |
Largest dollar losers
| Protocol | TVL before | Current TVL | 7-day change | Dollar move | Rank delta |
|---|---|---|---|---|---|
| Lido | $18.243B | $17.415B | -4.53% | -$827.1M | 0 |
| Babylon Protocol | $3.336B | $2.607B | -21.86% | -$729.3M | -5 |
| SSV Network | $9.599B | $9.172B | -4.45% | -$426.8M | 0 |
| Binance staked ETH | $7.250B | $6.875B | -5.17% | -$374.6M | 0 |
| Aave V3 | $14.162B | $13.830B | -2.34% | -$332.0M | 0 |
All rows above are derived from the same DeFiLlama protocol snapshot. The tables therefore establish the size and direction of the moves, not their cause. 2
Steakhouse Financial is the one large gainer with a concrete in-window product context. Kamino opened an isolated PAXG/USDG lending market on Solana on July 27, with Steakhouse Financial as curator and a reported opening borrow rate of 1%. That can explain why curator-linked TVL was in motion, but it cannot explain the whole $577.2M increase: the DeFiLlama record spans Base, Ethereum, Robinhood Chain, Solana, and other chains. Treat the Kamino launch as a plausible contributing event, not a proven attribution. 28
The loss side is more concentrated in staking and BTC-linked exposure. Lido, SSV Network, Binance staked ETH, and Babylon all fell, but the TVL feed cannot separate asset-price changes from withdrawals, redemptions, or accounting. Aave V3's decline has one piece of secondary context: TLDR Crypto reported on July 31 that Aave was deprecating 50 low-adoption reserves, winding down six chain deployments, and retiring 21 matured Pendle PTs. The Aave forum detail page was not readable in this pass, so treat that report as context rather than proof that the full $332.0M decline came from the wind-down. 29
Security: $8.788M, led by a BSC access-control drain
DeFiLlama's in-window hack records carry day labels, not hour-level event times. The three quantified records dated July 28 and July 30 total $8.7876M. No returned-funds value was present for any of them. 3
| Record | Chain | Amount | Attack vector | Status |
|---|---|---|---|---|
| Crypto DAO Pro Token Vault | BSC | $8.200M | A publicly callable Attack function and reserve manipulation repeatedly exchanged a small amount of Pro tokens for disproportionate USDT. | The incident report said no funds had been recovered and no project post-mortem or acknowledgment was visible as of publication. 10 |
| LULA | BSC | $578K | Price manipulation attack, according to the DeFiLlama classification. | Recovery and remediation were not disclosed in the returned record. 3 |
| Set Protocol | Ethereum | $9.6K | Malicious SetToken and Manager Contract, according to the DeFiLlama classification. | Recovery and remediation were not disclosed in the returned record. 3 |
The Crypto DAO event deserves more weight than its dollar figure alone suggests. The opened incident report reconstructs a public state-changing function on BSC, at least 12 repeated calls, and a total near $8.2M USDT. That is a basic permissioning failure combined with a pool-reserve drain, not a flash-loan label that can be dismissed as a one-block pricing accident. The report also says no recovery had been confirmed. 10
The two smaller entries should not be inflated into a broader claim. Their structured records identify the chain, amount, and technique, but not the contract, transaction, or remediation state. For a trader, that means the correct next step is to check the affected token and pool addresses before assuming the protocol is safe again. 3
Governance: one clear outcome, with the vote details missing
The in-window governance event with a usable detail page was Rarible's return to Solana. Rarible's official account said the community had approved the move and that development, integrations, and security audits were starting, with a four-week target. The supporting report explicitly says the vote margin and quorum were not disclosed. 46
| DAO / proposal | Close or announcement | Result | Voter blocs / quorum | Execution impact |
|---|---|---|---|---|
| RARI DAO: return Rarible to Solana | July 31, 2026 | Passed / approved, based on Rarible's official announcement | Vote margin and quorum not disclosed; voter blocs not identified. 4 | Development, integrations, and security audits began; the stated target was about four weeks. 6 |
This is an execution signal, not a completed deployment. The public announcement starts work and audits; it does not show a live Solana marketplace, an on-chain execution transaction, or a passed quorum threshold. The missing vote data is material for anyone pricing RARI governance impact. 4
Yield anomalies: rewards are doing the talking
The yield feed is most useful here as a screen for what needs checking. It does not prove that the displayed APY will persist, and its comparison fields are percentage-point changes against recent baselines. 5
| Pool | Chain | TVL | Current APY | 7-day comparison | 30-day comparison | What stands out |
|---|---|---|---|---|---|---|
| USDC-VELVET on Aerodrome Slipstream | Base | $2.981M | 14,839.31% | -19,925.83 pp | -845,542.98 pp | Base APY was 0.57% and reward APY 14,838.74%. 5 |
| ZBU on Zeebu | Base | $1.432M | 941.15% | -1,813.55 pp | -3,594.55 pp | The print is far above both recent baselines. 5 |
| USDC on Growihf | Hyperliquid L1 | $9.233M | 0.00% | -33.63 pp | -3,366.30 pp | A rate collapse, not a high-yield opportunity. 5 |
| WETH-AERO on Aerodrome Slipstream | Base | $2.216M | 67.68% | -1,177.03 pp | +46.04 pp | Current APY is still high while the seven-day comparison has fallen sharply. 5 |
The Aerodrome USDC-VELVET row is almost a pure incentive signal: 14,838.74 of the 14,839.31% headline APY comes from rewards in the returned data. Before sizing it, a farmer needs the reward token's sell depth, emissions schedule, pool volume, and the cost of exiting. The current APY says none of those things. 5
The opposite signal is Growihf's USDC pool, where the current APY is zero against a 30-day comparison of -3,366.30 percentage points. A collapsed rate can matter as much as a spike: capital may still be exposed to smart-contract, bridge, or stablecoin risk after the reward stream has disappeared. 5
Position screen
- Staking exposure: Lido, SSV Network, and Binance staked ETH all fell in the dollar ranking while total DeFi TVL declined. Separate ETH price effects from net deposits before calling this a staking exit. 12
- Curated lending exposure: Steakhouse Financial's $577.2M gain coincided with Kamino's new isolated PAXG/USDG market, but the gain spans several chains. Review each underlying vault and its curator-controlled risk parameters rather than treating the curator name as a single pool. 28
- BSC security: Crypto DAO and LULA account for $8.778M of the week's $8.788M quantified records. The common denominator is not a single exploit technique; it is concentration of observed loss on BSC and incomplete remediation data for the smaller incident. 310
- Yield farming: For the largest APY print, almost the entire displayed rate is rewards. Verify emissions, reward-token liquidity, pool volume, and exit paths before treating the number as income. 5
- Governance exposure: Rarible's announcement establishes a direction and a work plan, not a live Solana deployment. The absent vote margin and quorum should remain part of the risk note. 46
The week's cleanest signal is the split between a 4.08% aggregate TVL loss and a 23.45% rise in one curator-linked protocol. The next check is whether the vault inflow survives after the new-market launch and whether the staking losses stabilize without another security record adding forced outflows. 12
References
- 1DeFiLlama historical chain TVL
api.llama.fi
- 2DeFiLlama protocol TVL
api.llama.fi
- 3DeFiLlama hack database
api.llama.fi
- 4
- 5DeFiLlama yield pools
yields.llama.fi
- 6Rarible governance context
solanacompass.com
- 7DeFiLlama chain TVL
api.llama.fi
- 8Kamino PAXG market detail
solanacompass.com
- 9TLDR Crypto July 31
tldr.tech
- 10Crypto DAO incident report
0xposed.io
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.