DeFi's February 2026 Crash Test: TVL Fell, Fees Spiked
DeFi TVL fell 12.5% in February 2026's sell-off, from $120B to $105B. Fees at Aave, Hyperliquid, and Morpho jumped 1.8x to 5.1x on the crash's worst day.
DeFi's total value locked fell 12.5% during February 2026's sell-off, from $120 billion to $105 billion, while Bitcoin, Ethereum, XRP, and Solana all fell to multi-year lows, with Ethereum losing 21% of its value in the week alone (CoinDesk) and Bitcoin falling 18.1%, from $89,212 to $73,059, over the same January 29-February 5 span (CoinGecko market chart API). The more telling number sits underneath: on the worst single day of the crash, Aave and Hyperliquid's protocol revenue did not shrink, and neither did gross fees at Aave, Hyperliquid, and Morpho. Against the January 29-February 4 daily average, both spiked: revenue by roughly 1.8-2.2x at Aave and Hyperliquid, gross fees by about 5.1x at Morpho.
The Numbers Don't Lie
The Fear & Greed Index stood at 14, deep in "extreme fear" territory, during the February 2026 sell-off. Bitcoin, Ethereum, XRP, and Solana all touched multi-year lows in that window. Yet DeFi resilience told a different story.
| Metric | Change | Window | Notes |
|---|---|---|---|
| DeFi TVL | -12.5% | 7-day | From $120B to $105B |
| Bitcoin | -18.1% | 7-day, Jan 29-Feb 5 | $89,212 to $73,059, per CoinGecko |
| Ethereum | -21% | 7-day | Per CoinDesk |
| ETH in DeFi | +12% | Year-to-date, since Jan 1 | 25.3M ETH, up from 22.6M |
The two crashes are not measured the same way, and that gap matters more than any single number. In February 2025's selloff, three MakerDAO positions worth $340 million combined faced liquidation if ETH fell another 19%, below $1,900 (CoinDesk): one protocol's largest at-risk book, measured against a fixed price trigger. In February 2026's comparable stress, CoinDesk put DeFi-wide liquidatable positions within 20% of the then-current price at just $53 million (CoinDesk): every protocol, measured against a moving percentage band. Turning those two into a single reduction percentage would compare a MakerDAO-only figure to a sector-wide one, so this piece does not. Read separately, they still argue the same direction: 2026's sector-wide near-term exposure sits well below what a single 2025 protocol carried in one book alone. TVL itself fell a comparable or smaller percentage across both years depending on the measurement window: 2025 ranged from a 10.3% one-day drop to a 22.45% month-end decline, versus 2026's 12.5% over the week (CryptoSlate). Within the 2.7 million year-to-date increase in ETH deposits, DeFi added 1.6 million in the single week before the crash peaked. Deposits rose rather than fell during that window; who was depositing and why is not something this deposit count shows.
Why DeFi Resilience Looks Different This Time
The 2022 bear market destroyed DeFi confidence. Unsustainable yields backed by worthless tokens collapsed alongside Terra, Celsius, and FTX. Three years later, at Aave, Hyperliquid, and Morpho, the crash-day numbers below show real fee income rather than inflationary token rewards.
Which Protocols Actually Earned Fees Through the Drawdown
The claim of DeFi resilience during a crash is easy to make and hard to verify. DefiLlama's public fee-tracking API gives a protocol-by-protocol answer, and it forces a distinction that matters: gross fees (what users pay in) and protocol revenue (what the protocol itself retains) are different numbers, and they moved differently across these three protocols on the crash's worst day.
Aave's protocol revenue, the cut it keeps after paying suppliers, averaged $494,667 a day over the seven days from January 29 through February 4, 2026, a window that itself included a secondary spike of $1.11 million on January 31. On February 5, the day Bitcoin and Ethereum were bleeding hardest, it reached $1.10 million, about 2.2x that daily average (DefiLlama fees API). Aave's gross fees that same day, what borrowers actually paid before Aave's cut, hit $9.55 million against a $4.09 million average over the same window (DefiLlama fees API). Over the trailing year, Aave has generated $111 million in protocol revenue on $850 million in total fees, income that comes from borrowers paying interest and liquidation penalties, not from token emissions.
Hyperliquid's daily revenue averaged $3.40 million over the same January 29-February 4 window and hit $5.99 million on February 5, about 1.8x that average, as traders paid to open and close leveraged positions into the volatility (DefiLlama fees API). Its gross fees that day were $7.54 million against a $4.28 million average over the same window (DefiLlama fees API).
Morpho is a different case, and the difference is the point. Morpho's daily gross fees averaged $994,893 over the same January 29-February 4 window and spiked to $5.10 million on February 5, about 5.1x that average (DefiLlama fees API). DefiLlama's daily aggregate does not break that spike down by transaction type, so this update cannot confirm how much of it came from liquidations versus refinancing versus new borrowing; it records the size and timing of the spike, not its composition. DefiLlama's dailyRevenue series for Morpho reads $0 on every day in this window, including February 5. Morpho is a peer-to-peer matching layer: DefiLlama attributes none of the $5.10 million to protocol revenue, recording Morpho's dailyRevenue as zero, so none of it accrued to the protocol itself. Calling that figure "Morpho's revenue," as an earlier version of this article did, mixed a fees number with two revenue numbers and made all three look like the same kind of spike. They are not: Aave and Hyperliquid genuinely earned more for themselves in the crash; Morpho's crash-day number is a gross-fee increase, not a revenue figure, and not that Morpho got paid more.
A plausible mechanism fits this single episode: lending protocols would be expected to see liquidation penalties and interest-rate spikes when borrowers are stressed, and perpetuals exchanges would be expected to see trading fees when volatility drives volume. The DefiLlama daily aggregates used here show timing and magnitude only, not transaction composition, so this remains an association observed on one crash day, not a mechanism this update has confirmed.
At Aave, Hyperliquid, and Morpho, the crash-day numbers above came from fee income, not inflationary token rewards. For Aave and Hyperliquid that fee income became protocol revenue; for Morpho, DefiLlama records zero protocol revenue from it, as detailed below.
Hyperliquid's Buyback Turns Fees Into a Structural Bid
Hyperliquid routes 97% of protocol trading fees (99% for certain fee categories after a December 2025 governance vote) into its Assistance Fund, which buys HYPE on the open market and removes it from circulation. By May 2026 the fund had spent more than $1.3 billion on buybacks, and Hyperliquid's fees were running at roughly $1.3 billion annualized (crypto.news). As of that article's May 27, 2026 publication date, crypto.news put the buyback rate at roughly 7% of HYPE's market cap, annualized against market cap at the time of writing. Using the same annualized-against-current-market-cap method, it put Ethereum's token-burn rate near 1.5% of its market cap and Solana's priority-fee burn near 0.5% (crypto.news). None of the three underlying market caps are stated in the source, so treat the comparison as a same-method, same-source estimate rather than as three independently verified figures. The crash-day fee spike documented above feeds directly into this fund: more trading volume in stress periods means more forced buying of HYPE, not less.
Limited Near-Term Liquidation Exposure in This Snapshot
The sector showed limited near-term liquidation exposure in this snapshot. CoinDesk reports only $53 million in DeFi positions sitting within 20% of the then-current price, and names the specific threshold on the largest at-risk book: positions on Compound only become vulnerable if ETH slides below $1,800 (CoinDesk). That $53 million in near-term liquidatable exposure compares against more than $100 billion locked across the sector on the same day; TVL is not a collateralization denominator, so this figure is evidence for DeFi resilience in the narrow sense of low near-term liquidation risk, not a broader claim about how well-collateralized the sector was overall.
Index-Provider Recognition, With a Correction on Timing
Nansen launched its NX8 index on February 3, 2026, tracking eight major Layer 1 networks: Bitcoin, Ethereum, Solana, BNB, Tron, Hyperliquid, Avalanche, and Sui (Nansen NX8 launch coverage). Including a DeFi-native perpetuals platform alongside base-layer chains put Hyperliquid in the same index Nansen uses for the largest networks; what that categorization means to Nansen is not stated, and this update has no evidence other analytics firms made the same call.
HYPE's market cap did not overtake Cardano's ADA by early February, however. That flip happened weeks later, in March 2026, when HYPE traded in the $40-43 range against ADA near $0.29 (Coinpedia). The sequence is established: institutional index inclusion in early February preceded a real market-cap milestone a few weeks out. This update does not have evidence on whether any earlier recognition, before Nansen's inclusion, was already priced in.
The Standout Performers
While markets bled, some DeFi protocols kept earning through it, their fee engines running hotter even as prices fell alongside the broader market.
Hyperliquid (HYPE) was one of the protocols whose fee income spiked on the crash's worst day. Daily revenue rose to $5.99 million on February 5, about 1.8x its January 29-February 4 average, and gross fees rose to $7.54 million against a $4.28 million average, as traders paid to open and close leveraged positions into the volatility (DefiLlama revenue endpoint, DefiLlama fees endpoint).
Morpho (MORPHO) token price fell roughly 18% over the January 29-February 5 crash window, in line with Bitcoin's 18.1% decline over the same days, not ahead of it (CoinGecko market chart API). What stood out was not the price. The protocol optimizes lending rates between Aave and Compound, capturing spreads that benefit users, and its fee engine, shown above, recorded a gross-fee increase on the crash day, even though Morpho itself retains none of that fee flow.
The Investment Thesis
Aave, Hyperliquid, and Morpho's crash-day fee and revenue numbers, sourced from DefiLlama, describe three specific businesses on one day, February 5, 2026. Aave and Hyperliquid earned more in protocol revenue that day than their week-before average; Morpho's gross fees spiked too, though none of that flowed to Morpho itself. Together the three show that on this one crash day, at least some DeFi fee engines ran hotter under stress, a claim scoped to this single episode and to these three protocols, not to DeFi as a sector or to stress events generally.
The evidence here supports a narrow reading of DeFi resilience, not a sector-wide guarantee.
Bull Case
ETH accumulation signal. ETH deposits in DeFi rose 12% year-to-date, from 22.6 million to 25.3 million, even as price fell. Deposit counts identify neither who is depositing nor why: this data does not show whether the depositors are large "smart money" wallets or smaller holders, and it does not distinguish a deliberate accumulation strategy from any other reason ETH might sit in a DeFi protocol during a selloff.
Fees spiked on this one stress day. February 5 was the single largest day for gross fees in the entire month for all three protocols, checked against DefiLlama's full daily series rather than just the surrounding week (Aave, Hyperliquid, Morpho), and for Aave and Hyperliquid that translated directly into more protocol revenue. That is one crash day, not a pattern this update has tracked across multiple stress events, but it is evidence that Aave's and Hyperliquid's revenue held up, and grew, on February 5, 2026.
Bear Case
Still correlated to crypto. DeFi resilience did not mean decoupling from broader market movements. DeFi TVL fell 12.5% over the week. Bitcoin's price fell 18.1% and Ethereum's price fell 21% over the same week.
Sustainability concerns remain uneven. Aave and Sky post real, audited-by-market revenue; plenty of smaller lending and yield protocols still lean on token incentives; this article tracked three protocols, so it cannot say how concentrated sector-wide revenue is.
Macro headwinds persist. Hawkish Fed policy and institutional ETF outflows create ongoing pressure on risk assets independent of anything happening on-chain.
What the Data Shows Since
Sky (formerly MakerDAO) reported Q1 2026 gross protocol revenue of $124 million and net revenue of about $61 million, the highest income the protocol has generated since it launched in 2017 (DL News). Management credited growing institutional demand for risk-adjusted onchain yield, largely through Sky's USDS stablecoin. That result, reported in April 2026, shows Sky's revenue held up through Q1 2026, the same quarter that contained the February crash, not beyond it: no post-Q1 figure appears in the source. It does not confirm anything about Aave, Hyperliquid, or Morpho: this article tracked their single crash-day numbers only, and no post-crash revenue series for those three protocols appears here.
During extreme fear, protocols with measured, sourced revenue (Aave, Sky, Hyperliquid) are the ones worth watching. Protocols whose "yield" cannot be traced to a specific fee line on DefiLlama have unverified yield on DefiLlama: that gap means the claim needs corroboration from another auditable source, such as the protocol's own onchain fee contract or a separate tracker, before treating it as genuine rather than emissions-funded.
Risks to Monitor
DeFi resilience does not mean safety. Key risks require attention.
- ETH price floor. CoinDesk's February reporting named $1,800 as the then-current level where Compound's largest at-risk positions would convert from stressed to liquidated; that threshold described that specific book at that date, not a sector-wide cascade trigger, and this article has no updated figure for where it sits now.
- Token unlocks. Morpho and other protocols carry strategic partner vesting schedules that can pressure token prices independent of protocol revenue.
- Regulatory uncertainty. SEC and CFTC clarity on staking and DeFi protocols remains unresolved.
- Bridge exploits. Cross-chain DeFi adds attack surface that single-chain protocols avoid.
The Bottom Line
DeFi-wide, TVL fell 12.5% over the February 2026 correction week, while Bitcoin's price fell 18.1% and Ethereum's price fell 21% over the same week. Two sector-wide observations bound the DeFi resilience claim this piece makes: TVL declined right alongside those larger asset drops, and near-term liquidation exposure sat at just $53 million against more than $100 billion locked across the sector, evidence for resilience only in that narrow, near-term-liquidation sense, not a broader claim about how the sector performed overall.
The profit story, checked against the full month rather than just the surrounding week, is narrower and belongs to three named protocols. February 5 was Aave's, Hyperliquid's, and Morpho's single biggest day for gross fees of the month. For Aave and Hyperliquid, whose fee lines can now be checked against DefiLlama's public data rather than marketing claims, that fed directly into higher protocol revenue, and Hyperliquid's Assistance Fund converted its revenue spike directly into HYPE buybacks. Morpho's gross fees spiked that day too, but Morpho itself, by design, kept none of it. That is evidence that at least these three fee-generating businesses saw gross fees rise, and for Aave and Hyperliquid, got paid more, exactly when the market panicked.
Disclaimer: Nothing here constitutes financial advice. Cryptocurrency investments carry significant risk. Always conduct independent research and consult a qualified financial advisor before making investment decisions.
Related Reading
Weekly Crypto Insights
Market analysis and actionable insights. No spam, ever.