KoinlyticsDeFi total value locked has contracted to roughly $70 billion, down about 39% year to date and 59.6% from the November 2021 all-time high of $177.5 billion. Most of the 2026 decline came after the January cycle high. Ethereum remains the dominant venue with $38.24 billion, a 53.1% share.
The distribution underneath that headline is where the story is. Among the ten largest ecosystems by TVL, exactly two grew this year: TRON, up roughly 5%, and Hyperliquid, up about 7%. Every other top-10 chain shrank.
Two Growth Stories, Neither of Them Standard DeFi
TRON
TRON's resilience has almost nothing to do with speculative on-chain activity. It is the primary settlement rail for USDT, particularly across emerging market corridors where transfer cost and finality matter more than composability. TRON's TVL is anchored by stablecoin float doing payment work, not by yield farming or leveraged lending.
That is a fundamentally different demand driver than the rest of the category. Payment volume does not care about the DeFi risk appetite cycle. It cares whether the alternative, correspondent banking, is slower and more expensive, which it generally is. TRON grew because its use case was never the one that broke.
Hyperliquid
Hyperliquid's growth is concentrated in perpetual futures rather than lending or spot AMM liquidity. It has taken meaningful share in on-chain derivatives, a segment that has held up far better than the rest of DeFi this year. Its fully diluted valuation briefly overtook Solana's earlier this month, which is either a signal about where on-chain activity is going or a signal about how thin the comparison is, depending on your view of FDV as a metric.
What both growth cases share is that they are not the 2021 DeFi model. Neither is primarily about depositing tokens to earn emissions.
Why the Contraction Happened
Three forces compounded through 2026 and it is worth separating them because they have different fixes:
- Security. 121 hacks so far in 2026 with roughly $942 million in losses. Q2 alone accounted for 85 incidents and around $775 million, the worst quarter in the dataset. Depositors do not need to be hacked personally to conclude the expected loss is higher than the yield.
- Rates. With Treasury yields at multi-decade highs and September hike odds near 61%, the risk-free alternative to a DeFi lending position is paying real money for zero smart contract risk. That comparison has never been less favourable to DeFi.
- Price. TVL is denominated in dollars. When the underlying tokens fall, TVL falls without a single deposit being withdrawn. A meaningful portion of the 39% is arithmetic rather than behaviour.
That third point is routinely ignored and it changes the interpretation substantially. Distinguishing between a chain that lost deposits and a chain whose deposits lost value requires looking at native-denominated TVL, which most dashboards do not surface by default.
The Gap That Defines the Cycle
Stablecoin supply sits around $307.5 billion against $70 billion of DeFi TVL. That ratio, roughly 4.4 to 1, has widened all year.
Two hundred and thirty billion dollars of on-chain dollar liquidity exists and is not deployed in DeFi. That capital did not leave crypto. It is sitting in wallets, on exchanges, in payment rails, and in yield-bearing formats that do not require locking into a lending market or an AMM pool.
The optimistic read is that this is dry powder waiting for a catalyst. The realistic read is that a large portion of it is doing a job, settlement and payments, that was never going to migrate into DeFi in the first place. Both are partly true and the ratio between them is the actual open question for 2027.
DeFi did not lose its users to a competitor. It lost them to a risk-free rate and a hack calendar. Neither is fixed by better tokenomics.
What Is Actually Working
The parts of DeFi still growing share a profile. Lending has concentrated heavily in the most audited venues, with Aave alone controlling close to half of on-chain lending. Perpetual DEX volumes have held up better than spot. Tokenised Treasury collateral has grown because it imports the risk-free rate on-chain rather than competing with it.
The common thread is that survivors either offer institutional-grade risk profiles or serve a use case with no off-chain substitute. Mid-tier protocols competing on yield alone have been the clearest losers.
What to Watch
- Native-denominated TVL rather than dollar TVL, which separates outflow from price effect
- Whether the stablecoin-to-TVL ratio compresses, the cleanest signal of capital re-entering
- Q3 exploit count, after Q2's 85 incidents
- Hyperliquid and TRON share gains continuing, or reverting as their specific drivers mature
- Tokenised Treasury collateral growth as a share of total lending
If you run positions across multiple chains, the aggregate TVL number tells you almost nothing about your own exposure. What matters is the depth of the specific pools you are in and whether you could exit them at size without moving the price. Those are pool-level questions, and a falling category-wide TVL usually means the answer has gotten worse without any notification.