DefiLlama displays a single, comforting number on its homepage this morning: $71.10B of total value locked across DeFi. The panda has been reading that number for years, alongside how it is actually built. They are not the same exercise.
Why DeFi's TVL Number Stopped Telling the Truth
Total value locked was a useful metric back when DeFi meant "ETH sitting in a Compound contract." One deposit, one count, one risk surface. Since 2024, liquid staking, restaking, and LST-collateralised lending have stacked on top of each other in a way the metric never anticipated. According to DefiLlama's chains dashboard, DeFi TVL is now $71.10B, with Ethereum alone at $37.24B. Solana sits at $4.58B, BSC at $5.17B.
Those headline figures count every dollar of collateral inside every protocol, regardless of whether the underlying asset is the same one already counted upstream. In a 2020 DeFi, that produced almost no double counting. In a 2026 DeFi, where the same ETH can be a staking deposit, a liquid staking token, a liquid restaking token, and a lending collateral at the same time, the gap between gross TVL and net economic value has stopped being small.
Call it accounting drift. Call it brand inflation. The panda calls it the polite version of the truth.
How Does Restaking Triple-Count the Same ETH?
The mechanism is depressingly elegant. A user deposits 1 ETH into Lido and receives stETH, a liquid staking token. That stETH is then deposited into ether.fi or Renzo, which restakes the underlying validator credentials with EigenLayer and issues a liquid restaking token (weETH, ezETH). The user takes that LRT and posts it as collateral in Aave or Morpho, borrowing more stable assets to deploy elsewhere. One asset, four protocol counts.
According to DefiLlama's ether.fi protocol page, ether.fi alone routes billions through that pipeline. According to DefiLlama's EigenLayer page, EigenLayer's restaking TVL is reported alongside the LST TVL that feeds it, not netted against it. According to DefiLlama's Lido page, Lido's TVL is reported in full, even when its stETH is sitting inside another protocol's TVL.
Coverage from The Defiant on restaking accounting has been candid about this for over a year, noting that restaking is, by definition, a "re-use" of already-staked ETH rather than a new deposit. The word "re-use" is doing a lot of work. The TVL dashboard treats it as new deposit value, not re-used capital. So does most reporting downstream.
A reasonable adjustment, netting LST inside LRT inside lending collateral, deflates Ethereum's $37.24B figure by something between 30 and 50 percent depending on how aggressively you collapse the loops. The total $71.10B headline sits closer to $30 to $40B of distinct economic exposure once you do that. Nobody quotes that number because nobody wins by quoting it.
What the $71B Number Actually Measures
It measures protocol activity, not capital at risk. Each protocol's contract balance is real. The same dollar circulating between four protocols also generates four sets of fees, four sets of governance token emissions, four sets of marketing claims about "TVL growth." That part is unambiguously real.
What it does not measure is how much value would actually leave DeFi if the underlying ETH walked away. Strip the wrapping layers and the user only has 1 ETH to withdraw, not four. Spoiler: we saw this one coming when EigenLayer launched and every LST issuer immediately rebranded as an LRT.
For context on the underlying asset, ETH is trading at $1.66K according to CoinGecko's Ethereum page, with a market cap of $200.12B. DeFi TVL denominated in dollars is therefore extremely sensitive to ETH price moves: a 20 percent ETH drawdown shaves roughly $7B off the headline number before any actual unwind, simply by repricing the same collateral lower. The metric does not distinguish "TVL fell because users left" from "TVL fell because price moved."
For readers wanting the basics first, our explainer on what restaking is and why it exists walks through the EigenLayer mechanism without the accounting commentary. The wider DeFi cluster page catalogs the rest of the protocols feeding this stack.
Risks: When Layered TVL Becomes Layered Fragility
The triple-counting story is funny. The risk story is not.
Smart-contract risk compounds. A user sitting on weETH-collateralised Aave debt is exposed to bugs in Lido, ether.fi, EigenLayer's slasher contracts, and Aave simultaneously. Any one of the four breaking impairs the position. The recent ether.fi naked-yield analysis on this blog laid out how thin the post-points yield actually is once that risk is priced.
Depeg risk cascades. If stETH trades at a 3 percent discount to ETH, every LRT built on stETH inherits the discount, and every lending position using that LRT as collateral takes a margin call. The 2022 stETH discount lasted weeks. A similar event in 2026 would propagate through three layers of leverage before reaching the borrower.
Slashing risk migrates. When EigenLayer's AVS slashing finally goes live in production, a single validator misbehavior penalises the LST, the LRT wrapping it, and any lending position collateralised by that LRT. The slashing parameters are still being negotiated, and ongoing coverage at Blockworks on EigenLayer AVS economics regularly notes that the cascading impact is not yet fully priced into LRT prices.
Governance risk concentrates. veCRV, veAERO, and EigenLayer's restaking quorum all give outsized influence to whoever stacks the deepest LST positions. The token holders of one protocol can effectively vote with capital that has already been counted as TVL three protocols upstream.
What to Watch Next
Three signals matter over the next quarter. First, whether DefiLlama publishes a "netted TVL" view that collapses the LST and LRT loops. The team has discussed this on their public methodology page, and the political resistance comes from protocols that benefit from inflated headlines. Second, whether EigenLayer's slashing launch produces the first cascading LRT depeg, which would force the industry to face its accounting publicly. Third, whether BSC's slow TVL bleed of 1.74 percent over the last seven days, also per DefiLlama's BSC page, tempts protocols to migrate liquidity to chains where the same triple-counting trick is harder to pull off.
On the satirical-memecoin side of this blog, the takeaway is older than DeFi: a number is not the thing it claims to measure. Dadacoin is a memecoin on BSC that has never pretended to be a yield platform, in a sector where yield platforms increasingly pretend not to be repackaging the same ETH four times. The numbers say $71B. The panda raises an eyebrow.
