TVL Is Not Capital. Here’s What That Number Actually Measures and Hides

A billion dollars locked in DeFi does not necessarily mean a billion dollars of unique capital. Follow the assets, the claims built on top of them, and the economic activity the headline number leaves out.

Imagine depositing $100 million worth of ETH into a liquid staking protocol.

The protocol issues a token representing that stake. You then deposit that token into a lending market. Two protocols now have exposure to an asset position worth approximately $100 million, and each can report activity involving that value.

Add their individual figures without adjusting for overlapping claims, and the apparent total becomes $200 million.

Yet the original deposit was still $100 million.

No additional ETH entered the system. The asset simply acquired another financial representation.

This is one of the fundamental problems with interpreting Total Value Locked (TVL) as a measure of capital. The number can accurately describe assets associated with individual protocols while producing a misleading picture when treated as unique economic capital across an interconnected system.

The distinction matters because DeFi increasingly consists of assets being staked, wrapped, deposited, borrowed against, and reused across different applications.

Each additional financial layer can create a legitimate service, but it does not necessarily introduce additional underlying capital.

Understanding TVL therefore requires more than reading the number. It requires reconstructing the economic relationships behind it.

What TVL Actually Measures and What It Does Not

TVL is generally calculated by multiplying the quantity of eligible assets held in protocol contracts by their market prices, then adding those values together.

In simplified form:

TVL = Σ (Token Quantity × Token Price)

If a protocol holds 10,000 ETH valued at $2,500 each, its reported TVL is $25 million.

That calculation answers a specific question: What is the current market value of assets counted within the protocol?

It does not automatically answer several others.

How much capital has entered from outside the ecosystem? How much is available for borrowing or trading? How much has been pledged as collateral? How much activity does the capital generate? How much belongs to depositors rather than the protocol itself?

These are different economic questions that require different measurements.

According to DefiLlama’s official data definitions, TVL is intended to measure assets deposited into and exposed to protocols. Its methodology distinguishes several categories of assets and applies accounting rules designed to avoid misleading aggregation, including double counting.

That distinction is important. Not every TVL dashboard blindly counts every representation of an asset, and established analytics providers already apply safeguards against duplicate reporting.

But the existence of those safeguards also reveals the underlying analytical problem: counting financial positions is not always equivalent to counting independent capital.

Consider two hypothetical protocols with identical TVL.

Protocol A holds $500 million in stablecoins deposited by lenders, with active borrowers paying interest.

Protocol B holds $500 million in an eligible asset whose value has doubled, even though its token balance has remained unchanged.

Both could display the same TVL.

Their capital flows, economic activity, revenue potential, and exposure to market prices could nevertheless be radically different.

TVL provides the starting point for analysis, not the conclusion.

The $100 Million Test: How One Deposit Can Appear Twice

To understand overlapping capital, consider a hypothetical transaction involving two existing DeFi mechanisms: liquid staking and collateralized lending.

The example uses Lido and Aave because their documentation explicitly describes how tokenized staking positions can interact with lending markets.

It is a controlled illustration, not a claim about a particular historical wallet transaction or the current size of either protocol.

One Asset, Two Financial Layers

Step one: Deposit $100 million of ETH into liquid staking.

A depositor stakes ETH through Lido and receives stETH, a transferable token representing the underlying staking position.

The depositor can also convert stETH into wstETH, a wrapped representation that accrues staking value through its exchange rate rather than a changing token balance.

Lido’s official token integration guide explains that wstETH can be used as collateral in Aave V3 markets.

For this example, assume the initial staking position is worth exactly $100 million, with no price changes, accumulated rewards, or transaction costs.

Lido now accounts for $100 million of underlying staked ETH.

Step two: Deposit the corresponding wstETH into Aave.

The depositor supplies the wrapped staking token to an eligible Aave lending market.

Aave now holds a tokenized claim on the original staked ETH. The depositor receives a lending-position representation, while the underlying ETH remains associated with its staking arrangement.

Aave’s V3 protocol documentation explains how supplied assets generate aTokens and can be used as collateral for borrowing.

The two positions can now be represented as follows.

Ravel calculation. Illustrative exposure accounting, not reported aggregate TVL from either protocol.

The $200 million figure represents two exposures associated with the same underlying capital, not $200 million in independently introduced assets.

In this simplified example, the naive exposure sum is twice the original capital base.

This does not necessarily mean either individual protocol’s accounting is incorrect.

Lido provides staking infrastructure. Aave provides a lending and collateral management service. Both can expose participants to distinct risks and generate different kinds of economic value.

The problem emerges when someone adds their reported asset values together and interprets them as two separate contributions of original capital.

DefiLlama addresses aspects of this problem through double-counting classifications and adjustments to how different protocol categories are aggregated. Its treatment of liquid staking also helps distinguish underlying staked assets from financial claims built on those positions.

The example therefore illustrates what an unadjusted aggregation could imply, not how every professional TVL aggregator calculates its headline number.

It also exposes three separate concepts that should not be confused.

Underlying capital refers to economically distinct assets supporting the positions. Financial claims represent rights to those assets, while protocol exposure describes the value associated with each financial service or smart contract.

These categories overlap, but they are not interchangeable.

A derivative claim may create additional functionality without creating additional underlying wealth.

And a single asset can support multiple interconnected financial relationships without becoming multiple independent assets.

The Three Forces That Can Inflate the TVL Narrative

Overlapping claims are only one source of confusion.

A TVL chart also combines several economic forces that deserve to be measured independently: leverage, changes in token prices, and incentives designed to attract deposits.

Recursive Lending: More Exposure, Not Necessarily More Equity

Consider a depositor who supplies collateral, borrows against it, purchases or stakes another asset, and deposits the resulting position as additional collateral.

If the borrowing and collateral parameters permit it, the process can be repeated.

Each cycle increases gross exposure while also creating an offsetting liability.

For a simplified position beginning with $100 of equity, assume a strategy repeatedly borrows and redeposits at a fixed 60% ratio, ignoring costs and interest.

Its theoretical gross-supplied exposure approaches:

$100 + $60 + $36 + $21.60 + … = $250

The resulting gross debt approaches $150.

The equity remains $100 before fees, accrued interest, and market movements.

This is a geometric illustration of leverage, not a representation of any particular protocol’s reported TVL. Actual lending constraints, asset availability, collateral rules, and double-counting exclusions affect what appears in published data.

The important point is that $250 in gross exposure is not equivalent to $250 of investor equity.

Some borrowing uses independently supplied assets. Other systems create liabilities through token issuance or different credit structures. The underlying funding and accounting must be identified before determining how much new capital entered.

Nor should every reuse of a receipt token be labelled ‘rehypothecation.’

Holding a transferable claim, pledging collateral, and legally or technically reusing pledged collateral are all different mechanisms.

The analytical distinction is essential because gross exposure, borrowed capital, and net equity measure different things, even when they originate from the same financial strategy.

Price Appreciation: TVL Can Rise Without Fresh Deposits

Return to the protocol with 10,000 ETH.

At $2,500 per ETH, its TVL is $25 million.

If ETH rises to $3,000 and the token balance remains unchanged, TVL increases to $30 million.

That is a 20% increase in reported TVL, generated entirely by price appreciation.

No additional ETH had to enter the protocol.

The opposite also occurs: an ecosystem can attract new deposits while its dollar-denominated TVL falls because token prices decline.

This makes token-denominated balances and net inflows essential complements to USD TVL.

An analyst looking only at dollar growth might incorrectly attribute market appreciation to investor adoption.

Similarly, an analyst observing a falling TVL chart might overlook positive deposit flows during a broader market decline.

The distinction between capital flows and asset repricing is fundamental.

Token Incentives: When Deposits Are Rented Rather Than Earned

DeFi protocols may distribute newly issued tokens or externally funded rewards to liquidity providers.

Curve’s gauge incentive documentation provides a concrete example of how token emissions and additional reward programs can influence capital allocation.

There is nothing inherently illegitimate about subsidizing liquidity. Incentives can help a market function during its early development.

But a deposit motivated primarily by emissions should not automatically be treated as evidence that the protocol has found sustainable economic demand.

Consider a hypothetical pool attracting $100 million while distributing rewards worth $12 million annually.

If the pool generates just $1 million in annual user fees, the reward rate substantially exceeds the observed fee generation.

That does not prove the incentives are ineffective. The protocol might deliberately subsidize early liquidity to develop a useful market.

It does mean the pool’s reported TVL cannot establish that its current activity will economically support the same level of capital after rewards decline.

A stronger analysis asks whether deposits persist, borrowing and trading demand remain, and fee generation continues once subsidies are reduced.

Without those observations, high TVL may reveal an effective incentive program rather than durable economic adoption.

Aave’s September 2026 Risk Review Shows What TVL Hides

The distinction between deposited value and underlying economic behavior is not merely theoretical.

A September 7, 2026 risk review published by LlamaRisk on Aave’s governance forum provides a detailed view of how differently two lending reserves can be structured.

The LlamaRisk assessment of Aave V3 markets examined depositor concentration, outstanding debt, and position health across several markets.

Two reserves were particularly revealing: wstETH on Aave V3 Prime and USDC on Aave V3 Core.

Source: LlamaRisk, September 7, 2026. The figures describe two separate lending reserves, not the entire Aave ecosystem.

An additional calculation illustrates the difference.

The five largest wstETH depositors controlled approximately 3.3 times the share held by the five largest USDC depositors when comparing concentration percentages in their respective reserves.

That does not make one reserve automatically unsafe or the other inherently superior. The assets serve different purposes and have different borrower behavior.

It does show how much information disappears when positions are reduced to a single dollar figure.

In the wstETH reserve, most of the largest depositors also carried debt. The risk assessment identified borrowing against staking-token collateral, including WETH and GHO liabilities.

Such positions can be economically legitimate. Investors may borrow against productive collateral to obtain liquidity without selling their staking exposure.

But the deposits are not equivalent to unleveraged capital sitting independently in a vault.

The USDC reserve exhibited a different profile. Its top-20 depositor concentration was substantially lower, and only three of those depositors carried debt in the reviewed market.

LlamaRisk characterized much of that supply as capital provided to earn borrowing interest rather than to support collateralized leverage.

Both types of deposits contribute to lending activity.

Yet their concentration, leverage exposure, liquidity requirements, and sensitivity to market conditions differ.

These figures are a dated risk snapshot, not permanent attributes of the reserves. They also do not establish that every indebted depositor was running a recursive strategy.

What they demonstrate is a more durable analytical principle:

The same category of TVL can contain economically different kinds of capital.

To evaluate a lending market, the number of dollars supplied is not enough. Investors also need to know who supplied them, how concentrated the deposits are, how much borrowing they support, and how quickly liquidity could leave.

Those details are often more useful for understanding resilience than headline TVL alone.

Beyond TVL: How to Evaluate the Capital Behind the Number

Suppose a protocol reports $1 billion in TVL.

Instead of treating that number as proof of success, an analyst can reconstruct it through five separate tests.

Trace the Economic Origin of Deposits

The first question is whether the reported assets represent independent capital or financial claims built upon positions already counted elsewhere.

Identify whether deposits consist of native assets, stablecoins, liquid staking tokens, LP tokens, wrapped assets, or other derivatives.

Where possible, trace derivative tokens back to their backing assets.

This helps distinguish direct capital from claims that may already appear elsewhere in the system.

The objective is not to dismiss derivatives. It is to avoid treating every derivative representation as independently introduced wealth.

Separate Capital Flows From Market Appreciation

A rising USD balance may reflect additional token deposits, price appreciation, or both.

Track changes in actual asset quantities alongside dollar-denominated TVL.

Metrics such as USD inflows can help distinguish changes in deposited balances from price-driven movements, provided you understand their underlying methodology and coverage.

A protocol receiving no net deposits during a token rally should not be described as having attracted fresh capital merely because its dollar TVL increased.

Likewise, negative USD TVL growth does not prove investors withdrew capital.

The more useful question is how much capital actually entered or left, independently of asset prices.

Identify the Liabilities Attached to Collateral

For lending protocols, examine supplied assets alongside outstanding borrows, collateral composition, utilization, and concentration.

Determine whether deposited positions are associated with leverage or largely represent independently funded lending supply.

Borrowing does not necessarily mean capital is unproductive. Credit provision can be a meaningful financial service.

But gross deposits, outstanding debt, and net equity describe different parts of that service.

They should not be combined or substituted for one another without a clearly defined accounting method.

Measure What the Capital Actually Enables

TVL alone cannot establish whether assets are supporting active lending, facilitating trades, securing positions, or simply remaining deposited.

The relevant activity metric depends on the protocol.

For a lending market, examine borrowing demand, interest paid, utilization, and realized credit losses.

For a decentralized exchange, examine trading volume, fee generation, liquidity distribution, and the execution quality available to traders.

For a yield vault, examine realized returns, strategy exposures, and whether performance relies on external subsidies.

A large liquidity pool that executes few trades may still provide useful market depth. A lightly utilized lending reserve may retain value as available liquidity.

Low utilization is therefore not automatically evidence of economic failure.

The analytical task is to identify the service being provided and measure whether users demand it.

Follow the Money Paid by Users

Deposited value is a balance-sheet measure. Fees and revenue are economic flows.

They answer different questions.

Under common DeFi analytics conventions, fees represent payments made by users, while protocol revenue generally represents the portion retained after payments to liquidity providers or other supply-side participants.

Consider another hypothetical protocol with $1 billion in average TVL and $1 million in monthly user fees.

If that monthly fee rate persisted for a year, it would generate $12 million in annual fees, equal to 1.2% of the average TVL.

Annualized fee-to-TVL intensity = $12 million / $1 billion = 1.2%.

This 1.2% figure is an illustrative measure of fee activity relative to deposited value, not a yield received by depositors or revenue earned by tokenholders.

The protocol might retain only part of those fees. It might also spend heavily on incentives, operations, or security.

Nevertheless, the comparison provides information that TVL alone cannot: how much measurable fee activity occurs relative to the capital deployed.

It should be compared only across business models where the underlying fee definitions and capital requirements are sufficiently similar.

For analysts, such comparisons are where a more complete assessment begins.

TVL describes scale. Capital flows reveal funding. Utilization and fees help explain demand. Revenue and costs reveal how much economic value the system retains.

None of those measures is sufficient on its own.

Together, however, they provide a substantially richer picture than a single headline TVL figure.

The Capital Behind the Number

TVL remains useful.

It can show how much value is exposed to a particular protocol, how deposits change over time, and how the scale of a financial service develops.

But its usefulness depends on keeping the measurement separate from the conclusions drawn from it.

A billion dollars in eligible contract assets is not automatically a billion dollars of independent external capital.

A billion dollars of capital exposure does not automatically mean a billion dollars of liquid, withdrawable funds.

And neither figure proves that a protocol generates sustainable revenue or creates value for its tokenholders.

DeFi makes financial capital composable. One position can support staking, collateralization, lending, and other services through successive layers of tokenized claims.

That composability can improve capital efficiency. It can also create interconnected risks and make aggregate figures difficult to interpret.

The task is not to reject TVL, but to reconstruct what stands behind it: the underlying assets, the claims built upon them, the liabilities attached to them, and the services they actually finance.

TVL tells you where value is accounted for. It does not, by itself, tell you how much unique capital exists or how economically useful that capital has become.