Crypto Lending Platforms: Borrowing and Lending in DeFi

DeFi lending is open access, but it is not broad-based credit

Crypto lending has split into two different businesses. Aave and Compound are onchain lending protocols where users post collateral and borrow against rule-based risk parameters. BlockFi was a custodial lender that took user assets onto a corporate balance sheet and managed credit centrally. Grouping them together under “crypto lending” hides the most important distinction: who controls underwriting, liquidation, and the spread between what lenders earn and borrowers pay.

DeFi lending is open in interface terms and narrow in credit terms. Aave describes borrowing as maintaining overcollateralized positions, and Compound III expands borrowing capacity through collateral factors on supplied assets. The Bank of Canada frames the same structure more formally: DeFi loans are decentralized, anonymous, overcollateralized, and governed by rigid contract terms. That removes discretionary underwriting. It also means the system mostly serves users who already own the collateral base.

As of March 11, 2026, the market has voted decisively for Aave over Compound on scale. DefiLlama showed Aave at $27.343 billion in TVL and $16.744 billion borrowed, versus Compound Finance at $1.435 billion in TVL and $533.81 million borrowed. That puts Aave at roughly 19.1x Compound’s TVL and 31.4x its active borrow base. Scale matters because it improves liquidity depth and liquidation throughput, but it also increases the stakes of governance and oracle design.

Borrowing and lending in DeFi is a utilization market, not a negotiated credit market

Lender yield in DeFi comes from algorithmic liquidity pricing. Aave says supplied tokens earn at the current market supply rate, with rates determined by borrow utilization and governance parameters. Compound III uses the same core logic more explicitly: supply and borrow rates are functions of utilization, with a “kink” where rates rise faster above a threshold. Borrowers are not paying for credit scoring or relationship lending. They are paying for liquidity under a parameterized curve.

The design choice has distributional consequences. On Aave, supplied assets accumulate interest while remaining usable as collateral if enabled. In Compound III, users with a positive balance of the base asset earn interest, while collateral assets do not earn or pay interest. That makes Compound III simpler and often cleaner from a risk-engineering perspective, but it narrows where yield accrues and reinforces the importance of market design around the single base asset.

Liquidation is the hard edge of the model. Aave’s health factor is calculated from collateral value, liquidation thresholds, and total borrow value, and a position becomes liquidatable when the health factor falls below 1. Depending on how far the position has deteriorated, up to 50% or 100% of debt can be liquidated. Compound III uses separate liquidation collateral factors and lets a liquidator call an absorb function that takes the account’s collateral and returns value, net of penalty, in the base asset. These mechanisms are efficient. They are also explicitly pro-cyclical.

The fairness implication is simple. DeFi lending gives broad access to a market structure, not broad access to credit creation. If borrowing requires existing collateral, then economic participation still tilts toward early token holders, whales, treasury operators, and users with large spot balances. The interface is permissionless. The balance-sheet prerequisite is not.

Aave, Compound, and BlockFi solve different problems and expose different power structures

BlockFi belongs in the comparison because it shows what disappears when lending leaves the chain. The SEC said on February 14, 2022 that BlockFi would stop unregistered offers and sales of BlockFi Interest Accounts, and that it had operated for more than 18 months as an unregistered investment company while deploying customer crypto through institutional lending and other investments. That model gave users yield without governance. When the structure broke, users became claimants.

Ownership structure determines who actually governs the lending rails

Aave’s starting point is more community-inherited than founder-reset. Aavenomics set total supply at 16 million AAVE, with 13 million redeemed by LEND holders and 3 million allocated to the Ecosystem Reserve. That means 81.25% came through migration and 18.75% sat in the reserve. Voting power is proportional to AAVE, stkAAVE, or aAAVE holdings or delegation, and Aave’s governance process document states a 320,000-AAVE quorum requirement for votes. The upside is continuity with a pre-existing holder base. The trade-off is that treasury reserves and delegated blocs still shape power.

Aave is also moving toward direct economic participation for token holders. DefiLlama showed annualized holder revenue of $78.09 million on March 11, 2026, and its methodology states that holder revenue begins after the DAO started buying back AAVE on April 9, 2025. That is a meaningful shift because it routes protocol value toward token holders instead of leaving governance as a pure control token. But execution is not fully atomized. A 2025 Aavenomics implementation proposal created a four-member committee with a 3/4 signature threshold for parts of the treasury and liquidity program. That can improve speed and professionalism. It is still delegated power.

Compound’s origin is more evenly split, but also more visibly insider-aware. When governance launched, Compound Labs disclosed 2,396,307 COMP to shareholders, 2,226,037 to founders and team, 372,707 to future team members, and 4,229,949 reserved for users. A later update added 500,000 COMP for Coinbase Earn and 275,000 COMP to the Reservoir, bringing community-directed supply to 5,004,949 COMP and insider-plus-future-team supply to 4,995,051 COMP. That is almost a 50/50 split. The model is more balanced than many token launches, but it is still a reminder that “community governance” often begins with large pre-positioned blocs.

Compound’s current economics are weaker from a participation standpoint. Governance proposals require more than 25,000 COMP to submit and 400,000 votes for quorum, and DefiLlama shows zero holder revenue for COMP as of March 11, 2026. So token holders govern risk parameters and upgrades, but the token does not currently offer the same visible cash-flow linkage that Aave now does. Builder incentives may be cleaner when tokens are not used as yield claims. The cost is thinner alignment for smaller holders who are asked to govern without participating directly in protocol revenue.

The risk stack is larger than “smart contract risk”

Market volatility is the first risk, not the last. The BIS finds wallet leverage on major DeFi lending platforms typically ranges from 1.4 to 1.9, and that higher leverage increases the share of debt near liquidation. The Bank of Canada models a price-liquidity feedback in which rigid smart contracts can cause lending activity and asset prices to move with sentiment in self-fulfilling ways. The Financial Stability Board adds that DeFi can amplify familiar vulnerabilities through automatic collateral liquidation, leverage, and structural interconnectedness.

Code risk remains real even for blue-chip protocols. DefiLlama’s hack database lists Compound V2’s September 29, 2021 incident as a $147 million protocol-logic exploit. That does not mean DeFi lending is uniquely fragile. It means users should stop treating audits, age, and TVL as substitutes for adversarial review of smart contract vulnerabilities. Large protocols tend to fail less often because their systems are better scrutinized, not because code risk disappears.

Oracle risk is a governance risk wearing technical clothing. S&P Global argues that evaluating smart-contract risk also means evaluating oracle concentration, data quality, and technical risk. Aave’s own governance materials note that the protocol relies on oracles to price collateral and debt for normal operation. If oracle concentration or data quality is poorly governed, token holders may formally run the protocol while a smaller set of data and node operators effectively governs liquidation outcomes.

Custody and legal structure can dominate every other risk when lending is centralized. BlockFi’s history made that plain. The company filed Chapter 11 on November 28, 2022, its plan became effective on October 24, 2023, and even later distributions required coordination through Kroll, Digital Disbursements, and Coinbase. The lesson is not that CeFi is always inferior. The lesson is that opaque creditor hierarchy and discretionary asset deployment create a different and usually less legible fairness profile than transparent onchain collateral rules.

What matters for token design is not APY. It is the distribution of control, losses, and upside

The strongest DeFi lending protocols do four things clearly. They specify who gets initial ownership, who captures recurring spread, who absorbs bad debt, and who can change parameters during stress. Aave has improved the second bucket by routing visible value toward holders through buybacks, but that makes concentration analysis more important, not less important. Compound remains cleaner in some respects, especially in Compound III’s simpler market structure, but it asks token holders to govern without current holder revenue. BlockFi demonstrated the weakest version of the stack: users provided assets but did not control credit policy, liquidity management, or bankruptcy timing. That is the core of token economy design.

For teams building lending protocols, the hard question is not whether the protocol can generate yield. The hard question is whether the token economy distributes that yield, the associated governance rights, and the crisis-era powers in a way that remains legitimate after the founding team loses its moral monopoly. At FinDaS Tokenomics, that is where tokenomics consulting becomes more useful than headline APY modeling: allocation, quorum, reserve-factor routing, and emergency control rights determine whether a lending token can mature into infrastructure instead of reverting into a thinly decentralized franchise.