Curve DAO (CRV)

Curve Finance operates as an automated market maker optimized for assets that are expected to trade near a fixed exchange rate with one another, such as different U.S. dollar–pegged stablecoins (e.g., USDC, USDT, DAI) or wrapped versions of the same underlying asset (e.g., wrapped Bitcoin variants).¹ By combining features of constant-sum and constant-product market-maker formulas, Curve's underlying "StableSwap" invariant reduces slippage and impermanent loss for liquidity providers trading between closely correlated assets, relative to general-purpose AMMs such as Uniswap.²

Governance of the protocol is exercised by holders of CRV who lock their tokens for a period of up to four years in exchange for vote-escrowed CRV (veCRV), a non-transferable representation of voting power and a claim on a share of protocol trading fees.³⁴ The DAO's decisions determine which liquidity pools receive CRV emissions, how protocol fees are distributed, and what upgrades or integrations the protocol pursues.⁵

History

Origins and founding

Ticker	CRV
Category	Decentralized Exchange (DEX)
Website	https://www.curve.fi/
Twitter	@curvefinance
Telegram	curvefi
Contract Addresses
ethereum	0xd5...52Copied!
optimistic-ethereum	0x09...53Copied!
polygon-pos	0x17...afCopied!
fantom	0x1e...8bCopied!
arbitrum-one	0x11...78Copied!
energi	0xd3...9eCopied!
sora	0x00...d0Copied!

Curve Finance was created by Michael Egorov, a Russian-born physicist and software engineer.⁶ Egorov held a background in condensed-matter and quantum physics, having worked as a postdoctoral researcher at Monash University in Australia, and had previously worked as an infrastructure engineer at LinkedIn.⁷ In 2015, he co-founded NuCypher, a data-encryption and proxy re-encryption company that went through the Y Combinator startup accelerator, and he served as NuCypher's chief technology officer into 2020.⁸⁷ According to Egorov, his work on staking mechanisms and bonding curves for NuCypher's token led him to explore concepts that would later underpin Curve's AMM design.⁶

In November 2019, Egorov published the "StableSwap" whitepaper, which described a new AMM invariant designed specifically to minimize slippage and price impact when swapping between assets intended to maintain a stable exchange rate with one another.²⁶ The Curve Finance protocol, then commonly referred to simply as "Curve.fi," launched on the Ethereum mainnet in January 2020.⁹⁷ For the first several months, Egorov was reportedly the only person actively working on the project.⁷ The protocol quickly gained traction among traders and liquidity providers because its specialized formula offered materially lower slippage on stablecoin and pegged-asset trades than generalized AMMs of the period.²

Launch of the DAO and CRV token

In August 2020, at the height of that year's "DeFi Summer," Curve's governance token and DAO smart contracts were deployed.² Notably, according to Curve's own historical account, the contracts were not deployed by the core Curve team but by an anonymous community member using the handle "0xc4ad," who used code that had already been made public so that it could be reviewed by the community ahead of the token's release.² The Curve DAO's governance infrastructure was built using Aragon, an Ethereum-based toolkit for creating and managing on-chain organizations, which Curve used to connect the various smart contracts governing user-deposited liquidity.¹⁰¹¹

The CRV token's distribution was structured with a maximum supply of approximately 3.03 billion tokens, the majority of which (roughly 62%) was earmarked for community liquidity providers over time, with the remainder allocated to early shareholders (30%, subject to a two-year vesting schedule), employees (3%, also vested over two years), and a community reserve (5%).¹¹ There was no pre-mine, and the circulating supply was zero at the token's launch, with new tokens released gradually through a liquidity-mining emissions schedule; roughly 750 million tokens were expected to be in circulation approximately one year after launch.¹¹ The initial CRV emissions rate began at around 274 million tokens per year, declining by approximately 16% each year thereafter.¹²

Shortly after the DAO launched, Egorov drew attention for his own governance position: in a move he later described as an "overreaction" to the voting power that the yield-aggregation protocol yearn.finance had accumulated, he locked a large amount of CRV, which at the time gave him control of roughly 71% of the protocol's governance voting power.¹¹⁹ This concentration of voting power in the hands of the founder became a recurring point of discussion regarding the DAO's practical degree of decentralization.

Growth and expansion

Following its launch, Curve became one of the largest protocols in DeFi by total value locked (TVL), at times exceeding $1 billion, and it developed close relationships with other major protocols during the "Curve Wars," a period in which rival platforms competed to accumulate CRV and veCRV in order to direct Curve's liquidity-mining emissions toward pools that benefited them.⁹ Protocols such as Convex Finance emerged specifically to aggregate users' CRV locking and voting power, becoming significant forces in Curve governance in their own right.

Over time, Curve expanded beyond its original Ethereum mainnet deployment to other EVM-compatible chains, including via bridging arrangements to networks such as Avalanche and Polygon.³ The DAO also broadened Curve's product suite considerably. In 2023, Curve introduced crvUSD, an overcollateralized, crypto-backed stablecoin that uses a "lending-liquidating AMM algorithm" (LLAMMA) designed to gradually and continuously liquidate under-collateralized positions rather than triggering abrupt liquidations, thereby aiming to reduce liquidation risk for borrowers.¹ The protocol later added FXSwap-style pools intended to facilitate trading between crypto assets and pairs referencing traditional foreign-exchange markets.¹

The July 2023 exploit

On July 30, 2023, several of Curve's stablecoin- and pegged-asset liquidity pools were exploited due to a vulnerability in specific older versions (0.2.15, 0.2.16, and 0.3.0) of Vyper, the Python-derived smart-contract programming language used to write many of Curve's contracts.¹³¹⁴ The flaw affected the compiler's handling of reentrancy locks, meaning that pools written in the affected Vyper versions were vulnerable to reentrancy attacks despite appearing, on the surface, to have reentrancy protections in place.¹⁴

The attacks began with an $11.5 million exploit of the pETH-ETH pool associated with the NFT-lending protocol JPEG'd, and a wave of related exploits followed in rapid succession across other protocols that used the vulnerable Vyper versions, including Alchemix DAO's alETH-ETH pool (approximately $20–22 million), Metronome DAO's sETH-ETH pool (approximately $1.6 million), and Curve's own CRV/ETH pool, from which Egorov confirmed that roughly 32 million CRV tokens, worth more than $22 million at the time, had been drained.¹⁵¹⁶¹³ Total losses across the affected pools and protocols were estimated by various analysts at between roughly $61 million and $73.5 million.¹⁷¹⁸

The incident also triggered a rapid, unrelated financial contagion risk. At the time of the hack, Egorov personally held loans across several DeFi lending protocols—including Aave, Fraxlend, and Abracadabra—collateralized by very large holdings of CRV, reportedly around 427.5 million CRV tokens, or approximately 47% of the token's circulating supply, backing roughly $100 million in borrowed funds.¹⁹¹⁸ As CRV's price fell by close to 30% in the aftermath of the hack, market participants grew concerned that Egorov's positions could face liquidation, which in turn could have flooded the market with CRV and caused further price declines or "bad debt" for the lending protocols involved.¹⁵¹⁷ To reduce his liquidation risk, Egorov sold roughly 39.25 million CRV tokens over the counter, at a discounted price of about $0.40 per token, to a group of prominent investors, including Justin Sun, Machi Big Brother, and DWF Labs, raising approximately $15.8 million.¹⁵

In the hours after the exploit, so-called maximal extractable value (MEV) actors played an unusually prominent role. Some MEV bot operators, most notably one known by the pseudonym "c0ffeebabe.eth," front-ran the malicious transactions of the original attackers on several of the affected pools and subsequently returned recovered funds to the affected projects, acting in the manner of "white hat" rescuers.¹⁶¹⁹¹⁷ Separately, some individuals published technical details of the underlying Vyper vulnerability publicly while the exploit was still unfolding, a decision that was criticized within the community for potentially inviting further "copy-cat" attacks; a similar exploit was subsequently replicated against pools on the BNB Smart Chain.²⁰²¹ Within roughly a week of the initial exploit, a substantial share of the stolen funds—reported at around 73%, or about $52.3 million—had reportedly been recovered or returned.¹⁸

Governance and tokenomics

Vote-escrowed CRV (veCRV)

Central to Curve's governance system is its vote-escrow model. CRV holders may lock their tokens for a period of between one week and four years in order to receive veCRV, a non-transferable token that represents both governance voting power and a claim on a portion of the protocol's trading-fee revenue.³ The amount of veCRV received for a given quantity of locked CRV is proportional to the length of the lock: locking tokens for the maximum four-year period yields the greatest voting power and fee-sharing weight, while shorter locks yield proportionally less, and that voting power decays over time as the lock approaches its expiration.³

veCRV serves several interconnected purposes within the ecosystem. It grants holders the ability to vote on proposals, including which liquidity pools should receive CRV emissions, changes to fee structures, and other protocol-level decisions.⁵ Locking CRV to obtain veCRV also entitles holders to a "boost" of up to 2.5 times on the CRV rewards earned from providing liquidity to Curve's pools, incentivizing long-term alignment between liquidity providers and the protocol.³¹¹ In addition, veCRV holders are entitled to a share of the trading fees generated by the protocol; historically, roughly half of Curve's trading fees have been distributed to veCRV holders, either directly or by conversion into other assets.¹¹

CRV emissions and inflation

CRV's supply schedule was designed so that annual emissions decline over time, mirroring the diminishing issuance schedules used by some other proof-of-work and DeFi tokens.¹² Emissions began at roughly 274 million CRV per year in 2020 and have decreased by approximately 16% annually since.¹² According to Curve's own resources, a further tokenomics adjustment took effect in August 2025, when the DAO voted to reduce the protocol's annual CRV inflation rate to approximately 5.02%, a change reported to cut new annual token issuance by roughly 22 million tokens and reduce ongoing sell-pressure associated with continuous emissions.⁵

In a separate governance decision in June 2025, the DAO voted to direct 10% of all protocol revenue toward a dedicated treasury, described as the first time Curve had formally reserved a portion of protocol income for this purpose. The treasury was intended to provide the DAO with a financial reserve to fund audits, ongoing development, and other protocol needs.⁵

Multi-chain expansion

Beyond Ethereum, the DAO has pursued deployment of a lighter-weight version of the Curve exchange, referred to as "Curve-Lite," on a range of additional EVM-compatible blockchains. Following initial deployments on Monad and various OP Stack-based chains in 2025, Curve's DAO and development teams have targeted further deployments on chains such as Scroll, zkSync, and Polygon CDK-based networks, with the stated goal of expanding CRV's utility and emissions-driven demand across a broader set of ecosystems.⁵

Market data and token supply

CRV is issued as an ERC-20 token on Ethereum and has, through bridging arrangements, been made available on other networks such as Avalanche and Polygon.³ As of early 2026, CRV's maximum token supply stood at approximately 3.03 billion tokens, with a circulating supply on the order of 1.5 billion tokens.²²²³ The token has traded on hundreds of markets and exchanges, with daily trading volumes in the tens of millions of dollars, and it has generally ranked among the top several hundred cryptocurrencies by market capitalization.²⁴¹¹

Because a large share of CRV's total supply remains subject to ongoing emissions and vesting, CRV's fully diluted valuation (a hypothetical measure of market capitalization assuming the full maximum supply were in circulation) has typically been substantially higher than its circulating market capitalization, a common feature of tokens with long emissions schedules.²³

Governance criticisms and centralization concerns

Curve DAO has, since its founding, drawn recurring scrutiny over the degree to which governance power has been concentrated among a small number of large holders. Michael Egorov's early decision to lock a substantial share of the CRV supply gave him outsized control over DAO votes relative to the broader base of token holders, a dynamic that a number of commentators have pointed to as an example of the gap that can exist between a protocol's nominal decentralization and the practical distribution of its governance power.¹¹⁹ The so-called "Curve Wars," in which rival protocols and DAOs — most prominently Convex Finance — competed to accumulate CRV, veCRV, and voting influence over Curve's emissions in order to direct rewards toward pools benefiting their own users, further illustrated how governance power over Curve's liquidity incentives became a valuable and actively contested resource in its own right.

The July 2023 exploit also renewed scrutiny of governance-related risk at Curve, given that Egorov's personal, heavily CRV-collateralized borrowing positions across multiple external DeFi lending protocols created a channel through which a security incident affecting the Curve protocol could, in principle, have produced cascading financial stress across the wider DeFi lending ecosystem.¹⁹¹⁷ Some commentary following the incident also noted broader concerns about the extent to which smart-contract compilers such as Vyper had, at the time, received less rigorous security review than the application-level contracts built on top of them.¹⁹

Products built on Curve DAO governance

crvUSD

crvUSD is a stablecoin issued by the Curve ecosystem and governed through the same veCRV-based DAO structure that governs the exchange itself. Users generate crvUSD by depositing collateral, such as ETH or other approved crypto assets, into a lending market that employs Curve's LLAMMA mechanism. Rather than liquidating an under-collateralized position in a single transaction, LLAMMA is designed to convert a borrower's collateral into crvUSD incrementally as the collateral's value declines relative to the debt, and to convert it back if the collateral's value recovers, with the stated aim of smoothing liquidation risk for borrowers relative to conventional, single-event liquidation designs used elsewhere in DeFi.¹

FXSwap

FXSwap pools represent a further extension of Curve's core low-slippage AMM design, aimed at facilitating trading between crypto assets and pairs referencing more traditional foreign-exchange-style markets, such as crypto-to-fiat-tracking pairs.¹ This is presented by Curve as part of a broader strategic effort to extend the protocol's stable-asset trading expertise into a wider range of markets beyond conventional dollar-pegged stablecoins.¹

The "Curve Wars"

Because veCRV holders determine how CRV emissions are distributed across Curve's many liquidity pools, and because those emissions directly affect the yields available to liquidity providers, control over veCRV voting power became commercially valuable to any protocol whose own token or stablecoin depended on deep, low-slippage Curve liquidity. This dynamic gave rise to what market participants and commentators came to call the "Curve Wars." Rather than accumulating CRV and locking it directly, a number of competing protocols instead built dedicated "meta-governance" platforms that pooled users' CRV, locked it centrally for the maximum four-year term, and then allowed depositors to direct the resulting voting power, most prominently through the protocol Convex Finance. Over time, Convex itself came to control a very large share of all outstanding veCRV, effectively becoming a second, influential layer of Curve governance sitting alongside the DAO's own individual token-holder base. Other protocols, including Yearn Finance and various stablecoin issuers, similarly sought to accumulate voting influence, at times through direct token swaps, bribery-style incentive markets, or acquisitions of governance-focused competitors. The resulting competition illustrated how, in vote-escrow-based governance systems more broadly, voting power can become a tradable commodity in its own right, with implications for how genuinely decentralized such systems are in practice.

Reception and industry role

Curve DAO and the broader Curve Finance protocol are frequently cited in industry commentary as central infrastructure for DeFi's stablecoin markets, given their role in providing a venue for efficient conversion between different dollar-pegged assets, a function that other DeFi protocols, market makers, and centralized exchanges have relied upon for pricing and arbitrage.¹¹ The protocol's vote-escrow governance model, first popularized at scale by Curve, has since been adopted or adapted by numerous other DeFi projects seeking to align long-term token holders with protocol decision-making, a design pattern sometimes referred to generically as "ve-tokenomics."

At the same time, industry observers have periodically highlighted governance-related risks associated with Curve's structure, including the influence exerted by a small number of large veCRV holders and meta-governance platforms, the reliance of the protocol's stability on the personal financial positions of its founder, and the operational risks inherent in smart-contract systems written in less widely audited languages such as Vyper.¹⁹⁹

References