Skip to content
PROTOCOL RESEARCH

Uniswap risk

Immutable v2/v3 core with zero exploits across $2.75T of volume and real fee yield; the only blemish in the lineage is a small April 2020 reentrancy drain of a long-deprecated v1 pool. Residual risks are per-pool (v4 hooks, thin-pool TWAPs) and impermanent loss, not the core protocol.
SolidResearched Jul 3, 2026

Scorecard

strong: Audits & contractsv2/v3 core is immutable with zero exploits over 465M+ swaps; v3 audited by ABDK and Trail of Bits (verified in the v3-core repo); v4 shipped after nine audits, a $2.35M security competition, and a $15.5M bug bounty with no critical findings.
strong: Admin controlNobody can upgrade core pools or touch LP funds; the UNI DAO governs the fee switch behind a roughly 2-day timelock, and Uniswap Labs controls only the frontend, wallet, and API (interface fees zeroed since late Dec 2025).
strong: OracleThe core AMM consumes no external oracle; its TWAPs are manipulable on thin pools, which is a risk for integrators and hook pools, not for core LPs.
strong: Liquidity & exitDeepest DEX liquidity in DeFi ($2.94B TVL); LP withdrawal is permissionless and instant, no lockups.
watch: Yield (real vs emission)Fee yield is 100% real trading fees with zero emissions, but impermanent loss makes net LP yield position-dependent, and since Dec 2025 the fee switch clips roughly one-sixth of the LP fee on enabled pools.
watch: Holder concentrationTop-holder table unavailable this run (Etherscan Pro and Nansen both blocked, refresh); genesis gave team, investors, and advisors 40.0% exactly, fully vested since Sept 2024, and 100M UNI (~$596M) was burned in Dec 2025.
strong: Track recordLive since 2018 with over $2.75T of v2/v3 volume and no v2/v3 exploit across three cycles; v1 lost roughly $300k-$1.1M to an ERC-777 reentrancy in Apr 2020, and the 2025 hook incidents (Cork ~$11-12M, Bunni ~$8.4M) were third-party code on v4, not Uniswap itself.
🟢 strong🟡 watch / caveat🔴 weak / fund-loss risk
Verdict is a gate (worst flaw wins), not an average. Our read, not financial advice.

Live facts

auto-sourced now
TVL$1.42B
30d↓1%
Audits2
Last hacknone

DeFiLlama + our exploits feed. Cross-check the dated report against today.

TL;DR

Uniswap is the reference AMM. The v2 and v3 core contracts are immutable, which removes the single biggest smart-contract risk class: nobody, not Labs, not the DAO, can upgrade a pool or reach LP funds. Across more than $2.75 trillion of v2/v3 volume the core has never been exploited. The honest asterisk: Uniswap v1, long deprecated, lost roughly $300k to $1.1M in April 2020 when an attacker drained the imBTC pool via ERC-777 reentrancy, so "never hacked" is a v2/v3 claim, not an all-versions one. v4 keeps the clean record so far but changes the shape of the risk: a singleton contract plus permissionless hooks means every hooked pool carries its own third-party code, and hook projects (Cork, Bunni) already lost real money in 2025. Yield is real trading fees, no emissions, but impermanent loss decides whether your net is positive. Since December 2025 the UNIfication vote turned on protocol fees and a UNI burn, so governance now has a live economic lever over LP fee share.

Checklist

Audits & contracts. v2/v3 core is immutable and has facilitated over 465 million swaps with zero exploits (Uniswap's own claim, scoped explicitly to v2 and v3, and independent incident trackers do not contradict it). v3 was audited by ABDK Consulting and Trail of Bits (reports in the v3-core repo). v4 went through nine independent audits, a $2.35M security competition with 500+ participants, and a $15.5M bug bounty; no critical bugs were found and the bounty stays open. The caveat is structural, not audit coverage: v4 hooks are third-party code deployed per pool. A hooked pool is only as safe as its hook, exactly like an isolated Morpho market. Treat each hook as a separate protocol with its own audit status.

Admin control. There is no admin key over core pools. Governance is the UNI DAO, which controls the fee switch, treasury, and Unichain matters, executing through a timelock of roughly two days. The December 2025 UNIfication vote (passed Dec 25, 2025, with 125.3M UNI in favor vs 742 against) activated protocol fees on v2 and enabled v3 pools, and 100M UNI (~$596M at the time) was burned from the treasury after the timelock. Uniswap Labs removed all interface, wallet, and API fees in late December 2025 and contractually committed to align with the DAO entity. Protocol fees flow to a token-jar contract and are claimed by burning UNI through a firepit contract. Labs still controls the frontend, wallet, and API, which is a censorship and UX surface, not a fund-loss surface; the contracts remain usable without them.

Oracle. The core AMM needs no external price feed, which eliminates a whole failure class. Uniswap instead produces TWAP oracles that other protocols consume, and those are manipulable on thin pools. That history of manipulation losses sits with the integrators, not with Uniswap LPs. If you LP into a v4 pool whose hook adds its own oracle logic, you inherit that hook's oracle risk.

Liquidity & exit. $2.94B TVL and the deepest spot liquidity in DeFi. LP positions withdraw permissionlessly in one transaction, no cooldowns, no queues. Exit risk on majors is negligible; on long-tail pairs your exit price is your own pool's depth.

Yield: real vs emission. LP yield is trading fees only. No UNI emissions subsidize it, which puts Uniswap in the small honest-yield club: about $2.2M in daily fees, $43.3M over the last 30 days, $842M over the last year (DeFiLlama, pulled live). Two deductions apply. Impermanent loss can and often does exceed fee income on volatile pairs, so net LP yield is position-dependent, not a protocol constant. And since UNIfication, enabled pools route a protocol cut to the burn mechanism: 0.05% on v2 and reportedly around one-sixth of the LP fee overall, with the initial rollout covering pools that generate 80-95% of mainnet LP fees. The fee switch expanded to Optimism, Arbitrum, Base, Zora, and XLayer on March 8, 2026, and to Polygon, BSC, and Celo on June 2, 2026.

Holder concentration. Could not pull the on-chain top-holder table this run; Etherscan Pro is paywalled and Nansen returned 403 again, so the distribution is a refresh item. What is verifiable from the launch blog: genesis allocated 60.00% to the community, 21.266% to team, 18.044% to investors, and 0.69% to advisors (40.0% non-community), all on 4-year vesting completed September 2024. Today 621M of a 1B max supply circulates and total supply is down to about 893M after the 100M burn. The 2020 launch blog also scheduled 2% perpetual inflation starting after year 4; whether that was ever activated is a refresh item, though current supply figures suggest the burn dominates. UNI concentration threatens governance (who steers the fee lever), not LP principal.

Track record. v1 launched November 2018, v2 in May 2020, v3 in May 2021, v4 in January 2025. The full record: v1's imBTC pool was drained on April 18, 2020 via ERC-777 reentrancy (sources put the loss between $300k and $1.1M), a design incompatibility with a token standard that postdates v1; v2 and v3, where essentially all TVL has lived since, have zero exploits across three full market cycles. The 2025 incidents that get attributed to "Uniswap v4" were third-party protocols built on hooks: Cork Protocol lost $11-12M in May 2025 to missing access controls in its own hook contracts, and Bunni lost about $8.4M in September 2025 ($2.4M on Ethereum, $5.9M on Unichain) to a precision bug in its liquidity-accounting hook, then shut down permanently. News and X scans this run found no Uniswap incident in the last 90 days. Adoption is still compounding: Robinhood Chain launched mainnet on July 1, 2026 with Uniswap deploying a dedicated AMM as the chain's public liquidity venue (alongside a proprietary AMM from Pleiades).

Worst case

For a plain v2/v3 or hookless v4 position, the realistic worst case is not a hack. It is impermanent loss in a volatile pair grinding your net yield negative while headline APR looks fine, plus governance widening the protocol fee cut on your pool. For a hooked v4 pool, the worst case is real fund loss: a malicious or buggy hook drains or bricks the pool, and Cork and Bunni show this is not hypothetical. A compromised Labs frontend could also route users into bad approvals, which contract immutability does not protect against. There is no insurance fund in any of these scenarios.

Bottom line

Solid. The core protocol is about as safe as DeFi smart contracts get: immutable, battle-tested across cycles, real fees, no oracle dependency. The verdict applies to canonical v2/v3 pools and hookless v4 pools in deep pairs. Every v4 hook pool is a separate protocol wearing Uniswap's brand and must be assessed on its own. Size positions for IL, not for the fee APR.

Data appendix

  • TVL: $2.94B ($2,942,969,844; DeFiLlama, parent protocol "uniswap", 2026-07-03). 90d trend / ATH comparison: refresh.
  • Fees: $2.18M 24h, $9.25M 7d, $43.28M 30d, $842.1M 1y, $5.62B all time (DeFiLlama fees API, 2026-07-03). These are trading fees, mostly to LPs; post-UNIfication protocol cut on enabled pools reduces the LP share.
  • Audits: v3: ABDK Consulting and Trail of Bits (v3-core GitHub audits directory, verified this run). v4: nine independent audits, $2.35M security competition (500+ participants), $15.5M bug bounty, no critical findings (blog.uniswap.org/uniswap-v4-is-here, re-fetched this run).
  • Admin / governance: v2/v3 core immutable; v4 singleton with permissionless hooks. UNI DAO governs the fee switch and treasury behind a 2-day timelock. UNIfication passed Dec 25, 2025 (125,342,017 UNI YES vs 742 NO): protocol fees on (v2 plus v3 pools covering 80-95% of mainnet LP fees), interface/wallet/API fees removed late Dec 2025, 100M UNI ($596M) burned; fees flow to a token-jar contract claimed by burning UNI via the firepit contract. Fee switch expanded Mar 8, 2026 (Optimism, Arbitrum, Base, Zora, XLayer) and Jun 2, 2026 (Polygon, BSC, Celo). Protocol take: 0.05% on v2, roughly one-sixth of the LP fee per analyst coverage; exact v3 per-tier fractions: refresh.
  • Oracle: none consumed by core; produces TWAPs that are manipulable on thin pools (integrator-side risk). Source: Uniswap docs, incident history of TWAP-based exploits at other protocols.
  • Token: UNI $3.28, mcap $2.04B (rank 41), FDV $2.93B, 621.1M circulating, 893.2M total after the 100M burn, 1B max (CoinGecko, 2026-07-03).
  • Holder concentration: top-100 holder table: refresh (Etherscan Pro paywall, Nansen 403, both re-attempted this run). Genesis split 60.00% community / 21.266% team / 18.044% investors / 0.69% advisors, 4-year vesting completed Sept 2024 (Uniswap launch blog, re-fetched this run). 2% perpetual inflation clause from the launch blog: activation status refresh.
  • Incident record: v1 imBTC pool drained Apr 18, 2020 via ERC-777 reentrancy, loss $300k to $1.1M depending on source (PeckShield, ZenGo). v2/v3: zero exploits. Hook layer: Cork Protocol $11-12M (May 28, 2025, access-control flaw in Cork's own hook; Dedaub, Halborn, SlowMist), Bunni ~$8.4M (Sep 2, 2025, precision bug; shut down permanently Oct 2025; Halborn, The Block, CoinDesk). No Uniswap incident found in news or X scans in the last 90 days (2026-07-03).
  • Recent news: Robinhood Chain mainnet launched Jul 1, 2026; Uniswap deploying a dedicated AMM as its public liquidity venue, with Pleiades running a proprietary AMM alongside (Robinhood newsroom, The Block, The Defiant). June 2026 protocol revenue ~$5M: single X source, refresh; annualized protocol revenue estimates of roughly $27M added by the L2 expansion toward ~$61M total are analyst estimates (KuCoin coverage), treat as indicative.

Maintained monthly. Methodology: DeFi Research Instruction v2.

← all protocols