Skip to content
PROTOCOL RESEARCH

Maple risk

Maple is institutional onchain credit: real yield, Tier-1 audits, no protocol exploit, and this run the protocol's own contracts confirm zero recognised losses and a near-empty withdrawal queue, so the ~6% TVL decline is not credit stress. It stays a caution because you underwrite borrower credit and in-house underwriting, the pools are 96% lent out behind a documented 30-day redemption ceiling, and the collateral custody path is still not publicly documented.
CautionResearched Aug 5, 2026

Scorecard

strong: Audits & contractsSeven distinct Tier-1 firms verified this run against dated report links in maple-labs GitHub: Trail of Bits (2022-08), Spearbit (2022-10), Three Sigma (2023-04), Spearbit auditors via Cantina (2023-06), Sherlock and 0xMacro (both 2025-09, Governor Timelock), Dedaub (2025-11) and Sigma Prime (2026-01, cross-chain receiver). Cantina is the platform Spearbit delivered through, not an eighth independent firm. Immunefi bounty active; Maple appears in 0 of 612 records in DeFiLlama's hacks database; no protocol-level exploit. The February 2026 incident was a front-end compromise, not a contract break.
watch: Admin controlMaple Direct, Maple's own lending arm, is the delegate for the syrup pools, so underwriting is in-house judgement; term sheets are agreed offline and signed, then Maple proposes the loan on-chain. Governor is an on-chain GovernorTimelock held by Maple-managed multisigs; the Security Admin can pause every function and holds upgrade on the loan and manager contracts. Governor, Operational Admin and the delegate can process, remove and manually settle withdrawal requests. Maple's own assumption list states outright that pool delegates are trusted not to prevent withdrawals. Signer sets and thresholds still unverified.
watch: OracleDowngraded from green on evidence, not on doubt. Chainlink feeds, oracle wrappers against outage and manipulation during liquidations, withdrawal cooldowns and bootstrapMint were all re-verified in the docs this run, and no oracle incident occurred in the window. But the Governor's Globals permissions include setPriceOracle and setManualOverridePrice, so the price that drives liquidations is admin-replaceable and admin-overridable, and margin calls actually run off a proprietary three-source feed rather than the on-chain oracle.
watch: Liquidity & exitQueue state measured directly this run and it is calm: the syrupUSDC withdrawal queue holds 516,506 shares, about $608K or 0.055% of the pool, and the syrupUSDT, syrupUSDG and High Yield Secured queues are empty. The structural limit is unchanged and now quantified: the four material pools are 95.9% deployed into loan principal, leaving $82.8M of non-loan residual across $2.02B, and Maple's docs cap a redemption at 30 days on a FIFO queue. Uniswap and Balancer offer an instant exit at a discount.
strong: Yield (real vs emission)Independently corroborated. DeFiLlama's yields endpoint reports syrupUSDC 4.96%, syrupUSDT 4.14% and syrupUSDG 5.02% with apyBase equal to the full APY and apyReward exactly zero on all three, matching Maple's own 5.0/4.1/5.0 and its GraphQL weeklyApy of 4.96095% to five decimals. This is institutional loan interest on a 0.85% net interest margin, not emissions. SYRUP is a separate governance token from syrup principal.
watch: Holder concentrationSYRUP at $0.1554, about $193.4M market cap, 1.167B of 1.245B supply circulating, 76.2% below its June-2025 ATH of $0.6532 (CoinGecko, verified this run; total supply cross-checked on-chain at exactly 1,244,677,492.511099). Down 5.5% over 30 days, up 21.5% over 60. Wallet-level distribution still not pulled: the Etherscan top-holders endpoint requires a Pro key. Borrower-side concentration in the syrup pools is also undisclosed.
watch: Track recordThe 2022 Orthogonal Trading $36M default (~80% loss to remaining M11 USDC pool investors) remains real history under the old unsecured third-party-delegate model. Nothing new joined the loss record in the window, and every Maple pool reads unrealizedLosses of exactly $0 today. Two documented stress tests held. Not in the prior report and worth knowing: Core Foundation obtained a Cayman Grand Court injunction against Maple in November 2025 over the lstBTC partnership, settled and withdrawn on 22 May 2026 with no admission of liability.
🟢 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$2.47B
30d↑9%
Audits2
Last hacknone

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

TL;DR

The headline number moved down and the story did not. TVL is about $2.302B against roughly $2.446B at the start of July, a fall of about 6% and $147M. Maple's own reported AUM slid on the same slope, from $4.6B at the H1 close to $4.36B today, so the shrinkage is real and not a DeFiLlama artifact. What it is not, and this run I can say so from protocol state rather than from the absence of bad news, is credit stress. Every Maple pool reports unrealizedLosses of exactly zero. Loan principal outstanding across the four material pools is $1,932.6M right now, consistent with the record $1.9B Maple reported at quarter end, so the loan book has not shrunk at all. The syrupUSDC withdrawal queue holds about $608K against a $1.11B pool and the syrupUSDT, syrupUSDG and High Yield Secured queues are empty. The daily series swings 6 to 17 percent in a single day as large short-duration loans mature and get redeployed, and the 1 July reading happened to land three days before a local peak.

So the month's question has a clean answer, and the verdict still does not move. What you are underwriting is unchanged: borrower credit, one team's in-house judgement, a collateral custody path that is not publicly documented, and a pool that is 96% lent out behind a contractual exit the docs cap at 30 days.

Checklist

Audits & contracts. Seven distinct Tier-1 firms, verified this run against dated report files in Maple's own GitHub rather than taken on trust from a marketing page: Trail of Bits (August 2022), Spearbit (October 2022), Three Sigma (April 2023), Spearbit auditors via Cantina (June 2023), Sherlock and 0xMacro (both September 2025, on the Governor Timelock), Dedaub (November 2025) and Sigma Prime (January 2026, on the cross-chain receiver). Two small corrections to the roster as it has been printed before. Cantina is the platform through which Spearbit auditors delivered the June 2023 review, so it is not an eighth independent firm. And the coverage runs through January 2026, not "through 2025". The Immunefi bounty is active, the delegate fund-drain vector caught pre-mainnet in the Spearbit review stays bounded in code, and Maple appears in zero of the 612 records in DeFiLlama's hacks database. The February 2026 event was a front-end compromise of the website, in the same wave that hit OpenEden and Curvance. The contracts were not touched. Whether individual users lost funds by signing a malicious prompt is not something I could establish either way, and that is the honest version of the "funds unaffected" line that has circulated.

Admin control. This is sharper than it has been described. The delegate question has an answer: Maple Direct, Maple's own lending arm, underwrites, structures and manages the loans behind syrupUSDC, syrupUSDT and syrupUSDG. There is no third-party delegate marketplace any more. That is a genuine improvement on 2022, when outside delegates ran pools and one borrower was also a delegate, but it concentrates rather than removes the exposure: you are trusting one credit committee, and the term sheet is agreed offline and signed before Maple proposes the loan on-chain, so the underwriting is not visible to you at any point.

The on-chain powers are worth naming precisely. The Governor is an on-chain GovernorTimelock whose proposer, executor and role-admin roles are held by Maple-managed multisigs, with a security and ops multisig holding the canceller role. Its Globals permissions include setPriceOracle, setManualOverridePrice, the three platform fee rates, setValidBorrower, setValidPoolDelegate and setProtocolPause. A separate Security Admin can trigger a pause that the docs say "can pause every function in the protocol", and holds upgrade on the loan and loan-manager contracts. On the withdrawal queue, the Governor, the Operational Admin and the pool delegate can all call processRedemptions, removeRequest and setManualWithdrawal.

The most useful thing Maple publishes here is its own list of assumptions, which is unusually candid. Assumption 3 is that lenders "are inherently trusting a Pool Delegate with their funds". Assumption 5 is that "the Governor & Operational Admin are trusted actors". Assumption 7 is that "Pool Delegates don't try to prevent withdrawals", and it concedes that in theory a withdrawal manager could be configured with a one-second window, an oversized cycle, or an invalid address. Those are trust assumptions, not code guarantees, and Maple says so itself. Exact multisig signer sets and thresholds remain unverified.

Oracle. Downgraded to yellow on evidence, not on doubt. I re-pulled the stack this run rather than carrying it forward, and it is all there: Chainlink feeds, oracle wrappers built to stop outages and manipulation from causing problems during liquidations, withdrawal cooldowns against flash-loan front-running, and a Governor-set bootstrapMint against first-depositor front-running. No oracle failure or mispricing appeared anywhere in the window. What pulls it off green is that the Governor's permission list includes both setPriceOracle and setManualOverridePrice, so the price feeding the liquidation module is replaceable and manually overridable by an admin role, timelocked but discretionary. Separately, the margin calls that actually protect your principal run off Maple's proprietary three-source alert system with 24/7 monitoring, not off the on-chain oracle, so the operational layer sits outside what the audits cover.

Liquidity & exit. The dimension that defines the verdict, and now with numbers instead of adjectives. The docs are explicit that syrup assets normally have instant liquidity, that in rare cases a withdrawal takes around 24 hours, and that "the maximum possible time being 30 days". A developer-facing page in the same corpus puts it more conservatively at "typically less than 2 days" and up to 30 days during low liquidity. Redemptions run FIFO through requestRedeem; requestWithdraw and withdraw are disabled. You can cancel a pending request and get your shares back, and Maple Direct reports processing withdrawals in under 24 hours on average.

Today the queue is calm and I measured it rather than assuming it. The syrupUSDC queue holds 516,506 shares, about $608K at the current 1.1769 share price, or 0.055% of shares outstanding against a $1.11B pool. The syrupUSDT, syrupUSDG and High Yield Secured queues are completely empty. What keeps this yellow is the structure behind that calm: across the four material pools, $1,932.6M of $2,015.3M is out as loan principal, leaving $82.8M, about 4.1%, as non-loan residual. On syrupUSDC specifically the residual is $38.7M against $1.11B, about 3.5%, and that residual also has to cover accrued interest and the supporting basis-trade and DeFi-liquidity strategies, so the genuinely instant cash is smaller still. Instant liquidity is true for small tickets and stops being true somewhere well below 4% of the pool. Beyond that you are waiting on loan maturities, and the 30-day clause is what governs.

One honest counterpoint on the TVL comparison. Across this five-week window total DeFi TVL rose 8.3% while Maple fell 6.0%, so Maple lagged the sector by about 14 points. The likely mechanical reason is that sector TVL is largely denominated in volatile collateral that appreciated, and a stablecoin-denominated credit book cannot participate in a price rally. I did not decompose sector TVL into price versus flow, so treat that as the probable explanation rather than a proven one.

Yield: real vs emission. Real credit yield, modest, and independently corroborated this run. DeFiLlama's yields endpoint reports syrupUSDC at 4.96%, syrupUSDT at 4.14% and syrupUSDG at 5.02%, with apyBase equal to the full APY and apyReward exactly zero on all three. Maple's transparency dashboard shows 5.0%, 4.1%, 5.0% and 5.2% on Maple Institutional, and its GraphQL API returns a syrupUSDC weekly APY of 4.96095%, matching DeFiLlama to five decimal places. Three independent readings agree, and none of the yield is emissions. The net interest margin is 0.85%. July's MIP-021 vote replaced discretionary SYRUP buybacks with a rules-based mechanism taking 10% of monthly revenue below $1.5M per month and scaling to 30% above $2M, which affects the governance token and not your principal.

Holder concentration. SYRUP trades at $0.1554 for about $193.4M of market cap, with 1.167B of 1.245B supply circulating and the price 76.2% below its June 2025 ATH of $0.6532. It is down 5.5% over 30 days and up 21.5% over 60. The total supply checks out exactly against the on-chain contract at 1,244,677,492.511099 tokens. This is governance-token data and says little about pool risk. Wallet-level distribution was again not pulled: the Etherscan top-holders endpoint requires a Pro subscription this environment does not have. Borrower-side concentration inside the syrup pools is not disclosed by Maple at all, and that is the concentration number that would actually matter.

Track record. The scar is unchanged and still worth stating plainly. In December 2022 Orthogonal Trading defaulted on about $36M after concealing FTX exposure, costing remaining investors in the affected M11 USDC pool roughly 80 percent, and Auros missed a payment in the same wave. That was the old unsecured, externally-delegated model. Be careful with search results here, because at least one aggregator article pairs that $36M figure with the February 2026 website incident in a way that reads like a fresh default. It is not.

Since then, two documented stress tests. On 2 February 2025, 15 margin calls across the book, zero liquidations, roughly 165% average collateralization as of 6 February, $7.4M of extra collateral posted and $7.4M of loans repaid early. On 10 October 2025, the largest liquidation event on record, nine margin calls all cured within three hours, zero liquidations and zero losses, and $67M of lender redemptions processed through the turbulence, with syrupUSDC touching 136% collateralization at its low and syrupUSDT 140%. Maple's claim of zero credit losses since 2023 is its own, traced to a June 2024 page about the Secured Lending arm launched in August 2023 and repeated since; there is no third-party attestation of the loan book, so read it as a well-corroborated management claim rather than an audited fact. What is independently verifiable is today's protocol state, and it is clean: unrealizedLosses is exactly zero on Syrup USDC, Syrup USDT, syrupUSDG and High Yield Secured Lending.

One item belongs here that has been missing from this report. In November 2025 the Grand Court of the Cayman Islands granted Core Foundation an injunction against Maple entities, blocking the launch of syrupBTC and any dealing in CORE tokens pending arbitration, on a finding that there was a serious issue to be tried over alleged breaches of confidentiality and a 24-month exclusivity clause in their lstBTC partnership, a product involving more than $150M of user Bitcoin. It was settled on 22 May 2026: all claims and counterclaims mutually released, arbitration and court proceedings withdrawn, terms confidential, no admission of liability, and syrupBTC free to proceed. It is closed, and it predates this window, but when the thesis is "you are trusting one team's judgement", a partner taking you to court is a data point. It also means a widely-circulated third-party grade for Maple (Hindenrank, C-, 52 out of 100) is built partly on an overhang that has since been settled, and dates the injunction to March 2026 when it was November 2025.

Worst case

A borrower defaults into a fast, gappy market and posted collateral cannot be liquidated above loan value before it breaks the margin threshold. The documented waterfall is worth knowing before you need it: outstanding principal is written off the pool, accrued interest is removed, collateral is repossessed and marked as unrealizedLosses, collateral is liquidated back into pool assets, and recovered funds prioritize protocol fees before the remainder returns to lenders. You are last in that line, and there is one fewer buffer than the mechanism suggests. Maple's docs head that section "First-Loss Capital (Unused)" and state that pool delegate cover is not used because most pools are run by Maple Direct. So the delegate first-loss layer that is supposed to absorb losses before lenders does not exist in practice. The mitigation that does exist is that the delegate recovers no fees during a default.

The impairment mechanics deserve more attention than they usually get, because they punish panic specifically. Maple can impair a loan before any default when it expects non-payment, which marks your NAV down immediately. If you then withdraw while the impairment is active, the docs say you take a permanent impairment loss and forfeit any claim on later restitution, and any future recovery is pro-rated only to the lenders who stayed. The exit that feels safest is the one that crystallizes the loss.

The second scenario is liquidity rather than credit. The pools are 96% deployed, the non-loan residual is about 4%, and redemption demand beyond that waits on loan maturities behind a FIFO queue with a documented 30-day ceiling. Admins can also remove requests from that queue, and Maple's own security assumptions concede that the withdrawal manager could in theory be configured badly. None of that is happening now, and the queue today is essentially empty. It is the shape of the thing under stress.

A third surface is new this quarter. syrupUSDG went from launch on 2 July to $233.76M on Maple's dashboard in about a month, a fast-scaling asset on a new regulated-dollar rail with no stress history of its own, and its Ethereum pool is 95.5% deployed with a $5.0M residual. Position sizing and treating this as credit exposure rather than a savings account remain the mitigations.

Bottom line

Caution, unchanged, and now for better-evidenced reasons than last month. The question this month was whether a 6% TVL decline signalled credit trouble. It does not, and the answer no longer rests on the absence of bad headlines: zero recognised losses on every pool, loan principal outstanding at $1.93B and at its record, a withdrawal queue holding 0.055% of the syrupUSDC pool with the other three empty, and yield that three independent sources agree is entirely base interest. The decline reads as ordinary maturity and redemption churn in a lumpy book sampled near a local high. I still could not obtain per-day deposit versus redemption flow data, so the precise attribution of the $147M is an inference, but the stress markers that would matter are all absent and I checked them directly rather than by proxy.

The verdict stays at caution because the underlying structure is unchanged and, in two places, worse than previously written. Real borrower default risk sits behind in-house underwriting you cannot audit, agreed offline before it ever reaches a contract. The delegate first-loss buffer that the default waterfall describes is documented as unused. The oracle that drives liquidations is admin-overridable. The collateral custody path for the syrup pools is not publicly documented anywhere I could find, and the custodian names carried in previous versions of this report do not appear in Maple's current documentation at all. And the contractual exit can take a month. No audit closes any of that. Size it as credit exposure, not as a savings product.

Data appendix

  • TVL: ~$2.302B (api.llama.fi/protocol/maple returned $2,301,926,949 on 2026-08-05, verified). Two reference points for the decline, because they differ slightly: against the $2,445,026,544 the 2026-07-01 report printed from api.llama.fi/tvl/maple that is minus 5.9%, and against the $2,448,878,428 DeFiLlama's daily series carries for 2026-07-01 it is minus 6.0%, a fall of $147.0M. Context, all computed from the same series this run: 90-day average $2,160.9M, so today sits 6.5% above it; 90-day range $1,820.3M (2026-06-08 low) to $2,549.4M (2026-07-04 high); YTD minus 2.3% versus minus 34.7% for all DeFi; year over year plus 31.2% versus minus 45.2% for all DeFi. Same-window sector comparison, 2026-07-01 to 2026-08-05: Maple minus 6.0%, all DeFi plus 8.3%. ATH $3,246.0M on 2025-10-30. Chains: Ethereum $2,301.9M, Solana $0.0M; DeFiLlama also reports borrowed $1,833.5M and staking $29.8M.
  • Live pool state (read on-chain and via Maple's GraphQL API, 2026-08-05, Ethereum mainnet). Syrup USDC (0x80ac24aa929eaf5013f6436cda2a7ba190f5cc0b): totalAssets $1,110.67M, principalOut $1,071.98M, unrealizedLosses $0, totalSupply 943,762,744.68 shares, share price 1.1769, withdrawal queue 516,506 shares (~$608K, 0.055% of supply). Syrup USDT (0x356b8d89...): totalAssets $451.92M, principalOut $414.18M, unrealizedLosses $0, queue empty. syrupUSDG (0x87b65c4a...): totalAssets $110.65M, principalOut $105.70M, unrealizedLosses $0, queue empty. High Yield Secured Lending USDC1 (0xc39a5a61...): totalAssets $342.08M, principalOut $340.70M, unrealizedLosses $0, queue empty. Aggregate: $2,015.3M assets, $1,932.6M loan principal, 95.9% deployed, $82.8M (4.1%) non-loan residual. Contract decimals() confirmed as 6, and the API's weekly APY matched DeFiLlama to five decimals, so the readings are live.
  • Flow composition (the month's key question): No default, impairment, writedown or extended withdrawal queue in 2026-07-01 to 2026-08-05, confirmed from protocol state (unrealizedLosses zero everywhere, queues empty or near-empty) rather than only from news absence. Corroborating direction: Maple's own AUM fell from $4.6B at H1 close to $4.36B live, about 5.2%, matching the DeFiLlama slope, so the move is real rather than a methodology artifact. Loans outstanding were a record $1.9B at Q2 close and on-chain principal is $1,932.6M now, so the loan book has not seized. The daily series shows routine single-day moves of minus 6.4% (2026-06-27), plus 16.6% (2026-06-28), minus 8.4% (2026-07-05), minus 7.9% (2026-07-10) and plus 10.3% (2026-07-11), with plus 17.5% and minus 16.8% on consecutive days in early June. Per-day deposit versus redemption versus maturity decomposition: refresh (Maple does not publish daily net flows).
  • Scope warning on AUM. Three sources give three different sizes and they do not reconcile. Maple's transparency dashboard reports syrupUSDC AUM $2.46B; DeFiLlama's yields endpoint reports the same pool at $2,647.8M; the Ethereum Pool contract itself reports $1,110.7M. The same roughly 2x gap appears on syrupUSDT ($879.16M advertised, $451.9M in the contract) and syrupUSDG ($233.76M advertised, $110.7M in the contract). Summed, Maple's Ethereum pool contracts hold about $2.02B against $4.36B of advertised AUM. DeFiLlama lists only Ethereum and Solana for the maple slug and no other chain deployment with material TVL. Multichain deployments are the most likely explanation but I could not confirm it, so the reconciliation and therefore any exact utilization ratio against advertised AUM: refresh. Read the 6% decline as a movement within one measurement scope, not as "$147M left Maple".
  • Audits: Trail of Bits 2022-08, Spearbit 2022-10 (maple-core-v2/audits/2022-december/); Three Sigma 2023-04 and Spearbit auditors via Cantina 2023-06 (audits/2023-june/); Sherlock 2025-09 and 0xMacro 2025-09 (audits/2025-sept-governor-timelock/); Dedaub 2025-11 and Sigma Prime 2026-01 (maple-cross-chain-receiver, Chainlink integration). Seven distinct firms; Cantina is Spearbit's delivery platform, not a separate firm. A November 2025 release covered the Withdrawal Manager. Immunefi bounty active. Maple self-reports a 92% DefiSafety score (undated). Incident check re-run 2026-08-05: 0 of 612 records in DeFiLlama's hacks database mention Maple, verified.
  • Admin/governance (all verified in Maple's docs this run): Pool delegate for the syrup pools is Maple Direct, Maple's own lending arm, responsible for underwriting, structuring and management. Borrower and Maple team agree and sign the term sheet offline, then Maple proposes it on-chain. Governor is the on-chain GovernorTimelock with proposer, executor and role-admin held by Maple-managed multisigs and canceller by a security/ops multisig, executing on a schedule then delay then window. Governor Globals functions include setPriceOracle, setManualOverridePrice, setPlatformManagementFeeRate, setPlatformOriginationFeeRate, setPlatformServiceFeeRate, setBootstrapMint, setValidBorrower, setValidPoolDelegate, setProtocolPause, setTimelockWindows. Security Admin can pause every function and holds upgrade on FixedTermLoan and FixedTermLoanManager. Governor and Pool Delegate can both impairLoan and removeLoanImpairment; Pool Delegate can triggerDefault. On the WithdrawalManager queue, Governor, Operational Admin and Pool Delegate can all processRedemptions, removeRequest and setManualWithdrawal. Contracts are upgradeable to Maple-DAO-approved implementations, and the DAO "has the ability to introduce timelock requirements to sensitive functions", which is a configurable option rather than a guarantee. Stated trust assumptions: lenders inherently trust the Pool Delegate with their funds (3); Governor and Operational Admin are trusted actors (5); Pool Delegates do not try to prevent withdrawals, with an acknowledged theoretical ability to misconfigure the withdrawal manager (7). Exact multisig signer set, threshold and timelock delay values: refresh. Governance this window: MIP-021 rules-based buyback, 10% of monthly revenue below $1.5M per month scaling to 30% above $2M, vote 13 July 2026 (verified).
  • Oracle: Chainlink feeds plus oracle wrappers against outage and manipulation during liquidations, withdrawal cooldowns against flash-loan front-running, and a Governor-set bootstrapMint against first-depositor front-running, all re-verified in the docs this run. Margin calls run off a proprietary three-source price feed with 24/7/365 monitoring, separate from the on-chain oracle. Admin surface: setPriceOracle and setManualOverridePrice sit in the Governor's Globals permissions. No oracle incident in the window.
  • Model and custody: All syrupUSDC, syrupUSDT and syrupUSDG loans are overcollateralized (Maple FAQ, verified this run); current ratios and liquidation thresholds are not publicly disclosed, so live collateralization: refresh. Historical observed levels: roughly 165% average across pools as of 2025-02-06, 136% to 156% for syrupUSDC and 140% to 148% for syrupUSDT through the October 2025 stress, and a stated historical average above 150% for the Secured Lending arm (page dated 2024-06-05). Margin call flow: borrower notified at the Margin Call Level with 24 hours to restore the Initial Collateral Level; at the Liquidation Level Maple has full rights to liquidate, preferentially through partner OTC desks, otherwise by withdrawing to CEX or DEX venues. Custody: the Anchorage, BitGo and Copper/Zodia roster carried in the 2026-07-01 report does not appear anywhere in Maple's current documentation, and BitGo and Copper turn up instead in the separate lstBTC partnership context. The only custodians Maple currently names are J.P. Morgan, as banking provider, prime broker and custodian for the Cash Management Pool with the borrower on a whitelisted Fireblocks wallet, and Kraken Financial for the Kraken warehouse facility via a bankruptcy-remote SPV. Custody path for syrup-pool collateral: refresh.
  • Default handling: Impairment can be applied pre-default when Maple expects non-payment. Lenders who withdraw during an active impairment take a permanent impairment loss and forfeit any later restitution, which is pro-rated to lenders who stayed. Default waterfall: reduce principalOut, accrue and remove interest, repossess collateral and mark unrealizedLosses, liquidate collateral into totalAssets, then liquidate first-loss capital. Recovery prioritizes protocol fees before remaining funds return to the pool, and the delegate recovers no fees. Important qualifier verified this run: Maple's docs title that section "First-Loss Capital (Unused)" and state pool delegate cover is not used because most pools are managed by Maple Direct.
  • Withdrawals: FIFO queue via requestRedeem; requestWithdraw and withdraw disabled. FAQ: normally instant, in rare cases around 24 hours, "maximum possible time being 30 days". Developer docs: typically less than 2 days, up to 30 days during low liquidity. Maple Direct processes them and reports an average under 24 hours. Requests can be reduced or cancelled with shares returned; delegate and protocol admins can also remove any request from the queue. Instant exit alternative via Uniswap or Balancer at a discount. Instant-buffer size, measured this run: $38.7M non-loan residual on syrupUSDC (3.5%), $37.7M on syrupUSDT (8.4%), $5.0M on syrupUSDG (4.5%), $1.4M on High Yield Secured (0.4%); those residuals also cover accrued interest and supporting strategies, so true cash is lower.
  • Yield: Maple transparency: syrupUSDC 5.0%, syrupUSDT 4.1%, syrupUSDG 5.0%, Maple Institutional 5.2%. DeFiLlama yields, independent: 4.96%, 4.14%, 5.02%, with apyBase equal to full APY and apyReward zero on all three. Maple GraphQL syrupUSDC weeklyApy 4.96095%. Net interest margin 0.85%; yield outperformance 1.32%; monthly protocol revenue $1.29M; trailing-12-month revenue $20.72M; treasury $27.68M with $3.99M liquid and 77.45M SYRUP held (maple.finance/transparency, verified 2026-08-05).
  • Scale: AUM $4.36B live, split syrupUSDC $2.46B, syrupUSDT $879.16M, syrupUSDG $233.76M, Maple Institutional $786.79M, total deposits $2B (transparency dashboard, verified; see the scope warning above). Q2 2026 close: AUM $4.6B (up 81% YoY), loans outstanding $1.9B (up 123% YoY, record), YTD originations $5.4B, deposits $2.2B with $1.03B net H1 inflows, Q2 revenue $4.4M (up 47% YoY), ARR $17.5M, Borrower Hub first 300 users, $22B+ originated since 2022 (Maple Q2 2026 update and July 2026 memo, verified; Maple's docs elsewhere still say "over $20B" across 100+ borrowers).
  • Holder concentration: SYRUP $0.155407, market cap $193,357,798, circulating 1,167,185,890 of 1,244,677,493 total, 76.2% below ATH of $0.653229 set 2025-06-25, 7d minus 8.7%, 30d minus 5.5%, 60d plus 21.5% (CoinGecko id "syrup", verified 2026-08-05). Total supply cross-checked directly against the SYRUP contract at 0x643C4E15d7d62Ad0aBeC4a9BD4b001aA3Ef52d66: exactly 1,244,677,492.511099 tokens, matching. Wallet-level distribution: refresh (Etherscan top-holders is a Pro endpoint). Borrower concentration inside the syrup pools: refresh (not disclosed).
  • Legal (missing from prior versions of this report): Core Foundation obtained an injunction from the Grand Court of the Cayman Islands on 19-20 November 2025 against Maple entities, blocking the launch of syrupBTC or variants and any dealing in CORE tokens pending arbitration, on a finding of a serious issue to be tried over alleged breaches of confidentiality and a 24-month exclusivity clause in the lstBTC partnership, which involved more than $150M of user Bitcoin. Settled 22 May 2026: claims and counterclaims mutually released, arbitration and court proceedings withdrawn, terms confidential, no admission of liability, syrupBTC free to proceed (verified across CoinDesk, The Block, PRNewswire and settlement coverage). Note for readers who encounter it: the Hindenrank C- grade (52/100, dated 2026-03-30) treats this as a live overhang and dates the injunction to March 2026, both incorrect as of today.
  • Recent-news scan (2026-07-01 to 2026-08-05): 2 July, joined the Global Dollar Network and launched syrupUSDG, the first new Syrup asset in two years, on Ethereum and Robinhood Chain as collateral for Robinhood Earn, reaching $200M AUM in 8 days. 8 July, Q2 2026 ecosystem update. 13 July, MIP-021 buyback vote. Late June, Kraken onchain warehouse facility using a bankruptcy-remote SPV with Kraken Financial as custodian. Late July, "Against the Tide" memo. No default, impairment, writedown, liquidation or exploit reported in the window. Sourcing caution: at least one aggregator article pairs the December 2022 Orthogonal $36M default with the February 2026 website outage in a way that reads as a new 2026 default; it is the 2022 event, and the February 2026 incident was a front-end compromise. ainvest.com returned HTTP 403 this run; tidresearch.com root resolved but no Maple assessment was retrieved: refresh.

Maintained monthly. Methodology: DeFi Research Instruction v2.

← all protocols