TL;DR
Morpho Blue is a lending core with no admin keys over live markets. Every market's collateral, Oracle The price feed a protocol trusts to value collateral. A wrong or manipulated price can cause unfair liquidations or let bad loans through. and LTV and LLTV LTV is your loan as a share of your collateral's value. LLTV is the LTV at which the position is liquidated; the gap between the two is your buffer. is fixed when the market is created. Depositors usually reach those markets through a Vault and curator A shared deposit pool run by code or by a curator. The curator chooses which markets get your money and sets the limits, for a fee.'s vault. We re-read the core on chain today and diffed it against the block from our 2026-09-20 report. The deployed bytecode has not changed: 15,623 bytes with the same SHA-256 hash. owner() still returns the governance Safe, and both EIP-1967 proxy slots read zero.
The one change is in the Safe, and it was not there when we last wrote. On 2026-09-22 at 00:01:59 UTC the Safe executed a single transaction calling addOwnerWithThreshold on itself. It added a tenth signer, 0xad0e1b385a6e2d5aa4c72f37dcdd4dafd63ced8f, and raised the threshold from 5 to 6. Its nonce went from 281 to 282 and has not moved since. So governance has taken no other action in seven weeks: no fee switch, no new IRM, no new LLTV tier. The quorum got slightly stricter, from 5 of 9 to 6 of 10. We have not identified who holds the new key.
TVL Total value locked: the dollar value of everything deposited in a protocol or pool. It shows size, not safety. is $10.82B against $10.73B on 2026-09-20. That is flat in dollars while ETH fell from about $2,624 to about $2,225. Outstanding loans, which are mostly stablecoins, grew from $5.17B to $5.68B, so the flat headline hides real growth in lending. We found no new bad debt, oracle failure or curator loss since our last report. The 2026-10-29 Binance hack produced a sharp one-day outflow from Morpho, which the interest-rate curve absorbed with no reported freeze. The verdict stays solid as a curator call. Gate logic alone would turn on the Oracle grade, as it did in August.
Checklist
Audits & contracts. Still the strongest dimension. Morpho's Audit A security firm's review of a protocol's code. It lowers the odds of a bug but does not rule one out, so check what was in scope. registry now lists 34 entries, up from 32 in September. The two additions are audits of the Morpho Midnight Ratifiers by Blackthorn and Trust Security, both dated 2026-09-24. That is new audit work, not a backfill. The rest is unchanged. Spearbit and OpenZeppelin cover Blue, with a Cantina contest. Spearbit, ChainSecurity, Zellic, Blackthorn, Certora and a Cantina contest cover Vault V2. Spearbit, Blackthorn, Trust Security, StErMi and a Cantina competition cover Midnight. ABDK and Spearbit cover preliquidation. One Certora Vault V2 adapter entry is printed with the impossible date "2025-15-12", which is a typo on Morpho's page. Markets V2 still has no entry: not verified in this review. The Blue core is immutable, and today's bytecode matches September's byte for byte.
Admin control. We verified this by diffing state, not by quoting docs. The core at 0xBBBBBbbBBb9cC5e90e3b3Af64bdAF62C37EEFFCb returns 0xcBa28b38103307Ec8dA98377ffF9816C164f9AFa from owner() today on three public RPCs (drpc, publicnode, mevblocker). Its code is identical at block 26,014,612, the block of our September read, and at block 26,358,086. The Safe now answers getThreshold 6, ten owners and nonce 282, where in September it answered 5, nine owners and 281. It still reports VERSION 1.3.0. We found the change by bisecting the nonce. It is transaction 0xd7aae02089e6fe87453318fc49195d09f268c4fe7aaeb341228b7bc0870d4c1b at block 26,029,325, submitted by existing signer 0x264c86dbbd2e4165fbbf0c35b0ddf0e00aec6b31. It succeeded and emitted AddedOwner and ChangedThreshold. The nine September signers are all still present. The module list is empty and the guard slot reads zero. That rules out any other path for owner actions, and with the nonce at 282 we can say the core owner did nothing else in the window.
The owner's powers have not changed. It can turn on a fee capped at a quarter of borrower interest, whitelist LLTV tiers and IRMs for future markets, and handle treasury, token and ENS. It cannot change a live market's oracle or LLTV. The yellow remains the curator layer. Your vault's curator picks markets, sets caps, appoints allocators and takes a fee. Every loss in this protocol's history has happened at that layer.
Oracle. The market creator chooses each market's oracle, so its quality belongs to the market, not to Morpho. We found no new oracle event since 2026-09-20 in an X-wide scan, a web news scan or the DeFiLlama hacks list. That list holds 1,340 records and still names Morpho as a target exactly once, the LeadBlock PAXG market in October 2024. Morpho Blue appears in several other entries only as the source of flash-loan capital for attacks on other protocols.
What is new is evidence that failed oracles leave a permanent residue. A Bitquery investigation read Ethereum state at block 26,036,463 (2026-09-22) and found three cases. The sdeUSD/USDC market from Stream's November 2025 collapse still carries about $7.4M of unpaid loans, behind a deUSD feed that last updated on 2025-12-19. The PAXG/USDC market from 2024 carries about $95K, priced by a feed that reads PAXG in the quadrillions. Three Resolv markets (wstUSR, RLP, USR) carry about $8.5M borrowed, behind feeds that still price the tokens between $1.00 and $1.29 after the March 2026 crash. With no live price, Liquidation When a loan passes its LLTV or a perp position runs out of margin, it is closed by force and the collateral is sold, usually with a penalty. cannot run, and the core has no function to close a market. So these losses never settle on chain. We have not reproduced these figures.
This is not a new loss, but it sharpens the August argument for red: an oracle failure here is not a bad day, it is a permanent hole in a market. We are holding Oracle at yellow on doxe's 2026-08-11 call, for the reason he gave then. The depositor picks the market, the curator and the oracle, and the core cannot change any of them. If you read this dimension as red, the verdict becomes caution. The facts are the same either way.
Liquidity & exit. There is no protocol-level exit guarantee. You withdraw from your market's unborrowed Liquidity How much you can sell or withdraw right now without moving the price or waiting. Thin liquidity means a slow or costly exit.. We re-measured Vault V2 through Morpho's GraphQL API. The top 200 V2 vaults hold $5.384B out of 3,844 V2 vaults that exist. Instant liquidity is $1.640B, or 30.5 percent, of which $795M is idle cash.
That headline flatters the book. One vault, Sentora x Spark RLUSD at $500.0M, is 94.7 percent liquid and accounts for $473.5M of the liquid total. Without it, the rest of the book is 23.9 percent liquid. Grove x Steakhouse USDC on Base, now the largest vault at $546.6M, is 21.3 percent liquid. Steakhouse USDG on Robinhood Chain ($514.6M) is 10.0 percent, Steakhouse High Yield USDC ($461.4M) 5.4 percent, Sentora RLUSD Main ($411.1M) 5.0 percent, Paypal USD Main ($330.3M) 9.1 percent and Sentora PRIME Main ($190.9M) 2.1 percent. We did not re-measure force-deallocatable capacity this run: not verified in this review.
The 2026-10-29 Binance hack tested this. DeFiLlama shows Morpho Blue TVL falling from $11.203B to $10.466B between the 10-29 and 10-30 snapshots, down $737M or 6.6 percent, while ETH fell 4.6 percent. Borrowed fell $281M as positions were repaid. TVL recovered to $10.907B the next day. One X analyst (Sumcap) reported about $470M pulled from Morpho stablecoin markets in about eleven hours, which cut the unborrowed buffer from $616M to $444M. The same analyst reported $95M leaving the Base cbBTC/USDC market, Utilization The share of a lending market's deposits that is lent out. Near 100%, depositors may have to wait to withdraw and the borrow rate jumps. reaching 95 percent and the borrow rate peaking at 13.8 percent before easing to 6.8 percent. We did not reproduce those figures. We found no report of a frozen withdrawal and no report of frozen USDT used as collateral on Morpho. Press accounts say the hacker borrowed about 20,000 WETH against stolen USDT on Aave, not on Morpho. The rate curve did what it is meant to do.
Yield: real vs emission. The yield is real. Supply yield is interest paid by borrowers, and MORPHO incentives sit on top of it rather than replacing it. DeFiLlama shows $5.683B borrowed against $10.82B TVL, 52.5 percent, up from 48.2 percent in September. When a front end advertises a stablecoin yield far above ordinary lending rates, the difference is Leverage Borrowing to hold a position larger than your capital. At 3x you hold $3 for every $1 of your own, and gains, losses and liquidation risk grow with it. or a subsidy. The August PT (principal token) A Pendle token that redeems for one unit of the underlying on its maturity date. Bought at a discount, it locks in a fixed yield if held to that date.-reUSD cascade showed which of the two gets liquidated first. The roughly 18 percent APY Yearly yield with the earnings reinvested (compound interest). Most tables here show APY; funding and the exchange vaults show APR. on Bitget's Morpho-powered USDC vault, flagged in August, is still unexplained: not verified in this review.
Holder concentration. On the token side, CoinGecko shows MORPHO at $2.13, down 15.1 percent over 30 days. Market cap is $1.496B and FDV $2.126B, with 703,631,242 of 1,000,000,000 tokens circulating (70.36 percent). In September circulating supply was 698,882,077. So about 4.7M tokens entered circulation in seven weeks, against about 42.7M in the six weeks before. We do not know whether that slowdown is the unlock schedule or a change in how CoinGecko counts: not verified in this review. The allocation split is carried from August and was not re-pulled: not verified in this review.
The concentration that reaches deposits keeps rising. In the top 200 V2 vaults, the top ten hold 63.1 percent. Vaults with Steakhouse in the name hold 45.4 percent of that book, up from 41.8 percent. Sentora-named vaults hold 23.4 percent, up from 19.2 percent. Together the two brands hold 68.8 percent of the V2 book we measured. Co-branded vaults count under both names here: Grove x Steakhouse as Steakhouse, Sentora x Spark as Sentora. We did not re-measure Vault V1 this run: not verified in this review.
Track record. The core contract has never been the point of failure since its early-2024 launch. Its record now includes a CEX-hack stress day on top of the November 2025 liquidity squeeze and the April 2026 KelpDAO event. Above the core, the record is unchanged at six incidents through August 2026, and we found no seventh since 2026-09-20. The Bitquery work adds detail to two of the old ones. It reports that several vaults lending into the sdeUSD market were effectively written off, including Relend USDC at a share price of 0.018, Usual Boosted USDC and Avantgarde USDC Core. It also reports that two others, Adpend USDC and 1337 USDC, still count uncollectable paper interest as assets. Our September report described November 2025 as roughly 3.6 percent of one MEV Capital vault. For the smaller vaults Bitquery names, the loss was close to total. We have not verified those share prices ourselves: not verified in this review.
Morpho Midnight (new sibling). DeFiLlama now lists a separate child, Morpho Midnight, which it describes as Morpho's fixed-rate, fixed-term lending protocol. It holds $160.5M, of which $151.6M is on Ethereum and $8.8M on Base, with $87.2M borrowed. Press dates the Base launch to July 2026. The Defiant reports the Ethereum deployment on September 8, starting with USDC markets against cbBTC and WBTC and not yet open to Morpho vaults. Its TVL was $4.1M on 2026-09-21 and $15.9M on 10-22, then jumped to $113.9M by 10-24. We do not know what drove that jump: not verified in this review. Midnight is outside this report's TVL scope, but it has its own nine audit entries and should be graded separately if it reaches our boards.
Worst case
The worst case has the same two forms as in September.
As a depositor, your curator lists collateral whose oracle stops telling the truth. The asset collapses, borrowers draw against it at par, liquidations cannot run, and your vault absorbs the loss pro rata. The Bitquery data adds the tail: if the feed then dies, the hole sits in the market indefinitely. A vault that keeps booking paper interest on that hole shows a share price nobody can redeem at. Whoever exits first gets real cash, and whoever is last gets the hole.
As a borrower in a tight loop on thin collateral, an honest feed can liquidate you within minutes on a move that says nothing about the asset's health. August 2026 cost leveraged PT-reUSD positions about $36.4M that way.
There is no insurance fund behind either case. Your mitigations are choosing the curator, checking the oracle and sizing the position.
Bottom line
Solid, held as a curator call. The core is the same code we read in September, and governance acted once, to raise its own quorum. The loan book grew about 10 percent while ETH fell 15 percent. The protocol absorbed the largest CEX hack in DeFiLlama's hacks list ($6.36B, Binance) without a reported frozen withdrawal. None of this weakens the case for solid. If anything, the governance change and the hack-day stress test strengthen it slightly.
One thing pulls the other way, and we are naming it rather than burying it. Third-party on-chain work now shows that this protocol's oracle failures leave permanent holes, and some vaults that lent into them were wiped out or are still reporting inflated share prices. Gate logic alone turns on the Oracle grade. Read Oracle as red, as the August research pass argued, and the verdict is caution. Doxe's call keeps it at yellow and the headline at solid. What to watch: curator concentration, now about 69 percent of the V2 book under two brands; how liquid the largest vaults are, several between 2 and 10 percent; and Midnight, which grew about 40-fold in seven weeks.
Data appendix
- TVL: $10,822,182,661 (DeFiLlama
morpho-blue, api.llama.fi/tvl/morpho-blue). On 2026-09-20 it was $10,727,700,000 by the history series, so up 0.9 percent while ETH went from about $2,624 to about $2,225, down about 15.2 percent. Borrowed went from $5,170.9M to $5,683.1M, up 9.9 percent, which is the clearest sign of new lending rather than price. 30d mark (2026-10-08) $11,244.6M, down 3.8 percent. 90d mark (2026-08-09) $7,944.8M, up 36.2 percent. The series high is a one-day print of $11,465.8M on 2026-11-05, which fell back to $10,951.2M the next day. We treat it as a possible data artifact. The prior sustained high was $11,458.8M on 2026-10-06. Chain split (09-20 to today): Ethereum $5,006.8M to $4,863.2M, Base $4,179.4M to $4,293.3M, Robinhood Chain $553.0M to $578.2M, Arc $185.1M to $287.8M, Hyperliquid L1 $268.8M to $128.9M (down 52 percent), Arbitrum $182.6M, Monad $167.4M and Pharos $113.6M today. Parentmorphois $10,982.8M, which is Blue plus $160.5M of Morpho Midnight plus about $0 in legacy optimizers. - Morpho Midnight: DeFiLlama child, "fixed-rate, fixed-term lending protocol". $160,456,923 total: Ethereum $151.6M, Base $8.8M, borrowed $87.2M. History: $4.1M (09-21), $5.5M (10-08), $15.9M (10-22), $56.4M (10-23), $113.9M (10-24), $160.5M today. Launched on Base in July 2026 (press release dated 07-21 per The Block and Cointelegraph). The Defiant reports an Ethereum deployment on September 8 with USDC/cbBTC and USDC/WBTC markets, not open to vaults. The driver of the late-October jump is not verified in this review.
- Audits: docs.morpho.org/learn/resources/audits, 34 entries. New since September: Morpho Midnight Ratifiers by Blackthorn (2026-09-24) and Trust Security (2026-09-24). The latter is also linked from DeFiLlama as midnight/audits/2026-09-24-rate-ratifier-trust.pdf. Midnight now has nine entries in total. Blue: Spearbit (2023-10-16), OpenZeppelin (2023-10-13), Cantina contest (Nov to Dec 2023). Vault V2: Spearbit, ChainSecurity, Blackthorn, Cantina, Zellic, Certora. Also preliquidation (Spearbit, ABDK), public allocator (Spearbit), Vault V1 and V1.1 (Spearbit, OpenZeppelin, Cantina), token (Omniscia), infrastructure (Lexfo), frontend (Securing). Markets V2 is not listed: not verified in this review.
- Admin/governance: Read on chain today at block 26,358,086 via drpc, publicnode and mevblocker, with archive reads via drpc and mevblocker at block 26,014,612. Core 0xBBBBBbbBBb9cC5e90e3b3Af64bdAF62C37EEFFCb: owner() = 0xcBa28b38103307Ec8dA98377ffF9816C164f9AFa. Bytecode is 15,623 bytes with SHA-256 fd5aa16eea01735e0e33e51e913a27f7d5ed998adc0cec2bf1f9ea368d9e1bdf at both blocks. EIP-1967 implementation and admin slots are zero. Safe: getThreshold 6 (was 5), getOwners 10 (was 9), nonce 282 (was 281), VERSION 1.3.0 (unchanged). getModulesPaginated is empty and the guard slot is zero. The only transaction in the window is 0xd7aae02089e6fe87453318fc49195d09f268c4fe7aaeb341228b7bc0870d4c1b, block 26,029,325, 2026-09-22 00:01:59 UTC, from signer 0x264c86dbbd2e4165fbbf0c35b0ddf0e00aec6b31. It is execTransaction to the Safe itself, calling addOwnerWithThreshold(0xad0e1b385a6e2d5aa4c72f37dcdd4dafd63ced8f, 6), status success, with AddedOwner and ChangedThreshold events. All nine September signers remain, confirmed by an archive getOwners at block 26,014,612. The identity of the new signer is not verified in this review.
- Oracle: Per market, chosen by the market creator, with no protocol default. No new incident found from 2026-09-20 to 2026-11-06 (X scan via Grok x_search, web news scan, DeFiLlama hacks list of 1,340 records, whose only Morpho-targeted entry is LeadBlock's PAXG market on 2024-10-14 at $250K). Bitquery investigation "How $7.5 million of unpaid loans became $14.7 billion of debt on Morpho" (state at block 26,036,463, 2026-09-22, figures measured by Bitquery 2026-09-23, not by us): sdeUSD/USDC about $7.4M unpaid behind a deUSD feed last updated 2025-12-19; PAXG/USDC about $95K; Resolv wstUSR, RLP and USR markets $8.5M borrowed behind feeds pricing between $1.00 and $1.29. A separate secondary summary puts Resolv-related bad debt in Morpho vaults at about $6.2M, mostly in one curator's vaults. The two figures are not reconciled: not verified in this review.
- Liquidity: Morpho GraphQL vaultV2s, top 200 by assets, $5,384.2M of 3,844 V2 vaults. Instant liquidity $1,639.9M (30.5 percent), idle $794.9M (14.8 percent). Excluding Sentora x Spark RLUSD ($500.0M, $473.5M liquid): 23.9 percent. Largest vaults: Grove x Steakhouse USDC Base $546.6M (21.3 percent liquid), Steakhouse USDG Robinhood Chain $514.6M (10.0), Sentora x Spark RLUSD $500.0M (94.7), Steakhouse High Yield USDC Base $461.4M (5.4), Sentora RLUSD Main $411.1M (5.0), Paypal USD Main $330.3M (9.1), Sentora PRIME Main $190.9M (2.1), Steakhouse Prime EURCV $160.4M (71.3), Steakhouse Prime USDC Base $145.7M (79.9), Steakhouse Prime USDC Ethereum $137.9M (19.9). Force-deallocatable capacity: not verified in this review. Binance hack window: TVL $11,202.8M (10-29) to $10,466.1M (10-30) to $10,906.6M (10-31). Borrowed $5,757M to $5,476M to $5,513M. ETH $2,357 to $2,249 to $2,242. The X-sourced figures (about $470M of stablecoin outflow in about 11 hours, buffer from $616M to $444M, $95M out of Base cbBTC/USDC, 95 percent utilization, borrow rate peaking at 13.8 percent) were not reproduced.
- Holder concentration: CoinGecko, 2026-11-06T20:05Z: MORPHO $2.13, market cap $1,496,091,546, FDV $2,126,243,774, circulating 703,631,242 of 1,000,000,000 (70.36 percent), down 15.1 percent over 30d. Up from 698,882,077 on 2026-09-19. The reason for the slower release is not verified in this review. Allocation split carried from August: not verified in this review. On-chain top holders: not verified in this review. Vault V2 (top 200): top ten 63.1 percent, Steakhouse-named $2,443.4M (45.4 percent), Sentora-named $1,259.8M (23.4 percent). Vault V1 not re-measured: not verified in this review.
- Recent-news scan (2026-09-20 to 2026-11-06): No exploit, bad debt, oracle failure, curator loss or frozen withdrawal attributed to Morpho Blue or its vaults was found on X, in web news or on the DeFiLlama hacks list. The Binance hack of 2026-10-29 (DeFiLlama logs it at $6.36B, key leaked via infrastructure) hit Morpho through outflows only. Press places the hacker's borrowing against stolen USDT on Aave. Governance: one Safe transaction (signer added, threshold 5 to 6). Midnight expanded on Ethereum and grew from $4.1M to $160.5M. Re-check at the next refresh: PT-reUSD and Re status after August, and the Moonwell-curated vaults disabled in August.
- Carried forward, unchecked this run: composition of the Bitget USDC vault APY, the MEV Capital vault identity behind the November 2025 3.6 percent figure, the KelpDAO and Aave comparison figures from April 2026, and the deduplicated V1 plus V2 vault total.
Maintained monthly. Methodology: DeFi Research Instruction v2.