alphalend · may 2025 to sep 2026 · times utc

$17.1M of bad debt, 4 positions, & the fee wallet that gassed them

alphafi's tweet said borrow positions against ALPHA collateral "became substantially undercollateralised" & that the bad debt "has now been fully covered." here's the tx that covered it, the four positions it covered, who opened them, & what the protocol thought their ALPHA was worth at the time. spoiler: not what it traded at. position labels follow the oracle timeline.

the four positions

position ids are dynamic fields, so a plain object lookup returns nothing; the link opens the field object suiscan can show. creators found by scanning PositionCreateEvent on the original lending package 0xd631cd66…, then re-verified by fetching each creation tx.

positionopened (utc)opened bycollateral at creationwhat alphalend booked it atmarket that daydebt the bailout paidholds now (sep 27)
P1 0x9e9d0f9c…2025-05-02 12:16:11 CaGfwWZ2…0x4dde…819,104 SUI, then about 2.2M more SUI & 1,348,978 stSUI within the hour. no ALPHA until oct 2025$2.82M for the first deposit (SUI at $3.45, which was right). its 1,757,957 ALPHA on oct 23 2025 was booked $4.33M, $2.46 eachn/a at creation, it was SUI. $1.11 on oct 23 2025$3,163,830
2,104,629 USDC + 1,059,669 USDT
1,759,328 ALPHA + 35 WAL, no loans
P2 0x74d35b41…2025-06-25 08:37:35 21orUQ4Q…0xdfe2…2,391,005 ALPHA$6,632,011 ($2.77 per ALPHA)$1.03 (coinmarketcap daily), so 2.7x$7,878,752
43.63 xBTC + 4,342,663 USDC
nothing. 3,261,824 ALPHA withdrawn sep 25 11:39
P3 0xd439d8da…2025-08-24 17:06:37 F1LsEizq…0xa66f…1,065,807 ALPHA$3,893,418 ($3.65 per ALPHA)$1.07 (coinmarketcap daily), so 3.4x$4,776,283
41.08 xBTC + 1,446,940 USDC
2,082,087 ALPHA, no loans
P4 0xf0a41462…2025-12-03 21:52:13 3Zghi1qL…0x1d1c…2,236,291 ALPHA$31,658,871 ($14.16 per ALPHA)$0.95, market cap $9,273,008 (coingecko)$1,310,917
10.15 xBTC + 4,630,460 DEEP + 405,242 USDC
1,822,439 ALPHA, no loans

P1 was opened w SUI six days before alphalend's public launch ("now live," 2025-05-08); it took in 1,253,217 ALPHA in october 2025 (1,151,846 from 0xa3fd0ab6…, 20,000 from w9273, 80,000 & 1,371 from the fee wallet) & was ALPHA-only by the end. all four were backed by ALPHA & nothing else when the bailout paid them. together they held 5.7M ALPHA, 57% of the 10M supply.

what alphalend said the ALPHA was worth

each creation tx has a DepositEvent w the value the protocol booked (whole dollars in may 2025, cents from june on, a package upgrade changed the precision). P2's 2,391,005 ALPHA was booked at $6,632,011, so $2.77 each. P3's 1,065,807 at $3,893,418, $3.65 each. P4's 2,236,291 at $31,658,871, $14.16 each, on a day coingecko had ALPHA at $0.95 w a $9.27M market cap. that's the oracle marking one wallet's collateral at 3.4x the token's entire market cap. within two hours P4 borrowed about $9.3M against it.

how the mark got there is the oracle timeline: ALPHA didn't have its own price feed, so the admin mapped it onto other coins' pyth feeds w add_alternate_price_identifier. SUI first, then stSUI, then gold (XAUm) for about 16 hours, then APT, where APT itself was pointed at an AVAX/USD feed. here's one of those remaps, ALPHA to APT: FigVarLxwwAp6k4pwr7A5gBNJQyCxrz6ep4Q4ECM6k87. the timeline page charts the booked price against the market price for the whole 16 months, w the LTV underneath & a tick for each of the 59 times the deposit cap was raised. every remap w its tx, & the way the market's deposit cap, LTV & liquidation settings were opened for minutes at a time around these deposits, is the gate. it looks to me like the ALPHA market was open for specific wallets & closed for everybody else most of the time. that's a read, not a proof.

collateral, exactly

read from the position objects themselves (dynamic fields of the protocol's positions table), so this is the state on chain, not a valuation. "at bailout" is 2026-09-20 22:50 utc, "now" is 2026-09-27.

positionat creation & along the wayat bailoutnow
P1 0x9e9d0f9c…819,104 + 1,060,378 + 956,873 + 164,270 SUI & 1,348,978 stSUI (2025-05-02); SUI, stSUI, BLUE, USDC, USDT, EGUSDC & WAL cycled through later; all SUI gone by 2025-12-03; 1,757,957 ALPHA in on oct 23 20251,759,327.589 ALPHA + 35.479 WALsame, no loans
P2 0x74d35b41…2,391,005 ALPHA (2025-06-25); 311k, 40k, 15k, 250k, 5k, 3k & 12.85k ALPHA pulled out jun to aug 2025, more added later (1,215,262 by the fee wallet on dec 25); SUI, USDC, USDT, stSUI, BTC, tBTC, DEEP, IKA, BLUE & WAL used as side collateral at various times, all zero by 2026-09-053,261,823.818 ALPHA, nothing elsenothing, no loans
P3 0xd439d8da…1,065,807 ALPHA (2025-08-24); stSUI added aug 26 to 30; ALPHA topped up in 20 or so tranches; eSUI, eBTC, tBTC in dec 2025; 1,161,495 ALPHA (dec 14, from 0x1d1c), 1,000,008 ALPHA (dec 17, from the fee wallet out of P4), 196,502 ALPHA (dec 19); 266,748 ALPHA pulled dec 11; last ALPHA change 2026-01-162,082,086.714 ALPHA, nothing elsesame, no loans
P4 0xf0a41462…2,236,291 ALPHA (2025-12-03) plus 987,856 + 493,928 eSUI, 2,514,368 eWAL & 494,097 USDT the same night; eWAL & eSUI pulled dec 4 & dec 7; USDC, USDT, SUI, stSUI, DEEP, WAL cycled through; ALPHA down to 1,822,438.516 by 2026-05-111,822,438.516 ALPHA, nothing elsesame, no loans

so at the moment the foundation paid, every one of the four was backed by ALPHA & nothing else, apart from 35 WAL. three of them still hold 5,663,853 ALPHA between them, 56.6% of the supply, now worth a rounding error. the protocol's own stale valuations of that ALPHA, from the last time each position was refreshed: P3 $28.2M on 2026-01-16 ($13.56 each), P4 $18.4M on 2026-05-11 ($10.07 each), P2 $24.9M on 2026-09-06 ($7.63 each) & $36.5M when it was withdrawn on 2026-09-25 ($11.18 each). market that week: $0.65 to $0.90.

what each creator did next

from each wallet's own tx list.

  • P2, 0xdfe2…: the fee wallet sent it 2 SUI of gas at 08:08:06 (2xpHngAf…). w9273 sent it 2,391,005.24 ALPHA at 08:36:43 (2wkUnAxx…). position opened at 08:37:35, 52 seconds later. over the next two hours it ran a leverage loop: borrow stSUI, convert, deposit SUI, repeat, seven rounds (894,244; 768,292; 658,993; 564,500; 486,863; 420,674; 365,000 stSUI borrowed, about $11.7M at the protocol's prices, against 4.23M SUI deposited). w9273 is the same wallet that later sent a027 50,000 ALPHA.
  • P3, 0xa66f…: its first ever inbound tx is 2 SUI of gas from the fee wallet on aug 9 2025 (E9fWJ2qS…). on aug 24 it received 1,816 ALPHA from 6a49 & 1,063,991 ALPHA from 0x9d11564c…, a day-one ALPHA depositor, posted 1,065,807 ALPHA, then within two hours borrowed 607,510 SUI ($2,236,005 by the protocol's own event), 13,903,997 IKA ($457,614), 63.66 XAUM ($254,040), 855,047 ALKIMI ($134,195) & 150 UP ($122,001), about $3.2M, & forwarded it all to f71e (e.g. the UP at 18:53:08, AZHYqdqB…). f71e sent it 200,000 stSUI & 101,998 SUI during the session. then in november it did the loop below.
  • P4, 0x1d1c…: its first ever inbound tx is 1 SUI of gas from the fee wallet on nov 30 2025 (Eh3QkG1k…). three days later it posted 2,236,291 ALPHA, then within two hours borrowed 2,500,000 SUI ($4.21M), 6,200,000 WAL ($1.02M), 47,500,000 DEEP ($2.07M) & 2,000,000 USDC, about $9.3M, & forwarded it to 7c26 (500,000 SUI at 22:20, 3.6M WAL at 22:53, 27.5M DEEP at 23:07, & so on). it also posted 987,856 ESUI & 2,514,368 EWAL as extra collateral & parked borrowed SPRING_SUI & DEEP in suilend the same night. eight days later this wallet sent a027 18 BTC to bridge. this is the wallet i call "treasury" on the other pages, cuz 3.2M ALPHA went through it in six weeks.
  • P1, 0x4dde…: opened w SUI before the public launch & borrowed USDC & USDT in small steps that morning (1,333,222 USDC & 666,765 USDT among them). its first outbound transfer ever, 2024-08-09, was a dust transfer to the fee wallet. took in 1.25M ALPHA in oct 2025, some of it straight from the fee wallet.

the nov 26 to 30 loop

the money went in a circle through the fee wallet. all from the P3 creator's tx list, 2025.

utcwhattx
11-26 09:36 to 09:470xa66f borrows 15 + 13 xBTC7JKhaQPi… 9EUraNtN…
09:50 to 10:03borrows 3,697,813.8 USDC, 1,078,142.8 USDT, 500,000 USDT7j2kcmrR… 65fDbpKi… 3c2Ji3yk…
09:37 to 10:04sends all of it to f71eErD94Tpg… CRhLGxDX… FjivntBW… AwAJv7pb… CcnXRc3U…
11-27 05:18 / 05:19f71e sends 28 xBTC & 1,505,929 USDT to the fee wallet; the fee wallet sends 28 xBTC & 1,505,990.6 USDT to 0xa66f5kFpr6JQ… FB42QD6e…
11-27 05:19 / 05:200xa66f repays 28 xBTC & 1,505,990.6 USDTCf4ZgUju… 26GJRVgH…
11-28 07:18 / 07:19f71e sends 3,583,284 USDC to the fee wallet; the fee wallet sends 3,583,555.8 USDC to 0xa66fAUSix2D4…
11-28 07:20 / 07:210xa66f parks 350,000 USDC in an asset bank & repays 3,233,555.8 USDC9bMSmb3n…
11-30 06:08 to 06:10f71e sends 0xa66f 579,648 USDC; 0xa66f repays 466,650 USDC & 72,640 USDT
11-30 22:05 to 22:390xa66f borrows 20 + 22.1 + 5 = 47.1 xBTC, 1,984,979 USDC & 418,653.9 USDT & sends every coin straight to the fee wallet5Nvh6mEe… gQbr2eJh… 4DiPXdyp… Bw3T1pah… 26wmUy9M…

the bailout tx

FK6jAgAPEK7eEKbGJDZo9YweNBFyMvw47wxnBjbQWewf, 2026-09-20 22:50:17 utc, sender 0x3c2830dc…7a07, calls loan_bailout nine times. i got asked how i know this is a bailout & not just someone paying back their loans. fair question. five reasons:

  • loan_bailout(&mut LendingProtocol, market_id, position_id, Coin<T>, &Clock, ctx) is public & takes a position id, not a capability. the normal repay takes &PositionCap, the borrower's ownership object. paying another wallet's loan is the only thing loan_bailout exists for, & the name is the contract author's, not mine. the tx's inputs are the protocol object, the clock, the sender's coins, four position ids & market ids. no capability object of any kind.
  • the sender owns none of the four positions. they were opened by the four wallets above. the sender's wallet was created on sep 14 & funded from exchange-style wallets on sep 14 & 15.
  • all nine loans on all four positions were cleared in one tx (8,299,475 USDC, 1,059,669 USDT, 94.86 xBTC, 4,630,460 DEEP) & the sender got nothing back: no coins, no objects, no collateral.
  • the two functions that do need an AdminCap, loan_writeoff (which would socialize the loss on lenders) & admin_liquidate, were not used.
  • who stood behind the sender is not visible on-chain beyond the funding wallets. the attribution to the sui foundation rests on the public statements of sep 25, which is also the only reading under which a fresh wallet fronting $17.1M for positions it doesn't own, & taking nothing back, makes sense.
positionrepaidat oracle prices
P1 0x9e9d0f9c…2,104,629 USDC + 1,059,669 USDT$3,163,830
P2 0x74d35b41…43.63 xBTC + 4,342,663 USDC$7,878,752
P3 0xd439d8da…41.08 xBTC + 1,446,940 USDC$4,776,283
P4 0xf0a41462…10.15 xBTC + 4,630,460 DEEP + 405,242 USDC$1,310,917
total94.86 xBTC, 8,299,475 USDC, 1,059,669 USDT, 4,630,460 DEEP$17,129,783

values from the protocol's own RepayEvents. implied oracle xBTC price $81,053. the 94.86 xBTC leaving the wallet matches the nine events exactly. four positions borrowed $17.1M against a token whose entire market cap was about $9M.

where the bailout wallet's money came from

utcfromamount
2026-09-14 17:27 / 17:290x03b61aa7…e72e5 SUI, 5 USDC (a test)
2026-09-14 19:11 to 19:150x4eb81816…e5cd95.3 xBTC, 1,060,061 USDT, 4,651,500 DEEP
2026-09-15 16:25:280x60956e40…449b8,400,000 USDC
2026-09-18 16:42 / 16:430x4eb81816…e5cd5 USDT, 80,000 USDT
2026-09-23 18:32 / 18:330x03b61aa7…e72e8 SUI, 1,500,000 SUI

0x6095… & 0x03b6… behave like exchange hot wallets (dozens of distinct recipients per hour). 0x4eb8… is a dedicated wallet w 24 txs & two recipients. the sui foundation didn't state an amount in its announcement; the $17.1M is the chain's number. the wallet was created & funded eleven days before anyone was told anything, so the problem was known for at least that long.