Loading today’s stories
Invest1 distinct publisher3 min readPublished
The Solana neobank has acknowledged an issue with card withdrawals and nothing else, while the only visible equity behind any repayment is a $3.39m token that fell 39% on the very news that created the liability.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
Follow any of these and your For You feed starts watching them — no settings page required.
invest
A Solana DEX halted trading and says the loss stopped at its treasury. Nobody can check1 distinct publisher
invest
Phantom prunes Sui and Monad, and the exit fee points to Solana1 distinct publisher
invest
The agent payment rail cleared 8.7 million transfers last week, and $367,9501 distinct publisher
invest
500m SAND minted on a Base contract that held 14.7m, and Upbit posts a caution1 distinct publisher
AddCollateralAdmin is the permission that mattered here. The reported sequence ran SubmitSignatures against Avici's authorization program, then AddCollateralAdmin against its collateral program, then WithdrawCollateralAsset [4]: the permission layer doing exactly what it was written to do for whoever reached it first, unlike the price-gaming exploit that reportedly took as much as $9m from the lending protocol Moonwell days earlier [13]. The wallet that ended up holding roughly 10,005 SOL worth about $1.07m plus another $11,600 in stablecoins [5] implies a SOL price near $107 [6], and what it drained was collateral of the sort that backs spendable card balances [10], which is the product a depositor thought they were buying.
Now the balance sheet, such as it is. CoinMarketCap shows about $3.39m of market capitalisation across roughly 12.9m circulating tokens and some 12,400 holders [7], which divides out to $0.263 a token [1] and quietly settles which of the reported post-crash prices to trust, since the same report carries $0.26, $0.2175 and $0.2625 alongside an intraday range of $0.2189 to $0.4459 [14]. Back the 39% out [2] and the pre-breach capitalisation was about $5.56m, so roughly $2.17m of token value went away against a midpoint loss near $800,000, call it 2.7 times [3]. The selloff outpaced the theft itself, marking the franchise down by a multiple of it.
Per account the sums are small: 125 affected of about 12,400 holders is one percent [4], and $4,800 to $8,000 apiece across the reported range [5], which a chartered bank absorbs out of an operating line without a press release. This is probably wrong, but I read the $800,000 as the smaller number in this story and the $3.39m as the larger one, because against the upper reported loss of $1m the visible equity is that market capitalisation, about 30 percent of it [2], and it fell 39 percent on the same @SolanaFloor post that surfaced the liability [9]. Deposit insurance exists so that a bank's capital is not priced off the news that impaired its assets.
This could unfold three ways from here. Avici funds repayment 1:1 out of off-chain cash, as The Sandbox moved to do after a bridge exploit that Cryptopolitan put at about $697,000 [11], and the neobank wrapper survives its first stress. Or it publishes a post-mortem with no recovery plan, which is what MANTRA did on a roughly $3.6m exploit the same week [12]. Or it says nothing beyond the issue affecting card balance withdrawals it has acknowledged, while the partner work it says it is doing toward a fix [3] is engineering not going into the card product it sells, and the token, already about 96 percent below its $7.61 peak of 26 November 2025 [8], trades on as a claim against a service nobody can withdraw from. I lean to the second, and I would abandon that read the day Avici names the flaw, names the number, and pays from something other than the token.
Ranked by verification strength, evidence, and original report placement.
The AVICI token fell about 39% over 24 hours following news of the breach, hitting a new all-time low.
Avici said on X on August 28 that it was aware of an issue affecting card balance withdrawals, was monitoring the situation and was working with partners toward a fix; it did not label the event a hack, state a loss figure, or say whether users would be repaid.
According to reports, the attacker first ran a function called SubmitSignatures on Avici's authorization program, then called AddCollateralAdmin on Avici's collateral program, then ran WithdrawCollateralAsset to remove the funds.
The attacker's wallet held about 10,005 SOL, worth around $1.07 million at the time, plus about $11,600 in stablecoins.
A blockchain analyst named STACC created a real-time tracker that found 125 different user accounts were affected, with amounts taken ranging from about $9 USDC to over $26,000 USDT per account.
CoinMarketCap lists AVICI market capitalisation near $3.39 million, circulating supply of roughly 12.9 million tokens, and about 12,400 holders.
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 28, 2026
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
One outlet, no confirmed figure
Follow any number in this story and you arrive at the same place: a single Cryptopolitan report leaning on an X post from @SolanaFloor, a tracker run by an analyst identified only as STACC, and a CoinMarketCap snapshot. Avici has confirmed precisely one thing — that card withdrawals were affected — and the loss itself is still a range wide enough to nearly double. The instruction-call sequence is the sturdiest detail because it is checkable on-chain, but even it is relayed as what 'reports' say rather than traced by the outlet.
Small deposit base, real money in it
Small, but not theoretical. Somebody's actual spendable card balances were emptied: 125 distinct accounts, individual losses running from $9 to more than $26,000, out of roughly 12,400 token holders. That is a live product with real deposits and roughly a hundred-odd users directly harmed — meaningful for them, immaterial at the scale of the wider Solana ecosystem, and enough to tell you Avici was operating rather than announcing.
Upper bound in the headline, uncertainty in the footnotes
The tilt is modest and it runs one way. 'Up to $1 million' leads, while the admission that estimates vary and Avici has confirmed nothing waits for the FAQ. The striking 2.7x comparison between value erased and value stolen is arithmetic we stand behind, but both of its inputs — a snapshot capitalisation and a 39% move — come from one price feed on one day, and the same account can't keep its post-crash price straight to the nearest four cents. The finding survives; the precision implied by the framing does not.
Careful wording upstream, self-citation downstream
Note the word choices. Avici, facing a possible repayment obligation it may not be able to fund from a $3.39m token, described a theft as an 'issue affecting card balance withdrawals' and stopped there — the least costly true statement available. On the reporting side, Cryptopolitan builds its comparison set largely from its own prior coverage of Sandbox and Moonwell, closes with a newsletter pitch, and appends a trading disclaimer, which is the standard shape of price-driven crypto coverage rather than anything unusual to this story.
Direction firm, magnitudes soft
We are confident about the shape of this — a breach happened, users lost funds, the token fell hard, and the company has said almost nothing — and unconfident about nearly every figure attached to it. One publisher, an unconfirmed loss spanning a factor of nearly two, and a post-crash price quoted three ways in one piece leave the arithmetic directionally sound and decimally unreliable. A confirmation from Avici or a second outlet's on-chain tally would move this quickly.