Alan Lane's claim is a lagging indicator. The former CEO of Silvergate Bank now says the Biden administration ran a "coordinated attack" that forced the institution into liquidation. The statement has been recycled for weeks. It is partly true. It is mostly irrelevant.
Here is the datum that matters. In Q4 2022, Silvergate's deposits fell from $13.4 billion to $3.8 billion. A 71% single-quarter drain. Roughly 90% of that base came from digital asset customers. When your customers are one correlated cohort, your deposit base is one correlated cohort. The sentence is tautological. So is the death spiral.
The run did not begin with a regulator's phone call. It began with a balance sheet that had no diversification left to offer.
Silvergate was not an exchange. It was a bank. But its balance sheet behaved like a leveraged bet on a single sector, and its most valuable product was not lending — it was settlement.
The Silvergate Exchange Network, or SEN, was the piece that mattered. A 24/7 API that let crypto exchanges and institutional clients move fiat between each other in seconds, bypassing the ACH and SWIFT rails that close on weekends and holidays. For roughly six years, SEN was the closest thing the industry had to institutional settlement infrastructure. It was, in effect, a private payment network wearing a bank charter.
That is the technical reality behind the political story. Between 2017 and 2021, Silvergate's total deposits grew from under $2 billion to more than $14 billion, sourced almost entirely from exchanges, market makers, and stablecoin issuers. The bank earned fee income on SEN transfers. Then it redeployed the deposits into securities — mostly agency MBS and short-duration bonds.
The structure broke in the gap between those two facts. Deposits were callable on demand. Assets were not. This is the oldest failure mode in banking. It has a name. It is not "coordinated attack."
By the end of 2022, the bank was selling assets to meet withdrawals. Then it was selling assets at a loss. Then it was selling assets at a loss it could not absorb. In Q4 2022 the bank reported a net loss of $1 billion. It cut 40% of its staff. It discontinued its digital asset business. By March 2023, it announced a voluntary liquidation.
The proximate triggers were public: the collapse of FTX in November 2022, the resulting panic, and a bank run that no amount of goodwill could stop. The regulatory pressure came after the run, not before it.
From Bogotá, the settlement story looks different. Latin American exchanges and remittance corridors depended on SEN for dollar access more than their US counterparts did. Local banks could not provide 24/7 dollar settlement. Stablecoin rails were expensive relative to a SEN transfer. When Silvergate liquidated, the corridors did not find a substitute; they found a workaround. That distinction matters, and it is the reason I tracked the wind-down not as a US banking story but as a plumbing failure in the cross-border layer.
Here is where the "coordinated attack" narrative collapses under its own arithmetic. Compare Silvergate's deposit composition to a commercial bank of similar size. A mid-sized regional with $14 billion in deposits typically draws from thousands of unrelated customers — retail, small business, municipal, corporate. No single shock can drain more than 15-20% of the base. Silvergate drew from a few hundred institutions whose fortunes were correlated to the same asset class. That is not a bank. That is a closed-end fund with a checking account.
I spent part of 2017 auditing token models for projects that raised over $50 million in aggregate. The lesson then was the same lesson Silvergate taught later: concentration is not a feature. It is a time bomb with a variable fuse.
The Decay-Cycle, in Silvergate's case, ran like this. Deposit growth phase, 2017-2021: the bank added SEN customers, collected fees, and grew deposits faster than it could deploy them profitably. Peak concentration phase, 2021-2022: deposits reached a high of $16.3 billion in Q1 2022. The securities book grew to match. Duration mismatch widened. Stress phase, Q4 2022: withdrawals forced asset sales, which crystallized losses, which reduced capital, which accelerated withdrawals. Liquidation phase, Q1 2023: the bank admitted the model was unrecoverable.
Liquidity evaporates faster than hype. That was the sequence. Not a conspiracy.
Let me be precise about what regulation actually did. The OCC's interpretive letters from 2020-2021 had allowed national banks to custody crypto and hold stablecoin reserves. Silvergate was not an OCC-chartered bank; it was a California state-chartered institution under the Federal Reserve. Its crypto exposure was legal. Its SEN network was legal. Its stablecoin reserve relationships — with issuers like Circle — were legal.
The regulatory pressure that did arrive in early 2023 came in the form of joint statements from the Fed, FDIC, and OCC warning banks about crypto-related liquidity risks. Those statements did not shut Silvergate. They described the risk that had already materialized.
Regulation lags, but penalties lead. Lane's framing inverts the order. The penalty was self-inflicted. The regulation merely documented it.
Then there is the part he did not mention. SEN's economics were never sustainable as a standalone bank product. The network's fee income was modest relative to the deposit float it justified. In 2022, Silvergate's total fee revenue was roughly $48 million. The deposit base that supported it — $14 billion — generated interest income only if the bank held duration. Holding duration meant taking rate risk. Rate risk meant mark-to-market losses when the Fed hiked. The Fed hiked 425 basis points between March 2022 and December 2022.
Do the arithmetic. A $14 billion securities book with a modest duration of four years loses roughly $2.4 billion in mark-to-market terms across a 425 basis point move. Silvergate's entire equity capital was less than $1.5 billion at the start of that window. The bank was insolvent on a mark-to-market basis before the first withdrawal.
SEN's transaction volume at peak ran north of $50 billion per quarter. The network processed real settlement for real institutions. But volume is not the same as revenue. The bank took a per-transaction fee that generated a small fraction of the deposit float. When the float collapsed, the fee business had nothing to stand on. The rail needed the deposit base to be profitable; the deposit base needed the rail to be sticky. Each dependency made the other more fragile.
SEN's role in stablecoin reserves deserves specific attention. Circle held a portion of USDC reserves at Silvergate at the time of the wind-down. The bank was not merely a settlement rail for speculators; it was a reserve custodian for a top-three dollar stablecoin. When it entered liquidation, that reserve exposure did not fail because the deposits were segregated, but the operational complexity of moving billions of dollars of backing overnight was significant. The fact that the stablecoin did not break is a credit to segregated custody, not to the bank.
This is the second fact the political narrative cannot absorb. The bank was not killed by a coordinated attack. It was killed by its own duration book and by a deposit base that reflected the same asset class it was holding.
The correct design for a crypto settlement bank would have been narrow. Hold customer deposits in short-duration, high-liquidity instruments — T-bills, overnight repo, central bank reserves. Earn a thin net interest margin. Monetize the rail, not the float. Silvergate did the opposite. It monetized the float and treated the rail as a customer acquisition tool. The fee income justified the deposit relationships; the deposit relationships funded the securities book. The tail wagged the dog, and when the tail was cut, the dog died.
I saw the same mechanical failure in 2022, in the Terra-Luna system. The feedback loop between staking rewards and peg maintenance was not a design flaw in the sense of a coding bug. It was a structural artifact of a system whose parts all depended on the same assumption — that demand would keep flowing in. Remove the assumption and the system collapses at the speed of the withdrawal queue. Silvergate collapsed at the speed of the wire transfer. Same mechanics. Different wrapper.
The parallel extends to speed. In the Terra system, the UST redemption queue ran through the Burn/Mint mechanism at the speed of block confirmation. In the Silvergate system, deposit redemption ran through Fedwire at the speed of the banking day. Neither system had a circuit breaker. Neither system had a reserve buffer sized to the correlated flows. The difference was the wrapper — stablecoin versus bank charter — and the wrapper turned out to be thin.
Here is the counter-intuitive reading. The Silvergate liquidation did the crypto industry a favor it will not acknowledge.
The bank was a single point of failure for USD settlement across exchanges, market makers, and stablecoin issuers. If SEN had survived as a going concern into 2023 and 2024, it would have continued to be the hidden plumbing for a market that now manages trillions in notional flow. The industry outsourced its most critical infrastructure to one state-chartered bank with one asset class for a customer base. That is not resilience. That is counterparty risk on rails.
The industry's response — building direct fiat on-ramps through multiple banking partners, expanding stablecoin settlement networks, integrating non-USD rails — is slower, uglier, and safer. It is also unavoidable. Silvergate's death accelerated the transition that its existence had been delaying.
The 2024 spot ETF approvals did part of this work. Institutional dollars can now access Bitcoin exposure without touching a crypto-friendly bank. The custodian is a qualified institution, the settlement is T+1 through the DTCC, and the counterparty risk is to a systemically important institution with a duration book measured in hundreds of billions and a central bank backstop. Silvergate's fate is the argument for regulated wrappers.
Code is law until the wallet is empty. The corollary here is that rails are only as durable as their reserve base. Silvergate's SEN was fast. It was not robust. Those are different properties, and the market learned the difference at a cost of several billion dollars in depositor stress.
Nor is the "coordinated attack" argument likely to produce anything useful. Assume Lane is correct — assume the Biden administration targeted crypto-friendly banks. What changes for anyone building in 2026? Nothing structural. The lesson remains the same: if you build your settlement layer on a fractional-reserve bank, you inherit the bank's duration risk. The political party in power is a variable. The mechanics of a deposit run are a constant.
The signal to watch is not political. It is the composition of the next generation of crypto settlement rails. If they are consolidated into two or three large custodians with deep but single-sector deposit bases, the Silvergate failure was a rehearsal. If they are fragmented across stablecoin issuers, multiple banking partners, and non-USD corridors, the industry learned something.
Track the concentration ratio of USD settlement, not the election cycle. Volatility is the fee for entry. Structural fragility is the fee for forgetting.