Silicon Valley Bank: venture-cycle funding, duration and the two-day run

A specialized franchise with a linked asset-and-liability cycle

Silicon Valley Bank was more than a place where technology companies kept cash. Banking services connected venture investors, portfolio companies and founders. That specialization created knowledge and distribution advantages, but customers shared sources of liquidity and information. When venture fundraising slowed, many depositors needed cash at the same time. A bank can have thousands of accounts yet remain concentrated in one economic network.

The Federal Reserve’s review distinguishes SVB Financial Group, with approximately $212 billion of year-end 2022 assets, from its California state-member bank subsidiary, with approximately $209 billion. The holding company’s deposits were about 94% uninsured; loans represented 35% of assets and securities 55%. Its held-to-maturity securities had a weighted-average duration of 6.2 years. These are dated balance-sheet and portfolio measures, not descriptions of the bridge bank later sold to First Citizens. [1]

The balance-sheet implication is specific. A venture-backed company’s deposit is the bank’s liability, even if the customer also borrows from it. Equity funding received by that company becomes spendable cash, not permanent bank capital. Concentration therefore existed on the funding side independently of whether the associated loans were performing.

Why low credit risk did not mean low balance-sheet risk

Government and agency-backed securities reduced exposure to borrower default, but long-dated fixed cash flows remained sensitive to interest rates. Agency mortgage securities also carry prepayment risk: when mortgage rates rise, refinancing slows and principal may come back later. A portfolio that looked liquid in ordinary markets could lose considerable value precisely when depositors wanted repayment.

Held-to-maturity accounting records qualifying debt securities at amortized cost, subject to applicable loss accounting; available-for-sale securities are measured at fair value. Neither classification creates cash. Selling a security crystallizes its sale economics, while pledging it normally yields borrowing capacity subject to valuation, margins, operational acceptance and existing liens. A deposit withdrawal must be settled at its face amount even if an asset can only be monetized below its carrying amount.

The Fed’s executive summary records the March 8 package: $21 billion of available-for-sale securities sold, a $1.8 billion after-tax loss, proposed additional term borrowing and an intended $2.25 billion capital raise. This was a combined funding-and-capital response. The planned equity raise was not completed cash available to meet the subsequent run. [2]

March 8–10: separating the announcement, outflow and unfilled demand

The sale announcement exposed the trade-off between recognizing losses and finding replacement funding. For an uninsured depositor, waiting for the bank to complete a recapitalization offered little upside relative to transferring funds. The same announcement could reassure an investor about a repair plan and alarm a customer about immediate access to payroll cash.

The Fed reports more than $40 billion of deposit outflows on March 9. Management then expected more than $100 billion of additional outflows on March 10; the bank was closed that morning. Expected or requested withdrawals are not completed withdrawals. Adding the two figures and calling the result cash that actually left before closure would overstate the measured flow. [1]

The chronology also separates a declining deposit franchise from a terminal run. Cash burn and weaker fundraising had been draining customers’ balances before March 8. The subsequent surge was a change in behavior and speed, not merely another day of the same trend. Digital transfers reduced transaction friction, while the client network made decisions highly correlated.

Scroll horizontally to see all columns.

The missing link between collateral and usable liquidity

The Federal Reserve’s critical-risk review found that contingency funding and liquidity stress testing were inadequate. Shortfalls persisted; supervisors and management sometimes treated inability to monetize assets as an operational issue rather than a binding funding constraint. The report also describes changes in modeled assumptions that reduced calculated requirements without equivalently improving the actual liquidity position. [3]

This distinction matters because a bank pays withdrawals with immediately available cash or completed borrowing, not a theoretical inventory of eligible collateral. Preparing legal documents, moving securities, establishing counterparty access and testing draw procedures are part of the economic defense. If those steps take longer than a run, nominal capacity does not meet the obligation.

A simple hypothetical illustrates the mechanism without reconstructing SVB’s exact funding gap. A bank owing $10 billion today cannot satisfy that demand with $12 billion of assets that will only become pledgeable next week. Nor does a $10 billion face-value bond guarantee a $10 billion advance. Timing and lendable value must both match the withdrawal requirement.

Governance failures were broader than one vacant position

The Fed’s review found weaknesses in board oversight, independent risk management and internal audit. Its detailed chronology records the chief risk officer’s departure in April 2022 and replacement in December; senior risk officers managed the interim function. It also documents May 2022 immediate-attention findings and the August 2022 governance downgrade. The vacancy mattered, but assigning the entire failure to one unfilled role would obscure decisions by the board, finance, treasury and business leadership. [3]

A risk committee can meet and a model can produce detailed numbers while the organization remains vulnerable. The analytical question is whether the assumptions capture the business’s actual exposures and whether breaches lead to changes in positions or funding. Process volume is a poor substitute for a credible challenge to a profitable strategy.

The supervisory record and the limits of a single-cause explanation

The Fed’s supervision chapter records 31 open supervisory findings at the end of 2022 and 54 issued from 2019. Findings included governance, liquidity and interest-rate issues as well as other risk areas; counting them does not make each equally relevant to the failure. The record shows that supervisory awareness existed before closure, alongside delayed escalation and remediation. [4]

The inspector general’s September 2023 material-loss review found that examiner resources and expertise were insufficient for the larger, more complex institution and that the supervisory-portfolio transition was ineffective. It also found inadequate scrutiny of investment exposure to rising rates. The report cited an estimated $16.1 billion insurance-fund loss at that stage, demonstrating that the original $20 billion announcement estimate was already subject to revision. [5]

The defensible causal account combines a fragile funding mix, asset-value sensitivity, poor execution and insufficient supervisory intervention. It does not require claiming that every bank holding government securities was destined to fail, or that technology alone made failure unavoidable. The speed of the final run narrowed the rescue window; it did not create the pre-existing mismatch.

What emergency protection covered

The March 12 joint statement protected depositors of SVB and Signature through systemic-risk determinations and said depositors would have access on March 13. It explicitly excluded shareholders and certain unsecured debtholders from protection. Senior management was removed. The government said losses associated with protecting uninsured deposits would be recovered through a special assessment on banks. [6]

That distinction separates continuity of deposits from survival of the original enterprise. A bridge bank preserves payment and banking operations while a receiver arranges disposition. It is not a recapitalization of the old shareholders. Similarly, a bank subsidiary’s receivership and a holding company’s bankruptcy or creditor disputes involve different estates; a statement about one is not a universal recovery rate for the other.

First Citizens bought a specified business, not every old claim

The FDIC announced the purchase-and-assumption transaction on March 26. Approximately $72 billion of bridge-bank assets were purchased at a $16.5 billion discount; roughly $90 billion in securities and other assets remained for disposition. Commercial loans were subject to loss sharing, and the FDIC received equity-appreciation rights with potential value up to $500 million. The discount was not the entire purchase price and not a separately measured taxpayer loss. [7]

First Citizens’ March 27 announcement described approximately $110 billion of assets, $56 billion of deposits and $72 billion of loans assumed, based on the then-current information provided by the FDIC. These figures use a later transaction perimeter and date than SVB’s year-end accounts. They cannot be subtracted mechanically from the year-end balance sheet to calculate pre-failure flight. [8]

For the purchaser, acquired relationships and earning assets came with integration costs, financing obligations and retained risks. For the receiver, a negotiated discount and loss-sharing structure traded uncertain future recoveries against the value of a rapid transfer. Acquisition accounting gains, contract discounts, emergency lending and the insurance fund’s eventual loss are four different measurements.

The long tail: cost estimates, assessments and unresolved claims

GAO’s September 2026 disclosure report cites FDIC estimates of approximately $19 billion of combined SVB and Signature losses as of December 31, 2025. That combined number is not an SVB-only total, not exclusively the uninsured-depositor component and not a final liquidation result. The initial $20 billion SVB estimate in the acquisition announcement was an estimate at an earlier stage. [9, 7]

The FDIC’s special-assessment materials explain that losses attributed to uninsured-depositor protection are recalculated as assets are sold, liabilities satisfied and expenses incurred. Assessment recoveries finance that specified component; they do not erase the underlying economic loss or turn a borrowing balance into a loss estimate. [10]

The FDIC’s bank-specific information page, updated in February 2026, continues to distinguish the original bank and bridge-bank receivership claims. This case study does not assert that every claim, estate or litigation has reached final distribution. A final creditor recovery would require the relevant estate, claim class and distribution record, rather than inference from the successful continuation of the SVB brand. [11]

What the later evidence adds

The FDIC’s May 2026 transaction-level study found that large uninsured depositors were much more likely to run, while fully insured retail depositors generally did not run before failure. Some large depositors with pass-through insured balances also withdrew heavily. Insurance classification alone therefore does not fully describe the behavior of an operational account or a concentrated relationship. [12]

The institution-specific conclusion is that SVB’s franchise generated both its funding advantage and its vulnerability. Sector knowledge and integrated services helped deposits arrive quickly during the boom; the same economic network made them leave together. The decisive mismatch was between those mobile liabilities and assets whose value, cash flow and monetization capacity could not adjust on the same clock.

Sources

1. Federal Reserve, Evolution of Silicon Valley Bank, April 2023Official sourceBack to text: ↑1↑2↑3↑4
2. Federal Reserve, SVB review executive summary, April 2023Official sourceBack to text: ↑1↑2
3. Federal Reserve, Supervision by Critical Risk Areas, April 2023Official sourceBack to text: ↑1↑2
4. Federal Reserve, Federal Reserve Supervision, April 2023Official sourceBack to text: ↑
5. Federal Reserve OIG, Material Loss Review, September 25, 2023Official sourceBack to text: ↑
6. Treasury, Federal Reserve and FDIC joint statement, March 12, 2023Official releaseBack to text: ↑1↑2
7. FDIC, First Citizens purchase-and-assumption announcement, March 26, 2023Official releaseBack to text: ↑1↑2↑3
8. First Citizens, acquisition announcement, March 27, 2023SourceBack to text: ↑1↑2
9. GAO, Bank Financial Disclosures, GAO-26-107719, September 3, 2026Official sourceBack to text: ↑
10. FDIC, Special Assessment Pursuant to Systemic Risk Determination, checked October 4, 2026Official sourceBack to text: ↑
11. FDIC, Silicon Valley Bank receivership information, updated February 17, 2026Official sourceBack to text: ↑
12. FDIC, transaction-level depositor-flight research release, May 14, 2026Official releaseBack to text: ↑