Forty-Two Hours: The Bank Run That Broke Silicon Valley

Silicon Valley Bank was founded in 1983 above a pizza parlor in San Jose, the brainchild of Bill Biggerstaff and Robert Medearis, two men who saw an underserved niche: the founders and venture capitalists building technology companies in the orchards-turned-office-parks of Santa Clara County. Traditional banks didn't understand startups. They didn't know how to underwrite a company with no revenue, no collateral, and a pitch deck full of hockey-stick projections. SVB did — or at least, it was willing to try.

By the time Greg Becker took over as CEO in 2011, SVB had become something more than a bank. It was an institution, a rite of passage. If you raised a Series A from a top-tier VC firm, you opened your operating account at SVB. It was where you wired your payroll, where you parked your runway, where you got your venture debt line. The bank didn't just serve the tech ecosystem — it was woven into its connective tissue. SVB banked nearly half of all U.S. venture-backed startups. Its client list read like a CrunchBase leaderboard: Roku, Roblox, Etsy in their early days. The bank's annual wine-tasting events and ski trips were networking fixtures for the Sand Hill Road crowd.

Becker, a former commercial banker who'd joined SVB in 1993, understood the flywheel. VCs deposited their fund capital at SVB. Their portfolio companies deposited their raised capital at SVB. SVB lent back to those companies and earned fees on everything in between. It was a beautiful, self-reinforcing loop — as long as the tech ecosystem kept growing.

And grow it did. In 2019, SVB held roughly $60 billion in deposits. Then the pandemic hit, and something extraordinary happened. The Federal Reserve slashed interest rates to near zero. Venture capital flooded into startups at historic volumes. Companies that might have raised $20 million raised $200 million. And all that cash — billions upon billions of it — flowed into SVB's coffers. By the end of 2021, SVB's deposits had ballooned to over $189 billion, more than tripling in just two years.

Becker's bank was drowning in cash. The question was what to do with it. The answer he chose would destroy everything.

The problem with having too much money is that you have to put it somewhere. SVB couldn't lend it all out — there weren't enough creditworthy borrowers to absorb $189 billion. So the bank's treasury team did what seemed prudent, even conservative: they bought U.S. Treasury bonds and mortgage-backed securities guaranteed by the federal government. These were, by any traditional measure, the safest assets on Earth. The U.S. government had never defaulted on its debt. The bonds would pay a steady coupon. What could go wrong?

What SVB's team did not adequately account for was duration. The bank loaded up on long-dated securities — bonds that wouldn't mature for ten, fifteen, even twenty years. In a world of near-zero interest rates, these bonds offered slightly higher yields than short-term alternatives. The spread was modest, but on tens of billions of dollars, even a small yield advantage translated into meaningful income. SVB classified the bulk of these purchases as "held-to-maturity," an accounting designation that allowed the bank to carry them on its books at face value regardless of what the market was actually willing to pay for them.

This was the trap, though almost no one saw it at the time. Duration mismatch — the practice of funding long-term assets with short-term liabilities like deposits — is the oldest risk in banking. It's what killed savings and loans in the 1980s. It's what nearly brought down Continental Illinois. The math is brutally simple: when interest rates rise, the market value of existing long-duration bonds falls. If you can hold those bonds to maturity, you'll eventually get your money back. But if something forces you to sell early, you crystallize losses that were previously only theoretical.

Through 2021 and into early 2022, SVB's portfolio looked like a masterstroke. The bank was earning steady returns on government-guaranteed paper. Its stock price climbed above $700 per share. Becker was named Banker of the Year by a prominent financial publication. Inside the bank, the mood was triumphant.

Then, in March 2022, the Federal Reserve began raising interest rates. And it didn't stop. By early 2023, the Fed had hiked rates at the fastest pace in four decades. Each increase was like turning up the heat under a pot that SVB's executives hoped would never boil. The market value of SVB's bond portfolio cratered. By the end of 2022, the bank was sitting on more than $15 billion in unrealized losses — paper losses that didn't show up on the income statement because of that held-to-maturity classification, but that were disclosed in the footnotes of its financial filings for anyone who cared to look.

Almost no one looked. Until someone did.

The tech downturn of 2022 had been quietly draining SVB's lifeblood. Venture funding slowed. Startups that had raised enormous rounds in 2021 were now burning through that cash without raising new rounds. Deposits at SVB began to decline — not catastrophically, but steadily, as companies withdrew money to fund operations. SVB needed liquidity.

On Wednesday, March 8, 2023, SVB announced that it had sold $21 billion worth of bonds from its available-for-sale portfolio, booking a $1.8 billion after-tax loss. To shore up its balance sheet, the bank simultaneously announced a $2.25 billion capital raise — a stock and convertible debt offering that it hoped to complete within days. The message was supposed to be reassuring: we're being proactive, we're raising capital, we have this under control.

The market heard something entirely different. If SVB had to sell bonds at a massive loss and immediately raise capital, how bad were things really? How much more was hidden in that held-to-maturity portfolio?

The reaction was swift and merciless. SVB's stock dropped 60% on Thursday, March 9, erasing roughly $10 billion in market capitalization in a single trading session. But the stock price was almost beside the point. What mattered was what was happening in the group chats.

Venture capitalists — the very people who had built their businesses alongside SVB for decades — began texting each other, posting on Twitter, and sending urgent messages to their portfolio companies. The advice was blunt and unanimous: get your money out. Peter Thiel's Founders Fund reportedly advised its companies to withdraw deposits. Other prominent VCs followed. The counsel spread through the tight-knit startup community with the speed of a wildfire in dry brush.

The Guardian would later call it "the first Twitter-fuelled bank run". Social media compressed what would have traditionally taken days or weeks of growing unease into a matter of hours. There were no lines of depositors snaking around the block, no grainy news footage of anxious savers clutching bankbooks. Instead, there were founders sitting in their apartments, logging into SVB's online portal, and initiating wire transfers. The panic was silent, digital, and devastatingly efficient.

By the end of Thursday, March 9, customers had attempted to withdraw $42 billion from SVB — roughly a quarter of the bank's total deposits — in a single day. The bank's cash reserves were obliterated. SVB ended the day with a negative cash balance of approximately $958 million. The capital raise was dead. The stock was in freefall. And Greg Becker's reassurance video, posted just 24 hours earlier, now looked like the last words of a man who didn't understand the fire he was standing in.

The California Department of Financial Protection and Innovation moved before dawn. On the morning of Friday, March 10, 2023, regulators seized Silicon Valley Bank and placed it into receivership under the Federal Deposit Insurance Corporation. Trading in SVB's stock was halted. The bank's website displayed a terse notice. The doors of its branches — the ones in Palo Alto, the ones in San Francisco's SoMa district, the ones near the gleaming campuses of companies it had nurtured from infancy — did not open.

It was the third-largest bank failure in United States history and the largest since Washington Mutual's collapse during the 2008 financial crisis. The entire arc from announcement to seizure had taken approximately forty-two hours.

The immediate aftermath was chaos. SVB's depositor base was unlike that of a typical bank. Most of its clients were businesses, not individuals, and the vast majority of deposits exceeded the FDIC's $250,000 insurance limit. Estimates suggested that more than 90% of SVB's deposits were uninsured. This meant that thousands of companies — startups with ten employees, startups with ten thousand — suddenly could not access the money they needed to operate.

The New York Times documented the experience of founders scrambling to make payroll with frozen accounts, counting hours until direct deposits were due, trying to explain to employees that their paychecks might not arrive. One founder described having 48 hours until payroll and $200,000 to go. It was the human wreckage of a balance-sheet abstraction: real people, real families, real rent payments, all held hostage by a bank's bet on bond duration.

And SVB was not alone. Within days, two more banks — Silvergate Bank and Signature Bank — also failed. Silvergate, heavily exposed to the cryptocurrency industry, had already been wobbling. Signature Bank, a New York-based institution with its own concentration risks, saw depositors flee in sympathy. Over the course of five days, three American banks collapsed. Global bank stocks plunged. Credit Suisse, already fragile, saw its own crisis accelerate. The word "contagion" — a term most people associated with pandemic movies — was suddenly on every financial analyst's lips.

The question hanging over Washington that weekend was existential: was this 2008 all over again? Were the dominoes about to fall? Many regional banks held similar long-duration bond portfolios with their own unrealized losses. If depositors at those banks decided to run, the same math that killed SVB would kill them too. The banking system's structural vulnerability was not unique to one institution — it was systemic.

Something had to be done. And it had to be done before Asian markets opened on Sunday night.

The weekend of March 11-12, 2023, was one of the most intense periods of financial policymaking since the fall of Lehman Brothers. Treasury Secretary Janet Yellen, Federal Reserve Chair Jerome Powell, and FDIC Chairman Martin Gruenberg worked through Saturday and Sunday to construct a response that would stop the bleeding without explicitly bailing out SVB's shareholders or management.

On Sunday evening, March 12, a joint statement from the Treasury Department, the Federal Reserve, and the FDIC announced a sweeping intervention. All depositors at SVB — including those with uninsured balances above $250,000 — would be made whole. No taxpayer funds would be used; instead, the costs would be covered by a special assessment on the banking industry. The Fed simultaneously unveiled a new emergency lending facility, the Bank Term Funding Program, which allowed banks to borrow against their Treasury and agency holdings at par value — not market value. It was an elegant, if controversial, solution: banks sitting on unrealized losses could now access liquidity without being forced to sell bonds at a loss, breaking the doom loop that had killed SVB.

The intervention worked. When markets opened on Monday morning, the acute panic subsided. Deposits at regional banks stabilized. The immediate threat of cascading failures receded. But the rescue came with uncomfortable questions.

The moral hazard was obvious. If the government would step in to protect all depositors regardless of the insurance limit, what incentive did depositors have to monitor the health of their banks? What incentive did banks have to manage risk prudently if they knew the federal safety net would catch them? Critics pointed out that SVB's executives had sold millions of dollars in stock in the weeks before the collapse. Becker himself had sold $3.6 million in shares just eleven days before the bank failed. No laws were obviously broken — the sales were conducted under pre-arranged trading plans — but the optics were devastating.

The Federal Reserve later published a scathing self-assessment of its own supervisory failures. Examiners had identified risks at SVB as early as 2021 but had been slow to escalate concerns and even slower to force corrective action. The report acknowledged that the Fed's supervisory culture had shifted toward a lighter touch in the years before the crisis, and that this shift had contributed to the failure.

Greg Becker testified before Congress in May 2023. He expressed regret but stopped short of accepting full responsibility, pointing to the unprecedented speed of the bank run and the role of social media in amplifying panic. Lawmakers from both parties were unsparing. The man who had been Banker of the Year was now a cautionary tale.

The bank itself was carved up. First Citizens BancShares acquired the bulk of SVB's assets and deposits in an FDIC-brokered deal. The SVB brand, the relationships, the four decades of institutional knowledge — all of it absorbed into a regional bank headquartered in Raleigh, North Carolina. The pizza parlor in San Jose where it all began might as well have been a thousand years ago.

In the months after SVB's collapse, the startup ecosystem quietly rewired itself. Founders who had kept 100% of their operating cash at a single institution — because their VC told them to, because everyone did, because SVB threw great parties — began splitting deposits across multiple banks. Treasury management, once an afterthought for a Series B company focused on product-market fit, became a board-level conversation. The era of the single-bank startup relationship was over.

The banking industry adjusted too, though perhaps not as dramatically as reformers hoped. Many regional banks quietly reduced their holdings of long-duration bonds, shifting toward shorter maturities that were less sensitive to interest rate movements. The Fed's Bank Term Funding Program provided a bridge, but it was designed to be temporary. The underlying tension — that banks inherently borrow short and lend long — remained the foundational architecture of the system.

What SVB's collapse revealed most starkly was a speed problem. Traditional bank regulation was designed for a world where bank runs unfolded over days or weeks, giving regulators time to intervene, arrange mergers, or orchestrate orderly wind-downs. SVB's run happened in hours. Forty-two billion dollars moved with the tap of a screen. The depositors didn't need to drive to a branch and stand in line; they needed a WiFi connection and a routing number. The regulatory framework, built for the analog age, was confronting a digital reality it wasn't designed to handle.

The irony of SVB's demise contained a paradox that economists would debate for years. The very asset that destroyed the bank — U.S. Treasury bonds — was the same asset that investors fled to in the panic that followed the bank's collapse. As fear gripped the banking sector, money poured into Treasuries as a safe haven, pushing bond prices up and yields down. The instrument of SVB's death became everyone else's shelter from the storm it created.

Some in the crypto community saw vindication in SVB's failure — proof, they argued, that centralized financial institutions were inherently fragile. But the irony cut both ways: Silvergate Bank, one of the other two banks that failed that same week, had been a critical banking partner for the crypto industry, and its collapse actually made it harder for crypto firms to access the traditional financial system.

The structural vulnerabilities that killed SVB — duration mismatch, concentrated depositor bases, and the amplifying power of social media — did not disappear with the bank. They remained embedded in the architecture of the financial system, waiting for the next combination of rising rates, concentrated risk, and digital panic to activate them again.