SVB倒闭复盘：企业现金管理从单账户到多账户的实战转变

I got the email the other day. Frank Holding Jr., Marc Cadieux and Jesse Hurley, letting me know that starting in early October, the rollout of a “united brand strategy” begins. And the SVB brand will now be retired.

Translated: Silicon Valley Bank Technology and Healthcare Banking becomes First Citizens Innovation Banking. SVB Global Fund Banking becomes First Citizens Fund Banking. SVB Go becomes Go by First Citizens Bank. Your ABA numbers, account numbers and passwords are all fine. No action required.

What the email describes is the end of a 43-year-old brand that was, for most of those years, the most important financial institution in the startup economy. It did a lot of good.

It also blew itself up and took $10,000,000 of our money with it for a weekend.

1983 to 2019: The bank that would lend to companies nobody else would touch

The idea came out of a poker game at Pajaro Dunes in the early 1980s. Bill Biggerstaff, a Wells Fargo executive, and Robert Medearis, a Stanford professor, thought the companies getting funded up and down Sand Hill Road needed a bank that understood them. The first office opened in San Jose on October 17, 1983, with Roger Smith as founding CEO.

The first decade almost ended it. By the early 1990s roughly half the loan book was commercial real estate, and when the California market turned, SVB posted a loss in 1992. They pulled real estate down to under 10% of loans within three years and went all in on the innovation economy instead.

That pivot set up the next 25 years. SVB built an underwriting model for companies with no profits, often no revenue, and an asset base consisting almost entirely of a term sheet and a team. They invented venture debt as a real product category. They lent early to Cisco and Bay Networks. They opened in Israel in 2008, the UK and a China joint venture in 2012, then Europe and Canada.

By the end, SVB banked close to half of all US venture-backed technology and life sciences companies. It banked 55% of venture-backed tech and healthcare IPOs in 2021 and 44% in 2022. If you started a company in the US between 1995 and 2022, there is a very good chance your first business checking account, your first credit card, your first line of credit and your first venture debt facility all came from the same place.

Nobody else wanted that business. SVB took it for four decades and was good at it. Founders got a bank that didn’t ask why a company with $400K in revenue was burning $900K a quarter.

2019 to 2023: A $200 billion bank with no chief risk officer

Total assets went from about $71 billion in 2019 to more than $211 billion by 2021. Roughly tripling in two years.

The deposits came in faster than anyone could deploy them into loans, so SVB put an enormous share into long-duration securities at the exact bottom of the rate cycle. Then rates went up nine times in a year.

What the Fed’s own post-mortem found, in its words, was a textbook case of mismanagement:

• No Chief Risk Officer for roughly eight months heading into the failure. At a bank with more than $200B in assets and one of the most concentrated deposit bases in America.
• The interest rate hedges came off. The bank had them at one point. They were not in place when it failed.
• 31 open supervisory findings at the end of 2022, including a November 2022 finding that SVB’s own interest rate risk simulations were unreliable and gave a false sense of safety in a rising rate environment.
• Risk managed for short-run profit. The Fed’s review found SVB was positioned to protect against rates going down, not up.
• Deposit concentration that was total. Overwhelmingly uninsured, overwhelmingly from one industry, and every single client connected to every other client through group chats and VC portfolio-wide emails.

On March 8, 2023, SVB sold about $21 billion of securities at a roughly $1.8 billion after-tax loss and announced it would raise $2 billion. On March 9, clients attempted to pull $42 billion in a single day. On March 10, the California DFPI closed it and appointed the FDIC as receiver. Forty years of building, and the run and the seizure took two days. It was the second-largest bank failure in US history at the time.

Where is Powell? Where is Yellen? Stop this crisis NOW. Announce that all depositors will be safe. Place SVB with a Top 4 bank. Do this before Monday open or there will be contagion and the crisis will spread.

— David Sacks (@DavidSacks) March 10, 2023

SaaStr itself had $10 million there

We had roughly $10,000,000 sitting at SVB when the bank went down. That was, at the time, essentially all of our operating cash. Payroll. Vendor deposits for the Annual. Everything.

Nobody I know outsmarted this. There was no clever treasury policy that saved us. There was a Thursday where the wire didn’t go through, a Friday where the bank was gone, and a Saturday and Sunday where a very large number of people who run real businesses did the math on whether they could make payroll on Monday with 15% of their cash.

We got the $10M back. Not because we were smart, and not because the system worked as designed. We got it back because on Sunday, March 12, the Treasury Secretary, the Fed Board and the FDIC Board invoked the statutory systemic risk exception and guaranteed every depositor, insured and uninsured, in full.

That was a policy decision. It could have gone the other way. FDIC insurance covers $250,000 per depositor, per bank, per ownership category. Our $10M was uninsured. On the merits of the rulebook we had written our own exposure and we owned it.

Who actually paid: protecting uninsured depositors at SVB and Signature cost the Deposit Insurance Fund an estimated $16.3 billion, and the FDIC recouped it through a special assessment on the 110 banks holding more than $5 billion in uninsured deposits, at 3.36 basis points a quarter for eight quarters, with the first $5 billion exempt. Shareholders and certain bondholders were wiped out. So it was not a taxpayer bailout in the direct sense, and it was not a rescue of SVB as a firm. It was a depositor guarantee, funded by other banks.

It was still intervention. Without it, SaaStr takes a haircut on eight figures, and thousands of startups miss payroll in the same week. Every founder who says “we were fine” was fine because of a phone call in Washington on a Sunday.

TIMELINE OF SILICON VALLEY BANK, $SIVB, COLLAPSE:

1. Bank run begins with $91 billion in bonds facing interest rate risk

2. $21 billion bond portfolio sale announced, $1.8 billion lost

3. SVB announces $2.3 billion share sale to cover bond losses

4. SVB fails to raise capital…

— The Kobeissi Letter (@KobeissiLetter) March 12, 2023

Every startup now runs two bank accounts

Essentially every venture-backed company now runs at least two banking relationships. Sweep products that spread FDIC coverage across dozens of partner banks went from a niche treasury feature to table stakes. Cash that used to sit in one operating account now sits split across an operating bank, a fintech, and T-bills.

The share shifted with it. JPMorgan went after the vacuum aggressively, added hundreds of bankers to its innovation economy team, and is still building the business out in 2026. Mercury and Brex took the new-company formation cohort on user experience and speed of account opening. First Citizens kept a real business, especially in fund banking and venture debt, but mostly with the clients it already had rather than the ones being incorporated this quarter.

No single institution holds half of the startup ecosystem’s cash anymore, and none is likely to again. That costs founders an afternoon of treasury setup they didn’t used to have to do, and it’s worth it.

October: the name comes off, the FDIC note is still being paid