Video Script: “Why Silicon Valley
Bank Collapsed in 48 Hours”
Target length: 10-12 minutes Format: Case study / analysis —
written to be read out loud, start to finish, like telling a story
0:00 — Hook
“A bank holding over $200 billion just fell apart. Not over
months. Not over weeks. In 48 hours. That’s what happened to
Silicon Valley Bank in March 2023, and for a few days, people
genuinely feared it could take the rest of the banking system
down with it. So how does a bank that looked totally fine just
weeks earlier collapse that fast? To understand that, the story
actually has to start a few years earlier — with startups getting
rich, not banks getting reckless.”
0:15 — Setting the scene
Silicon Valley Bank wasn’t an average bank. It was THE bank for
startups — tech founders and venture capital firms overwhelmingly
kept their money there. About half of all VC-backed startups in the US
banked with SVB. And 2020 and 2021 were wild years for startup
funding — VCs were throwing money at tech companies left and right,
and a lot of that cash landed straight in SVB’s accounts.
That left SVB sitting on a mountain of deposits. And like any bank,
that money couldn’t just sit there — it had to be put to work. There
were two broad options: make a bunch of loans, which is riskier, or
buy something considered about as safe as it gets — long-term US
government bonds. SVB went with the “safe” option.
That decision is exactly what ends up sinking them. Here’s why.
2:00 — Domino 1: interest rate risk and
the mechanics of the bond loss
Through 2022, the Fed raised its main interest rate from near zero to
over 4% in about a year — the fastest pace of rate hikes in decades,
and the same hikes that pushed up mortgage rates and credit card
costs during that stretch. The core mechanism to understand: bond
prices and interest rates move in opposite directions. A bond locks in
a fixed rate — buy one paying 1%, and that’s the rate for the life of the
bond. If new bonds a year later pay 5%, nobody wants the old 1% one
anymore unless its price drops enough to make up the difference. So
when rates go up fast, existing bond prices fall.
There’s a second factor that made this worse for SVB specifically: how
long until the bond matures. Longer-term bonds lose more value when
rates rise than short-term ones do — the longer money is locked in at
the old rate, the more painful it is once new options pay better. SVB
had put a large chunk of its money into longer-term bonds during
2020-2021, exactly when rates — and the returns on offer — were at
rock bottom. That combination — long-term bonds, bought right
before rates spiked — meant SVB’s holdings took a bigger hit than
most.
Here’s the part that let the problem stay hidden for a while: under
standard accounting rules, a bank doesn’t have to report a loss on a
bond just because its market value dropped — only if it actually sells
the bond. So the loss was real, it just wasn’t “official” yet. It only
becomes official the moment the bank is forced to sell.
That’s the key word: forced. Because the next domino is exactly what
forces it.
3:30 — Domino 2: short-term cash needs
meet long-term investments
2022 was also a brutal year for startup funding — venture investment
dropped sharply from its 2021 peak, and cash-burning startups that
had raised big rounds during the boom now had no fresh money
coming in. Their only source of operating cash was the deposits
already sitting at SVB, so withdrawals started outpacing new money
coming in — a complete reversal of the growth SVB had seen the
previous two years.
This is where a basic mismatch became a real problem. Deposits can
be pulled out at any time — that’s the nature of a bank account. But
the bonds SVB bought with that money were locked in for the long
term, and could only be turned back into cash early by selling at a
loss. As long as deposits stayed steady, that mismatch never mattered,
because SVB never needed to sell early. But once deposits started
shrinking, the mismatch became the whole problem: SVB needed cash
now, but its money was tied up for years — and the only way to get
that cash was to sell the bonds and take the loss from Domino 1 for
real.
On March 8, SVB sold roughly $21 billion worth of bonds, locking in
that $1.8 billion loss, and announced it needed to raise $2.25 billion to
shore up its finances. That announcement turned a quiet, on-paper
problem into a public, confirmed one — and once a problem is public,
markets react instantly.
5:00 — Domino 3: the run, and why speed
was the real problem
The March 8 announcement worked like a public warning sign, and
what followed was people acting in their own self-interest, all at once:
anyone with more than $250,000 at SVB — the FDIC’s insurance limit
— had a real financial reason to pull their money out before the bank
potentially failed, since anything above that amount isn’t guaranteed
unless regulators later step in to cover it. Some of the biggest VC
firms openly told their portfolio companies to withdraw, and because
SVB’s customer base was so concentrated in the startup world, a
small number of influential firms could set off withdrawals across
hundreds of companies at once, instead of depositors acting on their
own one by one.
This is where the “internet-speed” bank run stops being just a colorful
detail and becomes the actual reason things moved so fast. Older
bank runs — think the 1930s, or even the slower institutional panic of
2008 — were limited by how quickly information and instructions
could physically travel. In 2023, one tweet or group chat message
reached thousands of people within seconds, and each of them could
move their money with a few taps on a phone. The result: $42 billion
in withdrawal requests in a single day — roughly a quarter of
everything SVB held — a scale and speed that had never really
happened before. Regulators shut the bank down the next morning
because there was no way to raise cash fast enough to keep up with
outflows that size.
7:00 — Why this wasn’t just SVB’s
problem
Once SVB went down, fear spread to other banks too — were they
sitting on the same hidden bond losses? Turns out some were:
Signature Bank collapsed just days later. Making it worse, the
government normally only guarantees deposits up to $250,000, and
most of SVB’s customers were companies holding way more than that
— meaning most of that money technically wasn’t protected at all.
To stop the panic from spreading further, the Fed and FDIC did
something unusual: they guaranteed all deposits at SVB, not just the
insured portion. Worth drawing a clear line here so this doesn’t just
sound like “2008 again” — 2008 was about banks making terrible
loans to people who couldn’t pay them back. This was different. SVB
never made bad loans. It simply never protected itself against the
possibility that interest rates could rise. Different disease, similar-
looking symptoms.
9:00 — What this actually teaches
Strip away the headlines, and what’s the actual lesson here? [Pick the
angle that feels like the strongest take — don’t just read all three, go
deeper on one]
One angle: SVB didn’t do anything that sounds reckless from the
outside. No crypto bets, no sketchy loans. It simply didn’t plan for a
fairly predictable scenario — interest rates going up — and that
“boring” oversight is what ended up sinking it. Sometimes the risk
that causes the most damage isn’t the flashy one, it’s the one nobody
bothered to double-check.
Another angle: SVB’s entire customer base was concentrated in one
industry — tech startups. When that industry had a bad year, the bank
had a bad year right along with it. It’s the same reasoning behind not
putting all investments in one place — concentration risk applies to
banks just as much as to a personal portfolio.
A third angle: this was arguably the first bank run that happened at
internet speed. Historically, bank runs built up over days or weeks.
This one happened over group chats in about 48 hours — which
means the old assumptions about how much time regulators have to
react may no longer hold.
11:00 — Close
“SVB looked completely fine right up until the moment it wasn’t
— and that’s the part worth remembering. It wasn’t fraud. It
wasn’t some obvious red flag everyone missed. It was a risk
nobody hedged against, catching up all at once. Next up: [next
topic] — drop a comment if there’s a company or a collapse
worth digging into next.”
Sources to pull real facts/figures from
SVB’s own public filings and March 2023 press releases
FDIC and Federal Reserve official statements from March 2023
Reputable financial news coverage (Reuters, Bloomberg, WSJ, FT)
for the day-by-day timeline
The Federal Reserve’s own post-mortem report on SVB’s failure
(published April 2023) — a great primary source for the “boring
risk” angle
Tip: keep a running doc of every source/link used per video — useful
for on-screen citations, and it can come in handy if this ever comes up
in a uni interview.