Credit Suisse AT1 Bonds: How CHF 16 Billion Went to Zero

March 2023. A Sunday evening in Bern.

The Swiss Federal Council signs an emergency ordinance. By the time markets open on Monday, CHF 16 billion of Credit Suisse bonds are worth nothing at all and the bondholders have lost everything. The shareholders, who in any ordinary resolution stand at the very back of the queue and absorb losses first, receive roughly CHF 3 billion in UBS stock.

That is not a market failure. Markets did not decide this. It is a rule change, executed over a weekend, and the clause that made it legally possible is sitting inside instruments that may well be in a bond ETF you own right now.

The queue, and who is supposed to stand where

When a bank fails there is a line, and it helps to think of it as the exit queue from a burning building.

Depositors and secured creditors get out first. Behind them, senior unsecured bondholders. Behind them, subordinated debt, which includes AT1 bonds. And at the very back, last out of the building, shareholders.

Shareholders stand last because they signed up for it. They own the upside, so they absorb the first losses. That is the deal, and the entire architecture of bank capital rests on it.

What Switzerland did that Sunday was pull the people at the front of the queue out entirely, and let the people at the back walk away with something.

What AT1 bonds actually are

Additional Tier 1 bonds were invented after the 2008 crisis, under Basel III, with a single purpose: absorb losses at a failing bank before public money is used. They are perpetual, they rank one step above shareholders, and because they carry a genuine risk of being written down they pay a high yield, in recent years frequently in the high single digits.

This next part is important and widely misunderstood. An AT1 bond going to zero is not a scandal in itself. It is the product working as designed. These instruments exist precisely to break, and buying one and being shocked when it breaks is like buying fire insurance and being shocked when the insurer pays out for a fire.

The scandal is not that they broke. It is the order in which they broke.

Two triggers, and only one of them fired

Every AT1 prospectus contains a viability event trigger, and it fires in one of two ways.

The mechanical trigger. If the bank’s core capital ratio falls below a defined threshold — often 7% — the bonds convert or are written down automatically. No judgement, no discretion: a number crosses a line and the instrument does what it says on the tin.

The regulatory trigger. The regulator declares the bank non-viable. This is not automatic. It is a judgement call.

Now hold those two against the facts.

Two triggers — only one can fire on its own

An AT1 bond can be written down two ways: a number crossing a line, or a regulator's judgement. At Credit Suisse, the number never came close.

Mechanical trigger fires at ≈ 7%

Credit Suisse's actual CET1, Q4 2022 14.1%

Roughly double the level that triggers an automatic write-down. The mechanical trigger was never within reach.

Mechanical trigger

CET1 drops below ~7% → automatic write-down. A number crosses a line. No judgement.

Never came close.

Regulatory trigger

The regulator simply declares the bank “non-viable”. A judgement call.

The door FINMA used — under an ordinance signed hours earlier.

Credit Suisse held roughly double the capital it was required to hold. The mechanical trigger never came within reach of firing.

So FINMA used the second door. It declared the bank non-viable, and to do so it did not rely on existing banking law. It relied on Article 5a of an emergency ordinance the Federal Council had signed hours earlier, the same Sunday evening. A legal basis created in real time, for a write-down executed the same night.

This was a liquidity crisis, not a capital crisis

The distinction matters more than almost anything else here, and it is the one the court would later fasten onto. Credit Suisse did not run out of capital. It ran out of confidence — and therefore of funding.

The chain is short and brutal:

Not a capital crisis — a run

1. Mar 2021

   Archegos defaults

   One client blows up. Credit Suisse loses ~USD 5.5bn in a week. The buffer is gone.

2. Q4 2022

   The run begins

   ~CHF 110bn of client assets walk out the door in a single quarter.

3. 15 Mar 2023

   “Absolutely not.”

   The Saudi National Bank refuses to add capital. CDS spreads blow past 1,000 basis points.

4. 16 Mar 2023

   A CHF 50bn lifeline

   The Swiss National Bank extends emergency liquidity. It buys days, not confidence.

5. 19 Mar 2023

   Gone in a weekend

   UBS buys Credit Suisse for ~CHF 3bn — about CHF 0.76 a share.

Capital never ran out. Confidence did — and confidence has no ratio on any balance sheet.

A bank can be perfectly solvent on paper and still die, because banking runs on the assumption that not everyone asks for their money at once. When that assumption breaks, capital ratios are cold comfort.

Archegos itself — how one banned trader built roughly $160bn of hidden exposure through Credit Suisse and five other banks, on a single instrument — is a story of its own.

But note what this means. The thing that killed Credit Suisse was not the thing the AT1 trigger was designed to detect.

What the Swiss court eventually said

For two and a half years this sat as a grievance without a verdict. Then, on 1 October 2025, Switzerland’s Federal Administrative Court ruled, and it did not split the difference.

The court found three things:

1. The write-down lacked a sufficient legal basis. FINMA’s order was revoked.
2. Credit Suisse met its capital requirements at the time. The court cited an internal Credit Suisse email from 19 March 2023 stating that the measures being taken were for confidence and liquidity, not capital.
3. Article 5a of the emergency ordinance was constitutionally invalid. The Federal Council had delegated emergency powers to FINMA without the authority to do so.

Around 3,000 claimants, across roughly 360 proceedings, had brought the challenge.

And yet nothing has been paid

Here is where it stands as of July 2026, and it is the part most coverage gets wrong by implying the case is over.

FINMA appealed. UBS appealed. The Swiss Federal Supreme Court granted the appeals suspensive effect, which means the lower court’s ruling has, for now, no legal force. The bonds remain at zero and not one franc has been returned to anyone.

Separately, in January 2026, Switzerland began facing investor-state claims under international investment treaties from foreign bondholders. That is a second front — and it could ultimately land the bill with the Swiss state rather than with UBS.

This case is not closed.

Was this a Swiss aberration or a global precedent?

This is the question that should determine whether you change anything in your portfolio, so it deserves a straight answer. Start with the comparison case.

Same instrument, two failures, opposite order

AT1 bonds are built to be written down. The question is never whether they break, but who breaks first.

2017

Banco Popular, Spain

AT1 bondholders

Written off in full

Shareholders

Written off in full

The queue held. Everyone at the back lost everything too.

2023

Credit Suisse, Switzerland

AT1 bondholders

CHF 16bn written to zero

Shareholders

Roughly CHF 3bn in UBS stock

The queue inverted. The back of the line walked out ahead of the front.

Losing everything is what an AT1 bond is for. Losing everything while the people ranked behind you keep three billion is a different event, and it is the one that needed an emergency ordinance signed the same evening.

When Spain’s Banco Popular failed in 2017, AT1 bonds and equity were written off together. Bondholders lost everything, but so did shareholders, and nobody at the back of the queue got out ahead of the people in front.

Then look at how fast everyone else distanced themselves. Within 48 hours, ECB Banking Supervision, the Single Resolution Board and the European Banking Authority issued a joint statement putting the principle in writing: common equity instruments are the first ones to absorb losses, and only after their full use would Additional Tier 1 be required to be written down. The Bank of England said the same. Singapore and Hong Kong followed.

That speed was not a courtesy. It was damage control on a European AT1 market worth hundreds of billions, and it worked: the market reopened within months.

So one jurisdiction, one exception, produced under emergency law. But do not take too much comfort from that. The lesson is narrower and more uncomfortable than “AT1 bonds are safe outside Switzerland”:

A government can rewrite the loss hierarchy faster than a market can reprice it. The contract you rely on is only as durable as the state’s willingness to be bound by it on a bad weekend.

That is not a reason to panic. It is a reason to know what you own.

What this means for your portfolio

Check

Open the factsheet, or better the full holdings list, of every bond ETF and bond fund you own, and search for three terms:

• AT1
• CoCo
• contingent convertible

Then read the mandate. If the fund does not explicitly say senior debt only, subordinated financial debt may be permitted.

The obvious place to find AT1 exposure is a specialist product, and an AT1 capital bond ETF does what it says. The non-obvious places are worth five minutes of your evening: broad subordinated financials funds, some European financials credit funds, and certain high-yield strategies, where AT1 sits quietly inside the sleeve because it offers exactly the yield the mandate is reaching for. It is the same read-the-factsheet discipline behind choosing any ETF well.

Then decide, honestly

If you hold AT1 exposure and can explain the write-down mechanics to another person without looking anything up, fine. You are being paid a high single-digit yield for a specific, understood tail risk, and that is a legitimate position for someone who has consciously chosen it.

If you cannot explain it, you are not being paid for risk. You are being paid for not having read the prospectus, and that is a fundamentally different trade.

Watching for stress, honestly

You cannot stop a government from rewriting the rules. You can sometimes see the pressure building before the decree gets signed.

• High-yield credit spreads (the ICE BofA option-adjusted spread, free on FRED). Sustained moves above roughly 300 basis points have historically signalled elevated stress; above 500, you are typically in crisis conditions. Credit Suisse’s own CDS spread had blown past 450bp weeks before the write-down. The signal was there.
• The VIX term structure. When spot VIX rises above the three-month future, the curve inverts, a pattern that has tended to appear ahead of major sell-offs including 2008, 2020 and March 2023.

Neither is a crystal ball, and anyone who tells you otherwise is selling something. They are context, not commands. But they are free, and free context beats none.

The uncomfortable conclusion

Credit Suisse existed for 167 years. It was dismantled over a single weekend.

CHF 16 billion of bonds written to zero by emergency ordinance, while shareholders walked away with roughly CHF 3 billion. A court that called it unlawful. A government that called it necessary. Three thousand claimants and, three years on, not a single franc returned.

The queue exists. The hierarchy is real, it is written into prospectuses and into law, and in the overwhelming majority of cases it holds. But when the building is genuinely on fire, it is governments rather than contracts that decide who gets out first.

You cannot legislate against that. What you can do is know, precisely, where in the queue each thing you own is standing. That takes about five minutes and a fund factsheet, and it is the most useful thing this entire story has to offer you.