KYC Breaks Our Revenue Neck

Fifteen million leaked Deutsche Bank documents describe a bank where compliance was a cost centre and revenue was the point. The memo that explains it is six words long.
KYC Breaks Our Revenue Neck

KYC Breaks Our Revenue Neck

In 2017, Deutsche Bank’s future chief executive wrote six words in an internal email that explain the largest banking leak in history. The documents that surfaced this week do not describe a bank that broke its rules. They describe a bank where the rules were never the point.

On 17 November, more than fifty newsrooms published the first stories from a cache of fifteen million internal Deutsche Bank documents. Nine million of them are emails.

The material did not come from a regulator or a court. Two whistleblowers handed it to the Indian journalist Josy Joseph and the South African journalist Khadija Sharife, who shared it with a coordination body called Barracuda Media. More than two hundred reporters have been reading it since. The files come from the parts of the bank that deal with problem clients: the units where money laundering, sanctions and reputational risk are supposed to be caught.

They were not caught. That is the story, and it is not new.

What the documents contain

The specifics are worse than the pattern.

Jeffrey Epstein’s accounts at Deutsche Bank generated suspicious transactions worth roughly 600 million dollars, according to ZDF. That is more than double what was previously public. An internal investigation found that some of the payments went to bank employees who may have been “witnesses or accomplices to the alleged crimes.” The bank kept him as a client for more than five years and closed his last accounts only in July 2019.

Donald Trump owed the bank about 331 million dollars when, in December 2016, its highest risk committee rejected his application for an additional credit line. The following year the bank accidentally mailed out his tax documents and, according to the reporting, did not tell him.

According to ZDF, the bank in Colombia processed thousands of suspicious transactions without filing the money-laundering reports it was legally required to file in Germany. In Russia, it was preparing to expand its business shortly before the invasion of Ukraine.

And in 2020, per the documents, an employee stumbled onto a laundering network through which hundreds of millions of euros had already moved. The bank’s surveillance systems had registered individual anomalies but never linked them. It took about three years for the pattern to surface. A manager wrote in an internal note: “Our systems simply did not find it.” No fine was imposed on the bank or its staff in that case.

The bank’s Frankfurt tower

The bank’s tower in Frankfurt. Client relationships sat inside it; the reviews of those relationships sat in the same building.

The six words

The sentence that matters most is not about Epstein or Trump or Russia. It is from a 2017 email by Christian Sewing, who became chief executive the following year, about the cost of strict customer due diligence:

“KYC bricht uns das revenue Genick.”

Know Your Customer breaks our revenue neck.

Read it as a complaint and it is banal. Every banker in the world has thought it. Read it as a description of how decisions were actually made and it explains fifteen million documents. The compliance department existed. The rules existed. The risk committee met and sometimes said no, as it did with Trump in December 2016. But the operating assumption inside the building was that customer checks were a cost centre standing between the bank and its income, and that when the two collided, income usually won.

That is not corruption in the criminal sense. It is something more ordinary and harder to prosecute: an organisation optimising the metric it is paid on. Banks are paid on revenue and return on equity. Compensation across the industry is built on those two numbers; refusal does not appear in either.

Corridors of client files

Every folder here is a relationship somebody was supposed to review. The leak came from the units where reviewing them mattered most.

Twenty billion euros, and then the same thing again

Deutsche Bank has paid more than 20 billion euros in fines and settlements since the 2008 crisis. That is not a rounding error on its balance sheet, which stood at 1.4 trillion euros at the end of 2025. It is roughly one and a half percent of total assets, extracted in cash, over fifteen years.

The list, compiled by Handelsblatt and RND, is long enough to be a curriculum: Libor manipulation, mirror trades in Moscow, sanctions violations, mis-selling of mortgage securities, the Danske Bank Estonian pipeline through which 200 billion euros of suspicious money flowed, and the supervisory failures that let Wirecard’s 1.9 billion euros of phantom assets sit on German balance sheets until the company collapsed.

Each case ended the same way. A fine, a deferred prosecution agreement, a monitor, a promise of cultural change. And then, four or five years later, a new case.

The leak describes a bank that had already paid for these failures and was still doing the thing it had paid for.

Why nobody stops it

The intuition is that a bank caught this often must be badly run. The documents suggest the opposite: it was run exactly as its incentives instructed.

Three mechanisms do the work.

The first is the gap between detection and connection. The 2020 laundering network was visible in pieces. Individual systems flagged individual transactions. Nobody joined them, because joining them was nobody’s job and the software was not built to do it. A bank processes millions of payments a day; the pattern that matters is never in one payment. It is in the tenth payment to the same beneficiary through a different correspondent. Machine learning is very good at the first kind of detection and was, until recently, poor at the second.

The second is the cost of saying no. A compliance officer who blocks a wealthy client is measured on the revenue he destroyed. A relationship manager who keeps the client is measured on the revenue he brought in. Both are behaving rationally. The organisation has simply decided, through its compensation structure, which of them it values.

The third is that the fines are affordable. Twenty billion euros sounds enormous. Spread over fifteen years and compared against a balance sheet of 1.4 trillion, it functions as a licensing fee. If the expected profit from a client relationship exceeds the expected fine discounted by the probability of getting caught, the rational decision is to keep the client. Deutsche Bank kept Epstein for five years. Whatever the internal reasoning, it did not end with the client being asked to leave.

The part the leak does not cover: what happens when the bank fails

Everything above is about misconduct. The larger risk is structural, and it sits in the same balance sheet.

Deutsche Bank holds 1.4 trillion euros of assets against a German economy of roughly 4.4 trillion. That is a third of national output inside one institution. Its common equity tier 1 ratio was 14.0 percent in the third quarter of 2026, per the bank’s own results, at the top of its own target range. On paper that is a well-capitalised bank.

The number that matters more is the leverage ratio, which measures capital against total exposure rather than against risk-weighted assets. Deutsche Bank reports 4.4 percent. Risk weights are the bank’s own estimate of how risky its book is; leverage is the blunter, harder number. When the risk models are the same systems that failed to connect a laundering network for three years, the blunt number deserves the attention.

Four point four percent means that a loss of about one twenty-third of the balance sheet would consume the equity. That is not a forecast. It is the size of the cushion.

And then there is the precedent nobody in Frankfurt likes to discuss.

Credit Suisse, March 2023

When UBS took over Credit Suisse, the Swiss regulator wrote down 17 billion dollars of the bank’s Additional Tier 1 bonds to zero. Those bonds were supposed to absorb losses before equity. In this case they absorbed losses while shareholders still received UBS stock.

That violated the absolute priority rule, the principle that says creditors rank above shareholders. European regulators immediately distanced themselves from the Swiss decision, which tells you they understood exactly what it meant: the next time a bank fails, AT1 holders in the eurozone will not assume they rank above equity, because they have watched a case where they did not.

Deutsche Bank issues AT1 instruments. So does every large European bank. The instruments are marketed as loss-absorbing capital that pays a coupon in exchange for a defined place in the queue. In March 2023 the queue was rearranged in a weekend.

This is the risk that never appears in a leak: not that the bank does something illegal, but that the resolution framework, tested once under pressure, behaved in a way nobody had priced.

The strongest case for Deutsche Bank

Deutsche Bank’s defence is not frivolous, and it deserves to be stated at full strength.

The matters in the reporting are old. Some date back two decades. Under Christian Sewing the bank cut its investment banking risk, exited equities trading, restructured its technology, and returned to sustained profitability: 8.4 billion euros of pre-tax profit in the first nine months of 2026, per its Q3 report in the first nine months of 2026, a CET1 ratio at the top of its range, and record net inflows. A bank that was the sick man of Europe in 2018 is now boring, and boring was the goal.

There is a real argument that the leak describes a museum. Compliance in 2026 is not compliance in 2015. Transaction monitoring has been rebuilt across the industry, the EU’s anti-money-laundering authority began operating in Frankfurt, and the supervisory intensity after Wirecard changed what German regulators demand.

And there is a fair point about journalism: fifteen million documents contain everything, including the ordinary. The fact that reporters found Epstein and Trump does not tell us whether such clients are still being taken. It tells us they were.

All of that is true. None of it changes the arithmetic.

What the leak actually proves

A bank can be profitable, well-capitalised, supervised by three authorities, advised by a monitor, and still have spent a decade deciding that the revenue mattered more than the rule.

The documents are not evidence of a few bad actors. Nine million emails are too many for that. They are evidence of an equilibrium: everyone inside the building understood which way the organisation leaned, and behaved accordingly without needing to be told.

That is why the fines did not work. You cannot fine an equilibrium. You can only change what it pays for, and the people who would have to change it are the same people whose bonuses depend on it not changing.

The question underneath

Strip away the names and the sums and there is one question: who decides what happens to your money, and on what grounds can you object?

At Deutsche Bank, over years, the answer was a risk committee that sometimes said no, a compliance function that sometimes warned, and a revenue line that usually prevailed. The customer whose transactions were processed for a laundering network had no way to know. Neither did the depositor whose bank was quietly deciding that a client was worth the risk.

The alternative that Bitcoin proposes is not that banks are evil. It is that the ability to reverse a transaction, to freeze an account, to grant an exception, is a power, and power that cannot be checked will eventually be used.

Bitcoin does not fix fraud. What it removes is the discretion. A transaction on a public ledger cannot be granted an exception, and the warning that should have been escalated is a matter of record rather than of who read it. That does not make the people better. It removes one place where a decision can quietly go the wrong way.

Whether that trade is worth it depends on what you think the alternative is. The fifteen million documents are an argument, not a proof. But they are a very long argument.


Sources: ZDF frontal, Der Spiegel, Manager Magazin, Private Banking Magazin (17 November 2026); Handelsblatt and RND on cumulative fines; Deutsche Bank Q3 2026 results (28 October 2026); Statista on 2025 balance sheet totals; reporting on the Credit Suisse AT1 write-down (March 2023); Danske Bank and Wirecard case documentation.

Write a comment