The books have finally turned black. N26, one of Europe’s most watched fintechs, posted its first annual net profit in 2025: €1.6 million, compared to a €42 million loss the year before. The German challenger bank, with 5.6 million paying customers, saw revenues rise 13 per cent to €501.6 million. New CEO Mike Dargan called the revenue milestone “a milestone”, while CFO Arnd Schwierholz credited “revenue growth, disciplined cost management, and a diversified earnings profile.”

The turnaround rests on three pillars. Surging card transaction volumes and subscription growth drove net fee and commission income up 21 per cent to €184.2 million, accounting for 53 per cent of gross profit. Direct costs – tied to money transfers, subscriptions, and insurance – fell 17 per cent. Headcount at the end of 2025 stood at 1,500, broadly stable, then edged up to about 1,600.

More primary customers, more deposits

Behind the commercial acceleration is a strategy all challenger banks chase: growing the number of “primary” customers who have their salary deposited into the account, becoming the hub of the financial relationship. Customer deposits surpassed €10.5 billion in 2025, a sign the bank is gaining trust and expanding wallet share. The rise in subscription and transaction revenue shows that the business model, built on paid services and fee margins, is strengthening after years of losses.

The leadership change and regulatory shadows

What makes the story less straightforward is the managerial backdrop. Dargan was appointed in December 2024, effectively closing the chapter of co-founders Maximilian Tayenthal and Valentin Stalf, who founded N26 in 2013 and served as co-CEOs. Stalf had already stepped down in 2025 after reported tensions with some investors over the handling of regulatory issues. Tayenthal also moved on. Shortly before, the German financial regulator had levied new sanctions on N26 for compliance shortcomings. These upheavals are no footnote: governance soundness and the ability to keep operational risks in check are now decisive parameters for any institution that wants to grow without regulatory roadblocks.

Where are the bank’s data running?

N26 has never publicly detailed its IT architecture, but the regulatory story shines a light on a topic that goes beyond a single case. Banks, especially those holding a European license, operate within a web of norms – GDPR, Payment Services Directive, anti-money-laundering requirements – that make the physical location of data a non-negotiable variable. The BaFin sanctions are a reminder that compliance isn’t just about corporate processes: it is also embedded in infrastructure choices.

Any organization now evaluating the deployment of large language models in a financial setting – for customer support, transaction analysis, or fraud detection – faces the same fork in the road. The cloud promises scalability and lower operating expenses, yet raises questions about actual data residency and lock-in risks. On-premise or hybrid solutions, in contrast, return granular control and ease audit trails, provided the institution is ready to absorb capital outlays for hardware and specialist skills. The Total Cost of Ownership, in these scenarios, becomes a complex calculus where the benefits of data sovereignty must be weighed against the fixed costs of local infrastructure.

The takeaway for those designing the next stack

For the fintech sector, N26’s first profit signals that even a challenger bank born in the cloud era can achieve stable profitability by keeping operating costs tight. Yet the regulator’s shadow shows that the road to maturity also runs through deep technical choices. For those building AI platforms inside a bank, the balance among control, performance, and spending is no longer a mere architectural detail: it is a competitive factor that determines how quickly new services can be launched without breaking legal constraints.

Even without details of N26’s infrastructure, for the CTOs and decision makers who read AI-RADAR the German bank’s trajectory is a reminder: every growth sprint brings the need to revisit where and how information is processed. When the time comes to put LLMs into production, the cloud versus on-premise question stops being theoretical and becomes the bedrock of a solid deployment strategy.