Until the latter half of 2024, core banking applications (CBAs) attracted little public attention outside Nigeria’s banking industry. That changed when several major commercial banks switched or upgraded their CBAs, disrupting banking services for millions of customers.
But five years before core banking was on Nigerians’ lips, the founding team at Kuda, a Nigerian neobank which boasts over seven million customers, knew that building its CBA could be the difference between success and failure.
“It was, of course, one of the hardest and probably the most daring decisions I’ve ever taken in my professional life,” Musty Mustapha, Kuda’s co-founder and chief technology officer, said about the decision to build an in-house CBA. “I knew I had to stick my neck on the line with my founding partner, the investors, the board, and everybody else.”
Banking’s economic logic has not changed much since its early days: institutions collect deposits from customers, guarantee access to the deposits, and finance the economy by extending credit. What has changed is the mechanics and machinery powering it.
Today, the core banking application, an always-on ledger, is the software that records and powers every deposit, loan, and transaction in a bank. As the engine of modern banking, it ensures that accounts are updated in real time, that bank operations are reliable and scalable, and that the bank can launch and manage products.
How it happened
In 2019, Kuda was still small enough that its key decision-makers could fit on a small table. There were no headline venture capital firms like Target Global yet. The neobank was live and growing fast, but running on a third-party core banking provider.
This is the model most financial institutions in Nigeria rely on because building core banking infrastructure takes time, effort, and capital that could be better spent elsewhere. Apart from Kuda, only a handful of Nigerian financial institutions, like Moniepoint and Sterling Bank, have developed their own core banking applications in-house.
However, inside Kuda, the third-party CBA was starting to crack under the weight of the neobank’s ambition. More customers and product releases drove demand and a growing intolerance for instability, placing pressure on the CBA.
Mustapha knew that this tension could jeopardise his young startup’s future. A series of conversations with his engineers and Kuda’s CEO, Babs Ogundeyi, led to what eventually became NERV, Kuda’s in-house core banking system, named after the human nervous system because it sits underneath everything and quietly controls how Kuda operates.
Mustapha insisted on one point above all else during our hour-long conversation in October 2025: when Kuda started building NERV, Nigeria’s banking or fintech market had not built an in-house core banking system like Kuda was attempting to.
“At the time when we built NERV, there was no fintech or banking-related company that had built its own core banking system,” he said. “NERV is the pioneer in-house-built core banking solution by a bank in Nigeria.”
However, Moniepoint began its in-house core banking development in 2020, and several Nigerian companies like Appzone’s Qore already sold core banking products to Nigerian microfinance banks.
NERV began as most banks think about CBAs: shopping. Musty approached multiple local and foreign providers, seeking a CBA robust enough for what Kuda wanted to become and cheap enough for what Kuda could afford in 2019. The startup eventually launched in August 2019 on a local core banking provider.
Then reality set in as Kuda grew to a million customers quickly, and the system felt brittle under this weight. Mustapha described three issues that kept compounding: it could not scale with customer growth, could not be adapted to Kuda’s needs, and above all, could not be relied upon to stay up.
“If you are offering a financial service, the minimum that you can do is to ensure that your systems are stable for your customers,” he said. “If you cannot achieve this, you can as well just pack your load and go home.”
So, Kuda did what most institutions avoid: it kept looking for alternatives before deciding there weren’t any that met its needs within the constraints it had. At that point, Mustapha said that the question became existential rather than technical: what kind of company are we building?
His answer was simple. “Kuda is an engineering-first company,” he said. “That is the kind of company we’re trying to build: using engineering resources to solve financial services problems in Nigeria.”
Mustapha added that Kuda never needed to run a long persuasion campaign, even though building a core banking system could cost tens of millions of dollars and is slow.
In September 2019, the company had raised only $1.6 million in a pre-seed round from angel investors and family offices. This lack of institutional capital meant fewer layers of approval, less red tape, and fewer people who could veto a high-conviction decision.
But even then, Kuda avoided betting the business on a rebuild while it was still learning to exist. Instead, NERV ran in parallel to the existing CBA. Kuda continued operating on the third-party core while engineers built NERV in the background. That way, if it failed, the bank would not die. It would keep running, find a new provider, or endure the one it had.
“It wasn’t like the company stopped operating,” Mustapha said. “We were still on the third-party provider system, still serving customers, still growing, but we’d decided and were treating it as a side project.”
Kuda had another advantage; its engineering talent was willing to take on a long, invisible build while still maintaining production systems.
“There’s no way I would have even attempted to go ahead with it if that quality was not there,” Mustapha said. “I’m speaking about an actually world-class team.”
The mechanics of NERV
NERV’s architectural choices were less exotic than the outcome itself. Kayode Ilesanmi, one of the engineers who built it, said the team started with a clear conclusion: if Kuda wanted scalability and flexibility, it couldn’t build a monolith.
“We just had to say, ‘You know what, for us to be able to achieve that kind of scalability and flexibility we want, we have to go with a microservices architecture,’” he said.
What stands out with this approach is the ambition behind the throughput targets. Kayode described a guiding principle that became a kind of internal dare: 1,000 transactions per second, meaning 1,000 customers hitting the system simultaneously should feel like one.
They were nowhere close in the beginning. “We started the first time, and we couldn’t even get up to 10,” Ilesanmi said.
Then they did what infrastructure teams should do. They traced the bottleneck, inspected logs and metrics, and questioned whether the issue lay in the database, the cache, the application logic, or the compute.
He identified the cache connection handling as a weak point, studied how leading implementations managed it, rewrote their approach, and saw the load test jump to over 500, then eventually to the 1,000 goal.
By the time they were near go-live, the team says they were hitting 5,000 transactions per second in tests.
Migration day: when production refused to cooperate
If you want to understand the cultural differences between big-bank and startup migrations, start with risk. Musty referenced Nigerian banks’ public migrations, noting that large banks spent millions of dollars hiring global vendors and still faced messy outages and backlash.
If you apply that pressure to a startup barely a year old, with a small team, and no option to hide behind institutional distance, you get a failed migration day.
“Yes, absolutely,” Mustapha told me when I asked if it was painful. “It wasn’t all rosy.”
What Kuda did differently, he believes, was to treat customers like participants rather than victims. In the weeks before they went live, Kuda began telling users they were switching cores, when it would happen, and that the outcome would be fewer failures and faster service.
When migration day arrived, communication intensified as emails, real-time updates, and a social media thread turned into something rare in Nigerian banking: customers defending the bank while it was down.
“If anyone was even saying, ‘Oh, what’s happening, Kuda?’ it was customers that would be responding to them,” Mustapha said. However, the plan still failed in the beginning.
Kuda had built the system in a test environment, then in a staging environment designed as a near-production rehearsal space. But when they moved into production, the migration failed.
So they reverted to staging and pushing production data into it, because staging behaved like the environment they had tested and load-tested repeatedly.
“The production environment was slow, which was really weird,” Ilesanmi said. “It was an environment that had higher specifications… and the expectation was that requests… should be fast. But they were slow.”
When they switched back, the latency normalised, and the bank launched in the environment that worked.
What does owning core banking change?
When fintech leaders talk about in-house core systems, they often lead with margins. However, Musty said the move wasn’t a cost-cutting one.
“We didn’t take this decision… because I felt, ‘Oh, we’re going to reduce cost,’ or ‘We’re going to make lots of money out of it.’ No,” he said. “The goal… was to be in a position where we can serve our customers rightly.”
Everything else, he argued, becomes a downstream benefit: better unit economics, product velocity, and fewer vendor constraints.
Owning the core also changes a bank’s risk profile. With a third-party core, your operational risk is no longer just your risk; you also inherit your provider’s downtime, constraints, roadmap, and security rating.
“By having our own system, we’ve kind of removed that third-party risk,” Musty said.
From an engineering and product standpoint, the team described immediate wins that users could feel. The app became faster, and transfers were completed more quickly, especially during the COVID-19 pandemic when digital transaction volume surged.
With NERV, Kuda scaled internal overdrafts into customer-facing credit, built business banking products, launched term loans, and expanded into multi-currency wallets, capabilities the team says would have been slower, more expensive, or structurally harder on an external core.
From an operations standpoint, the changes were even more concrete. Before NERV, running end-of-day (EOD) routines (ledger balancing and banking close processes) could cause the system to freeze.
With NERV, they automated EOD so it runs without locking the bank’s ability to operate during the day.
“It became very flexible, where you don’t even need somebody to go there to run EOD, and everything runs by itself,” Ilesanmi said.
