advertisement
Selling To Strangers
You can buy a SIM card from a bank in South Africa. You will soon be able to open a bank account at a Shoprite till. A Nigerian family holding company that made its name in banking, power and oil now sells outsourced IT support and would like to compete with India for it. And a Kenyan insurer that has been writing policies since before independence runs an AI underwriting engine it will licence to anybody who asks.
None of this is a secret. All of it has been announced, covered, and filed away one press release at a time, which is exactly why nobody has stepped back to look at the whole picture.
Step back and it looks like this: across the continent, old companies are building new technology businesses that sit beside the original business rather than underneath it. Not IT departments. Businesses, with their own profit lines, their own customers and, in a few cases, their own investors.
advertisement
Whether that is a good idea is a separate question, and the evidence on it is more interesting than the announcements.
The Bank That Noticed What It Had Become
Start in Nairobi in 2018, because that is where this begins.
On 22 August that year, Equity Group took Finserve, the outfit behind its Equitel mobile network, and made it a standalone commercial subsidiary with permission to sell to anyone, including companies that banked elsewhere. “We have taken a deliberate strategy to make Finserve an independent commercial subsidiary,” James Mwangi said at the time. He also said something that explains the whole decision in nineteen words. “Thanks to fintech innovations by Finserve, 97 per cent of all Equity transactions are now outside the branch.” Ninety-seven per cent. At that point you are not a bank with a good app. You are a technology company that happens to hold a banking licence, and the only remaining question is whether your accounts say so.
advertisement
Finserve went on to build the Jenga payment gateway and 64 application interfaces covering finance, regulation and insurance, and sold them across East Africa.
Eight years later, Equity is doing it again, and bigger. Newsroom Kenya and The Africa Report have both reported that the group is preparing to spin its technology operations into an independent company by late 2026 — this time including Azenia, the Nairobi software outfit that builds Equity’s digital products, alongside Finserve Africa and its 1.5 million-plus Equitel subscribers. The point is to let the technology arm raise money and sign partnerships without being tethered to a bank’s balance sheet.
A caveat: that reporting rests on people who were not named, Equity did not respond to requests for comment, and nothing has been announced. No chief executive has been named for a company that does not formally exist yet.
advertisement
What is on the record is where Mwangi’s head is. “The future of financial services will not be built on brick-and-mortar,” he told the Inclusive FinTech Forum in Kigali this year. The group wants 100 million customers by 2030.
So: carve out the payments business. Run it for eight years. Then separate the entire technology stack and open it to outside money. Half the continent is at step one of a journey Equity is finishing.
Lagos Works It Out
Nigeria’s banks reached the same conclusion later, and then moved faster than anyone.
The thing that made it possible was structural rather than visionary. Holding company structures let lenders run non-banking businesses as separate, distinctly reported subsidiaries, and the Central Bank of Nigeria’s expanding payments rulebook gave those subsidiaries somewhere legitimate to stand.
GTCO started Habari in 2018, which went nowhere in particular, and relaunched it as HabariPay in 2022 around a merchant payments product called Squad. Squad worked. In 2025 HabariPay turned over ₦12.9 billion, up 122 per cent, made about ₦9.7 billion, and pushed ₦8.1 trillion through its rails. In the first half of 2026 it nearly doubled profit again, to ₦7.8 billion. Across GTCO’s whole non-banking stable, nine-month profit climbed 70 per cent to ₦23.9 billion.
Access Holdings built Hydrogen — switching, gateways, point-of-sale terminals, the plumbing — and moved ₦41.1 trillion through it in the first half of 2025 alone. Profit has been bumpier: roughly ₦1.65 billion after tax for 2025, down from ₦1.89 billion. Stanbic IBTC’s Zest Payments is younger and smaller, posting ₦543 million in the third quarter of 2025 against a ₦1.89 billion loss a year earlier, which is at least the right direction.
Wema’s ALAT has been running since 2017 and does not report its own profit. Zenith’s ZenPay publishes almost nothing at all.
Five banks, five fintech subsidiaries, and three of them will not show you the numbers. Draw your own conclusions about why.
The Insurer That Became A Software Company
It is tempting to read all this as a banking story. It is not.
Jubilee has been selling insurance in East Africa for over 50 years. Its technology arm, J-Hub, now describes itself as an insurtech engine, and its product list reads like something out of a software catalogue: an AI risk and underwriting engine that scores on alternative data, white-label embedded insurance interfaces for fintechs and online retailers, USSD journeys for people without smartphones, micro-insurance and micro-savings, automated claims intake with AI adjudication, and a marketplace for partners to build on. 14 African markets, more than 5,000 agents and brokers.
There is commercial substance underneath the brochure language. In July, Jubilee and FSD Africa started building affordable cover for underserved Kenyans, later picking seven insurtechs to work with — Dukatech, Aura Insure, Inclusivity Solutions, DPE, Fiinovate, Agrails and Ibisa Network — across embedded insurance, digital underwriting, small business protection, climate intelligence and satellite data. A separate tie-up with bolttech, launched on 27 July, pushes Jubilee’s health products through embedded distribution.
Juan Cazcarra takes over as group CEO on 1 December, having run operations, finance and risk as deputy, and having been in the room for the FSD Africa work.
An insurer selling underwriting software to other insurers is the same move as a bank selling payment rails to other merchants. Which is to say: the licensed product is no longer the only thing on the price list.
Renting What You Cannot Build
Then there are the companies that simply walked into somebody else’s industry.
South African banks have spent a decade becoming mobile operators. FNB Connect, Capitec Connect, Standard Bank Connect, Nedbank — all live, with Absa circling. The country’s virtual operator market held 4.4 million active SIMs last year and is forecast to hit 14.4 million by 2030, most of it, bank-owned.
These are no longer vanity projects. Capitec Connect cleared R442 million in net income in the year to February 2026, more than double the R193 million before it, off 1.5 million subscribers burning through 768 million voice minutes. FNB has sold over R600 million in devices and watched data usage climb 98 per cent in a year — more than 40 petabytes between July 2025 and May 2026.
The banks all say the same thing about why, and it is not really about airtime. “Multi-product customers show higher retention, lower propensity to switch and greater lifetime value,” says Sashin Sookroo, who runs FNB Connect. Kartik Mistry at Standard Bank Connect puts it more bluntly. “Digital banking depends on reliable connectivity.”
Kenya, for the record, got there a decade earlier. Equitel was the country’s first virtual operator, launched by a bank, while South Africa’s lenders were still thinking about it.
Traffic now runs the other way too. On 10 November, Shoprite announced it is partnering with Investec to expand into transactional banking. Investec supplying the licence, the infrastructure and the regulatory know-how, Shoprite keeping control of what actually gets sold in its stores. The grocer already does basic banking for more than four million customers over its Money Market counters, runs over 3,700 stores and turned over R270.8 billion in the year to June.
“This partnership allows us to scale our established financial services capabilities,” said Investec CEO Fani Titi. Shoprite’s Pieter Engelbrecht talked about giving customers “greater choice and control over their money.”Notice what Shoprite did not do. It did not apply for a banking licence. It rented one. For a retailer with four million customers already queuing at a Money Market counter, that is the cheapest possible way to find out whether the idea works.
The Expensive End
Cassava Technologies, out of Strive Masiyiwa’s Econet group, is doing the version of this that almost nobody else can afford: building the compute itself. Nvidia systems in South Africa, a facility running in Cape Town, 20 megawatts planned for Johannesburg, and Dan Kwach brought back in October to drive the data centre push.
A telecoms group turning itself into an AI infrastructure provider for a market that did not exist when it started. Few imitators, for the obvious reason.
Selling To Strangers On Another Continent
Heirs Technologies is the odd one out, and the most interesting for it.
Part of Tony Elumelu’s Heirs Holdings, it was incorporated back in 2004 but only started trading in January 2024. What it sells is basic: outsourced IT support for companies whose business is not technology. Infrastructure management, monitoring, patching, cybersecurity, cloud, networks, analytics, backup and disaster recovery. The ambition is not basic at all. The company is going after a global outsourcing market it sizes at $250 billion, explicitly copying how India and Eastern Europe built their services industries, and betting on Nigeria’s supply of young developers.
“We want to make Nigeria a serious outsourcing hub,” says Executive Director Anant Rao.
What is refreshing is how clear the board has been about what this is not. Chairman Fumbi Chima has said the company will not compete on products and will put its money into technical support instead. CEO Obong Idiong talks about giving clients “a seamless and secured IT environment,” which is about as far from a pitch deck as it gets.
And it is a genuinely different proposition from everything else here. A carve-out sells something the parent already built. Heirs is competing with Tek Experts and with offshore providers on three continents, on price and delivery, for clients who have never heard of its parent and do not care. Its answer is people: cohorts of 75-100 in training, campuses planned across the country, an academy on university grounds, riding a federal push to train three million technical workers in four years.
Ask again in three years.
Not everyone is building. KCB got competition approval early this year for a Kshs 2 billion purchase of 75 per cent of Riverbank Solutions, a payments and agency banking firm, and a Tanzanian filing later revealed a 22.23 per cent stake in Pesapal. Buying is a perfectly respectable answer. It swaps a three-year build for an integration headache, which anyone who has done both will tell you is a real trade rather than an easy one.
The One That Went Wrong
Now the part the conference panels skip.
MTN Nigeria wrote off ₦62.56 billion against its fintech subsidiaries in its 2025 audited accounts, describing them as loss-making. It then sold 60 per cent of MoMo Payment Service Bank and Y’ello Digital Financial Services to MTN Group Fintech at a ₦95.5 billion valuation, implying about ₦152 billion in fresh capital, with KPMG signing off on fairness. Part of the reasoning was regulatory housekeeping: the restructure leaves MTN Nigeria answering to the Nigerian Communications Commission alone, rather than straddling telecoms and central bank supervision.
Read that again. The company with the biggest subscriber base on the continent, in its biggest market, impaired its own fintech venture and shifted it elsewhere in the group. It is a reminder that this is harder than the slide suggests.
The bank-owned mobile networks have a trap of their own, and it is a beauty. They run on identities they do not own. The IMSI inside every SIM belongs to the host network, and South Africa’s regulator only hands IMSI ranges to companies holding spectrum, which no virtual operator does. Want to switch host? Reissue every SIM, port every number, one at a time. Standard Bank moved roughly 300,000 subscribers from Cell C to MTN in June 2024, so it can be done. At 1.5 million subscribers, telecoms consultant Pambos Soteriades reckons it becomes commercially impossible, with passive churn alone costing six figures of customers. His summary of where that leaves you at contract renewal runs to four words. “The terms belong to the host.”
What Nobody Says Out Loud
Underneath a fair number of these decisions sits a valuation hope. Banks get valued on book value and return on equity. Technology companies get valued on revenue or transactions. Separate the fast-growing bit, the thinking goes, and the market re-rates the parent.
The record does not support it.
Emirates NBD pulled around $1.4 billion out of selling down Network International and booked a $1.2 billion gain, and got no lasting re-rating, with the eventual take-private valuation roughly where the IPO had been five years before. Santander kept 89.9 per cent of Getnet, the float was thin, the shares dropped about 34 per cent, and it went private. SBI Cards trades on technology multiples, but the stake is too small to move SBI, which the market still values as a bank.
Frontier Fintech, which assembled that record, draws the conclusion that ought to be printed on the inside cover of every strategy document on this subject. “Changing where revenue is reported creates visibility without improving the underlying economics,” adding that “Separate valuation follows economic independence.”
So, What Separates Them
Line up the winners against the strugglers and one thing does most of the explaining.
Finserve sells to companies that are not Equity. Squad serves merchants who bank elsewhere. Hydrogen supplies the wider market. J-Hub licences underwriting to firms that are not Jubilee. Heirs is chasing clients who could not pick Tony Elumelu out of a line-up. They have customers of their own.
A subsidiary that exists to serve its parent is a cost centre with better stationery. It might be excellent. It will still be valued as part of the parent, because economically that is all it is. The ones posting real profits are the ones that went out and sold to strangers.
Which sounds obvious, and is brutally hard. Selling outside means service credits owed to people who can walk, support hours that do not respect your calendar, a roadmap driven by customers who do not report to your CEO, and pricing that has to survive being compared with somebody else’s. Internal customers put up with things. External ones switch.
What It Means If You Are The One Running IT
Here is where this lands for anyone sitting in a technology seat inside an African company.
Sooner or later someone senior is going to look at a platform your team built and ask whether it could be sold. It is a flattering question and a dangerous one, because the thing that works beautifully inside your organisation is usually held together by people who know where the gaps are and quietly work around them.
Productising it means writing down what nobody wrote down. It means surviving integration with systems designed by people who never met you. It means a support function answerable to a contract rather than a colleague down the corridor.
The companies in this story that are making money did that work. The ones taking impairments, on the published evidence, had built something that was not yet a business. Which makes the board question rather simpler than it usually gets framed. Not whether to have a technology subsidiary.
Who, apart from us, is going to pay for it?