Why banks launch MVNOs, and what it takes

Banks launch MVNOs because a mobile line adds a second revenue stream and a daily reason to open the app. Capitec Connect, the mobile arm of a South African retail bank, reported net income of R442 million for the year to February 2026, up 129 per cent, on 1.5 million three-month-active subscribers, having launched in 2022; Nubank’s NuCel passed one million customers in June 2026. What it takes is a host agreement and a BSS the bank controls rather than rents.

Capitec Connect, the mobile arm of a South African retail bank, reported net income of R442 million for the year to February 2026, up 129 per cent, on 1.5 million three-month-active subscribers. It launched in 2022. In Brazil, Nubank’s NuCel passed one million customers in June 2026, having become the country’s largest credentialed MVNO by Anatel accesses in April. In 2025 Revolut, Klarna and N26 all announced or launched mobile plans. A mobile line inside a bank has stopped being an experiment. The question for a bank strategy team is no longer whether the model works, but which version of it fits and what has to be in place before launch.

Why now

South Africa has the most complete evidence because two banks publish numbers. Capitec’s figures come from its annual results, reported by MyBroadband in April 2026, and its active base grew 67 per cent in the year. FNB Connect, according to TechAfrica News and ITWeb in November 2025, has passed one million active subscribers, with eSIM now 10 per cent of monthly SIM sales. TechCentral’s June 2025 list of the country’s MVNOs shows four South African banks and insurers running mobile brands.

Latin America produced the fastest recent example. Nubank’s NuCel reached one million customers by June 2026, per the Nubank newsroom and TELETIME, and was the largest credentialed MVNO in Brazil by Anatel accesses in April 2026. The distribution was the bank’s app: no shops, no dealer network, none of the acquisition spend a greenfield MVNO would recognise.

The European and North American fintechs followed in a cluster. Revolut announced mobile plans for the UK and Germany in April 2025. TM Forum Inform reported in September 2025 that Klarna had launched in the US on AT&T in June 2025, that N26 had launched an eSIM plan in Germany on Vodafone in May 2025, and that Monzo was reported to be preparing a similar move. None of these has published subscriber figures in the sources we use, so they are evidence of intent rather than of results.

The idea itself is not new. Equity Bank’s Equitel in Kenya was licensed in 2014 on Airtel’s network and, according to Communications Authority data reported by Citizen Digital in June 2026, holds 1.51 million subscriptions. What has changed since 2014 is that eSIM removes the shop, the bank app has become the channel customers open most often, and analysts treat banks as a permanent category of operator. Omdia’s Matthew Reed said in February 2025 that “banks, supermarkets, and broadband and TV service providers continue to be key players” in the MVNO market, and Omdia expects Middle East and Africa MVNO subscriptions to grow 17.6 per cent a year to 2029, driven by South Africa and Nigeria.

Bank and fintech MVNOs: what is public

Two patterns stand out. The banks with published results serve a mass, mobile-first customer base in markets where prepaid is the norm. The fintechs entered with eSIM and app-only distribution, the cheapest way to test the proposition. Neither group launched to compete with its host on price.

Three business cases

The first is retention and engagement. A customer who pays for mobile from a bank account, and earns data for a card purchase or a salary deposit, has one more reason to keep that account as the primary one. Balance-based rewards turn deposit stickiness into something the bank can price: a gigabyte for holding a minimum balance costs the wholesale price of a gigabyte and is visible in the app every day. This case is measured in the banking P&L, in primary-account share, deposit retention and app activity, and it is the reason most banks give internally for starting.

The second is cross-sell and data. The mobile line generates events the bank did not have: SIM activations and swaps, roaming country, top-up behaviour, plan changes, usage patterns. Combined with transaction data these become segmentation attributes for offers and, within a lawful basis and with consent, fraud and risk signals. A SIM swap before a large transfer is a stop signal; a card transaction during an active call on the same number is a fraud flag. These are added features, never a sole decision, and the bank’s risk and compliance functions decide how they are used.

The third is a new revenue line. Capitec’s R442 million shows a bank MVNO can become a material telecom business, and SMS, USSD and push on the bank’s own routes reduce a cost the bank already carries. Our view is that this should be the proof the board sees in year three, not the reason for starting. A plain airtime margin is thin, and a bank that launches to sell cheap data is competing with its host. The banks with strong published telecom results built the first two cases first. The general economics are in the MVNO business model.

What it takes

Five decisions determine whether a bank MVNO delivers those cases or ends up as a plan on a shelf with the bank’s logo on it.

Choosing the operating model

A reseller or Light arrangement puts the bank’s brand on the host’s or an enabler’s plans. Billing, pricing logic and subscriber data stay in the host’s systems, so the bank cannot build a reward rule the host does not offer, cannot see event-level usage and cannot reuse its own onboarding. A Medium MVNO runs its own BSS (catalogue, charging, billing, CRM, self-care) on the host’s network; the data and the commercial logic are the bank’s. Full adds the network layer later, for autonomy and roaming control. For a bank, Medium is the usual answer, with Full as a planned step rather than a rebuild. The ownership split by model is on the operating models page; the trade-offs are discussed in Full MVNO, Light MVNO or MVNE.

KYC reuse and the regulatory perimeter

Every market requires SIM registration, and the rules differ. South Africa’s RICA requires full name, ID number, address and MSISDN before activation, and treats an eSIM as a SIM. Nigeria links every SIM to a National Identification Number. Kenya’s SIM Registration Regulations 2025, in force since 30 May 2025, require verification against government databases. A bank already holds this data, verified to a higher standard than a telecom dealer applies. Whether it may reuse that verification for SIM registration, and whether the mobile business sits inside the bank or in a subsidiary, are questions for the telecom regulator and the banking supervisor together. Settle them in discovery, because the answer shapes the onboarding flow and the legal entity that signs the wholesale contract.

Core-banking and card integration

Reward rules only work if the BSS sees banking events as they happen. Card authorisations, salary credits, loan repayments, balance thresholds and account-tier changes have to reach the telecom catalogue in near real time, and telecom events have to flow back into the bank’s CRM, data platform and risk engine. That means an event integration with the core banking or card platform, a payment integration so the account can pay the bill and fund top-ups, and the bank app as the single front end for eSIM activation, plan changes and balances. None of this is exotic; the effort is in the number of integrations and the bank’s own change calendar.

Billing logic the host cannot provide

Balance-based rewards, card-spend-to-data conversion, plans reserved for salary customers, mobile included in a premium account tier, rollover of unused allowances, family lines under one account, and sponsored billing where the bank or an employer covers part of a customer’s usage all need a catalogue and online charging that accept banking events as inputs. Caps, eligibility rules and expiry matter as much as the reward itself, because an uncapped reward is a liability. Business-banking clients add cost centres, per-employee limits and pooled allowances. A bank that owns the catalogue configures and tests these in weeks; a bank on a reseller plan waits for the host’s roadmap.

Team and managed operations

A bank has no telecom operations team and does not need one to launch. Year one can run as a managed service, with the vendor or a partner operating the BSS, the host interfaces, number porting and regulatory reporting, while the bank’s teams work on offers, channels and the customer. What cannot be outsourced is ownership of the proposition, pricing, compliance and the customer relationship. On Avante’s planning assumptions a Medium launch takes four to six months once the host agreement and legal framework are in place, and Full eight to ten: Discovery two to three weeks, host and legal on the critical path, BSS platform four to eight weeks, integrations six to twelve, launch two to four. The stage-by-stage checklist is in How to launch an MVNO.

Mistakes to avoid

The most common is launching as a reseller because it is quick, then finding that the data, the pricing logic and the onboarding all live with the host. Moving to an own BSS later means a subscriber migration that could have been avoided. If the business case depends on rewards or data, start with the model that gives you both.

The second is treating the MVNO as an IT project. A mobile brand needs a product owner with a P&L, a pricing calendar and a care plan, and it needs the bank’s marketing to sell it in the app. Banks that bury it in the technology division get a working service nobody promotes.

Third, uncapped rewards. A gigabyte per card transaction with no monthly ceiling is generous until a customer splits a purchase into fifty payments. Every rule needs a cap, an eligibility test and an expiry, agreed with finance before launch.

Fourth, underestimating the critical path. The host agreement, the regulator’s view on KYC reuse and the legal structure of the mobile business take longer than the platform, and they cannot run in parallel with a launch date already announced.

Fifth, using telecom data in credit decisions without a lawful basis, consent and an agreed data-processing model. As added features the signals are valuable. As sole decision inputs they are a regulatory problem waiting to happen.

Finally, measuring one side only. A bank that tracks only the telecom P&L will call the line a failure in year one; one that tracks only engagement will never know whether the line covers its costs. Both numbers belong on the same dashboard.

Where Avante fits

Avante supplies the BSS a bank runs in a Medium or Full model: product catalogue, online charging, convergent billing, CRM and agent workspace, self-care, partner management and the integrations to the host, the bank app, core banking and payment gateways. Five fintech and bank MVNOs run on the platform. A leading banking MVNO in the CIS was delivered in six months to the bank’s security and integration requirements, and Medium to Full runs on the same BSS with no second subscriber migration. Year one can run as a managed service with 24x7 support, so no in-house BSS team is needed at launch. Avante does not run a network or resell airtime; the host agreement remains the bank’s. The banking-specific detail, including the scenario library and the data available by model, is on the banking MVNO page.

Frequently asked questions