Capitec and Le Roux: Three Big Bets

Capitec is priced for a world where nothing goes wrong

The co-founder's R3.7 billion insurance policy

On 3 July 2026, Michiel le Roux, the man who built Capitec from a startup into a R550 billion ($30B) giant, borrowed R3.71 billion ($202M) against his own shares. He used a structure called a collar.

Le Roux has done this before. In 2019 he said he was "growing older and needed to acquire a little bit of balance" in his portfolio. Reasonable. The collar floor sits 37% below the price at which the deal was structured, and Capitec hit its all-time high of R4,833.32 in February 2026. This does not read as a founder betting the company falls next month. It reads as someone in their seventies protecting against catastrophic downside while accessing cash without selling shares he has held since the company was worth close to nothing.

One caveat worth noting: when Snowflake CEO Frank Slootman sold $69.2 million in shares in early February 2024, he announced his resignation weeks later and Snowflake's stock dropped roughly 29%. But Slootman was exiting a role. Le Roux is financing a balance sheet. Borrowing and selling are not the same thing.

The three bets — and where they're cracking

Capitec trades at 7.8 times its book value. Book value is the bank's net worth on paper: assets minus debts. Standard Bank trades at 1.48x. FirstRand at 1.95x. Capitec trades at roughly five times the premium of its nearest rival. That premium rests on three bets.

Bet 1: mass-market customers will get richer and stay

Capitec's growth story depends on pulling lower-income South Africans up the wealth ladder and keeping them in its ecosystem as their spending power grows. Two rivals are now after the same customers.

Running stores and running a bank are two entirely different problems. Capitec posted R16.8 billion ($915M) in headline earnings for FY2026. Pepkor is entering with a tech platform built on an acquisition and a CEO who left Capitec, not the one running it. Building credit-scoring infrastructure and managing default risk on a low-income lending book takes years. African Bank collapsed in 2014 after building exactly that kind of book without first building daily transactional relationships with its customers — no daily data, no early warning, no leverage when consumers came under stress. Pepkor faces the same trap if it pushes lending before building the data layer. The threat is real, but at meaningful scale it's probably a 2028-2030 problem.

Bet 2: ATMs and branches lock customers in

While other banks cut ATMs, Capitec added 3,787. Competitors cut a combined 8,249. That calculation made sense when cash was king.

PayShap — South Africa's instant payment system that moves money between any two bank accounts in seconds for almost nothing — could change this equation. When a customer can pay anyone from any bank instantly, the convenience advantage of being inside the Capitec ATM network shrinks. But South Africa's informal economy still transacts heavily in cash, and a meaningful share of Capitec's customer base is highly security-sensitive about digital-only services. The ATM network becomes less valuable incrementally, not overnight — which at a 30x earnings multiple still matters.

Bet 3: owning the data loop

This is the most interesting bet and probably the most fragile.

Capitec runs an MVNO called Capitec Connect. An MVNO (Mobile Virtual Network Operator) is a phone company that sells data and calls without owning the towers — think of a supermarket selling a coffee brand it does not grow or roast. Capitec Connect generated R442 million ($24M) in net income in FY2026. Every bundle a customer buys generates behavioural signals that Capitec uses to assess creditworthiness without requiring a formal payslip.

The Cell C dependency is the one risk on this list that Capitec's balance sheet cannot simply buy its way out of overnight. Capitec has both the financial resources and the strategic incentive to migrate or acquire infrastructure if the relationship deteriorates — but the free-call lock-in means its own customers would be the biggest obstacle to leaving. Across Africa, this data race is the defining battleground of the next decade. Safaricom in Kenya uses M-Pesa transaction data to build credit profiles for millions of customers who have never held a formal loan. Whoever owns the daily data loop most completely wins the credit-scoring war.

The numbers

The 31% return on equity at this scale is the number that justifies the premium. Banks in developed markets consider 15% exceptional. The credit loss ratio — the share of loans that turn bad — rose from 7.5% to 8.1%, real pressure from a consumer base squeezed by rising costs and global trade disruption. Not a crisis level, and Capitec has already tightened its lending criteria in response.