Why modernisation starts underneath
TechCentral reported this week that the JSE has sharply increased what it spends on software. The exchange put R75.5-million into intangible assets in the six months to June, against R24.4-million in the same period last year. Total capital expenditure rose more than fourfold to R110-million, with full-year guidance of R190- to R230-million.
The destination is ambitious: a pan-African digital marketplace by 2031, part of a strategy the exchange calls Forge 2031.
The detail worth sitting with is not the number. It is what a large part of the money is doing.
The exchange is moving its Broker Dealer Administration system off a legacy mainframe.
The unglamorous half
Not a new customer app. Not an AI product. Taking a core system that already works and moving it onto infrastructure that other things can actually connect to.
That is the half of every digital strategy nobody puts in the press release, and it is the half that decides whether the rest of it works.
It is worth being clear about what “legacy” means here, because it is often used as an insult. A legacy system is usually not a bad system. It is frequently the most reliable thing a business owns, precisely because it has been running for twenty years and every edge case has already been found. The problem is rarely that it works badly. The problem is that it works alone.
Why the order matters
A system that cannot be reached is a system nothing else can build on.
You can buy the best analytics tool available, but if it cannot see your order data, it analyses nothing. You can add an AI assistant to your website, but if it cannot look up a real customer’s real order, it can only guess convincingly. You can plan a marketplace, but if settlement lives inside something that only speaks to itself, the marketplace has no floor to stand on.
This is why serious modernisation programmes so often start in the least interesting place. Not because the organisation lacks ambition, but because it has learned the order.
The failure mode is familiar to anyone who has watched a few of these. A business buys the visible thing first, because the visible thing is what got approved. It works in the demonstration, where the data was prepared by hand. Then it meets the real estate of systems underneath, discovers half of what it needs is unreachable, and quietly becomes a project nobody mentions.
The same shape, at a different size
Very few businesses are moving a mainframe. Almost every business has the same problem in miniature.
An accounting package here. A stock system there. An online store somewhere else. And a person in the middle, keying the same order into all three.
That person is doing integration work by hand. It is slow, and it is where the errors come from, because an order typed three times can be wrong three times, and the wrong version is usually found by the customer. It is also invisible on any budget line, which is why it can continue for years without anyone deciding to allow it.
The question that surfaces it is simple enough to ask in a meeting: when a sale is made, how many systems need to know, and who tells them?
If the answer is a person, that is the joining problem. It is the same one the JSE is spending nine figures on. The difference is that at three systems rather than three hundred, it is a matter of weeks rather than years, and it costs a fraction of what people expect.
What this does not mean
It does not mean rip everything out and start again. That is the advice businesses are most often given and it is usually wrong: expensive, disruptive, and unnecessary. Your accounting package is probably fine. Your stock system is probably fine. What is missing is the wiring between them.
Nor does it mean AI is overrated. It means AI is downstream. An agent is only as good as what it can reach, and the reaching is a solved problem that most smaller businesses have simply never been offered.
The JSE is not spending that money because mainframe migrations are exciting. It is spending it because the thing it wants to build in 2031 cannot exist without it.
The order is the lesson. The scale is just detail.
Source: JSE triples software spend in push for digital marketplace, TechCentral, 5 August 2026.