South Africa’s 2025 Tech Leaders Shaping AI Commerce

A funny thing happened in South African tech this year: the biggest “AI story” wasn’t a flashy chatbot launch. It was a set of hard, structural moves—fibre consolidation, mobile turnaround plans, platform scale plays, and policy tweaks—that decide whether AI-powered e-commerce and digital services can actually work at scale.

TechCentral’s South African Newsmakers of 2025 list reads like a map of that infrastructure layer. If you run an online store, a marketplace, a fintech product, a subscription service, or a delivery-heavy business, these decisions show up in your unit economics. Better connectivity lowers acquisition costs and raises conversion. Better data capacity makes personalisation affordable. Better policy clarity reduces time-to-market.

This post is part of our “How AI Is Powering E-commerce and Digital Services in South Africa” series, and I’m going to take a clear stance: most AI projects fail here not because the model is wrong, but because the pipes, pricing, and incentives around the model are wrong. The 2025 newsmakers are the people changing those pipes.

Why telecom consolidation matters to AI in e-commerce

AI in e-commerce runs on three inputs: data, distribution, and trust. Telecoms touches all three.

When fibre and mobile markets consolidate or become more competitive, you don’t just get different pricing plans—you get different customer behaviour. Faster, more reliable internet means:

• Shoppers browse more product pages and watch more product videos (higher intent signals for recommendation engines).
• Customer support shifts from voice to messaging (more text data for AI-driven support and quality monitoring).
• Digital services can offer richer UX—real-time delivery tracking, fraud checks, dynamic pricing—without breaking.

South Africa’s AI adoption curve is increasingly constrained by connectivity costs, last-mile reliability, and platform competition. That’s why the Telkom/Vodacom/Cell C stories matter even if you’re “not in telecoms”.

Serame Taukobong (Telkom): the infrastructure-first AI enabler

Telkom’s turnaround under Serame Taukobong is a reminder that operational integration beats fancy roadmaps. His “OneTelkom” approach—integrating mobile, fibre, and data centres—pushes Telkom toward a more coherent data-led infrastructure business.

For AI-driven commerce, that has two knock-on effects:

1. More affordable data and stronger prepaid competition: AI-enabled marketing (especially video and social commerce) becomes viable only when customers can actually consume content without friction.
2. More data-centre and edge capacity: retailers and digital services can place workloads closer to users, cutting latency for real-time recommendations, payment risk scoring, and customer chat.

TechCentral notes Telkom’s 25% rise in EBITDA (year ended 31 March 2025) and dividend reinstatement after earnings growth (including the Swiftnet sale proceeds). Those aren’t just investor headlines. They signal more room for capex discipline and network stability—both prerequisites for digital services that can’t afford downtime.

Shameel Joosub (Vodacom): scale, fintech, and the “AI + payments” loop

Shameel Joosub landed two big strategic moments in 2025: the fibre transaction involving Maziv and a move to take control of Safaricom. Even if your business only operates in South Africa, the logic is familiar: bundle connectivity with financial services, then use data to improve both.

AI becomes more powerful when it can learn from the full funnel:

• ad exposure → click → browse → cart → payment → delivery → repeat purchase

Telecom operators and fintech ecosystems see more of that funnel than most retailers do. The practical result? They can:

• price risk more accurately (fraud prevention and chargeback reduction),
• offer smarter credit or “pay later” decisions,
• personalise offers based on usage patterns, not just purchase history.

This matters because South African e-commerce growth is increasingly shaped by payment confidence. If customers fear fraud or failed transactions, conversion collapses. If payment flows are smoother, AI-powered personalisation actually has a chance to pay for itself.

Jorge Mendes (Cell C): MVNO growth is a distribution story for digital services

Cell C’s JSE listing in November 2025 capped a multi-year restructuring and an “asset-light” strategy. The part e-commerce operators should care about: Cell C’s expanded role as a host for MVNOs, including fast-growing digital-first offerings.

MVNOs (and app-first telecom propositions) are quietly becoming distribution channels for AI-powered services:

• embedded shopping benefits (data-free access, discounted bundles),
• loyalty programs tied to mobile usage,
• identity and verification hooks for onboarding into financial products.

I’ve found that many online businesses underestimate partnerships with telco-adjacent brands. But if your acquisition costs are rising, bundled distribution through MVNO ecosystems can be cheaper than pure paid media—especially in December’s post-Black-Friday hangover when ad prices stay elevated.

Media and platforms: AI’s next battlefield is attention

AI in e-commerce doesn’t win at checkout; it wins in attention markets. Whoever controls viewing and discovery can steer purchasing behaviour.

That’s why Maxime Saada making the list (via Canal+ and the MultiChoice takeover) matters to digital services. A scaled media platform influences:

• what gets promoted,
• what content formats become standard (short-form, live commerce, shoppable video),
• how subscriptions and bundles are priced.

Maxime Saada (Canal+/MultiChoice): commerce will follow video

The strategic bet behind the Canal+/MultiChoice deal is scale in a fragmented content world. But from an AI commerce lens, the bigger point is this:

The more time South Africans spend in streaming ecosystems, the more shopping shifts toward recommendation-driven discovery.

Streaming normalises algorithmic feeds. That spills into commerce expectations: “show me what I want” beats “let me search.” Retailers then need:

• stronger product data (attributes, images, variants),
• better creative testing (AI-generated variations that still match brand rules),
• sharper measurement (incrementality, not vanity metrics).

If you’re a retailer or digital service provider, treat 2026 as the year to get serious about product information management (PIM) and creative operations. AI can’t recommend what your catalogue can’t describe.

Policy, trust, and IP: the unglamorous foundations of AI adoption

AI adoption in South Africa is also throttled by trust: who owns the idea, who owns the data, and who is accountable when systems fail.

Nkosana Makate: why IP clarity matters to AI products

The “please call me” saga involving Nkosana Makate is a long-running reminder that ideas become valuable only when rights are enforceable. The dispute dragged on for 17 years and ended with an out-of-court settlement—then new litigation followed.

AI teams should read that as a warning sign. As companies roll out:

• AI-generated marketing assets,
• recommendation models trained on customer interactions,
• automated customer support scripts,

you need clean agreements on:

• ownership of training data,
• ownership of outputs,
• who can reuse learnings across clients (especially if you use agencies or external vendors).

A practical rule that saves pain later: treat your prompts, fine-tuning datasets, and evaluation reports as IP-bearing assets—because they are.

Solly Malatsi: policy momentum reduces AI “time tax”

TechCentral highlights communications minister Solly Malatsi for injecting momentum into policy areas that affect investment. For digital services, the value of policy clarity is simple: it reduces the “time tax”—the hidden cost of delays, uncertainty, and compliance rework.

If you’re building AI-driven customer engagement (chat, WhatsApp commerce, identity verification, credit scoring), slower policy execution doesn’t just annoy you. It blocks partnerships and expands legal review cycles.

My take: South Africa doesn’t need perfect digital policy to grow AI commerce. It needs predictable policy. Predictability lets operators invest; investment improves networks; better networks make AI experiences feel instant instead of fragile.

What e-commerce teams should do next (a practical checklist)

The leaders on TechCentral’s list are shifting the environment. Your job is to shift your operating model to match it. Here’s a checklist I’d use heading into 2026 planning.

1) Build for a “messaging-first” customer journey

Most South African customers already behave this way. Now formalise it.

• Put WhatsApp and in-app chat at the centre of support and sales.
• Use AI to classify queries, draft replies, and route to humans.
• Track outcomes: resolution time, repeat contacts, refund rates.

2) Treat connectivity variability as a product requirement

Even with improvements, you must design for uneven networks.

• Make product pages lighter; optimise images.
• Cache key flows; reduce steps at checkout.
• Offer “save cart” and “resume” flows that don’t punish drop-offs.

3) Fix your product data before you “add AI”

AI personalisation fails when catalogues are messy.

• Standardise attributes (size, colour, compatibility, warranty).
• Enforce image rules and naming conventions.
• Create a feedback loop from returns/support into catalogue fixes.

4) Upgrade fraud prevention with AI—but keep humans in the loop

Use AI for scoring and pattern detection, not as a final judge.

• Set thresholds for manual review.
• Monitor false positives (blocked good customers) weekly.
• Tie risk decisions to customer lifetime value, not just order value.

5) Negotiate distribution, not just ads

As telcos and media platforms rebundle, distribution deals get more interesting.

• Explore bundles with MVNOs, wallets, and subscription platforms.
• Test “data-sponsored” journeys for key landing pages.
• Prefer partners who share measurement data you can actually use.

Snippet-worthy truth: AI makes marketing cheaper only when distribution and measurement get tighter.

The real opportunity for 2026: AI that earns trust

The 2025 newsmakers list is a leadership story, but it’s also a roadmap for where South African digital services are heading: more consolidation, more platform power, and more pressure to prove value fast.

If you’re trying to generate leads or grow revenue with AI in e-commerce, don’t start with tools. Start with the constraints these leaders are changing—connectivity, payment ecosystems, media attention, policy clarity, and IP rules. That’s where AI projects either compound… or quietly die.

If you want help pressure-testing your 2026 AI roadmap—customer engagement automation, AI content workflows, personalisation, or fraud controls—build it around a single question: where will AI measurably reduce cost or increase conversion within 90 days, given South Africa’s real-world network and customer behaviour?

What would you change first: your catalogue data, your messaging experience, or your payment-risk workflow?