AI Newsmakers: Who’s Shaping SA Digital Commerce

Telkom’s earnings (EBITDA) jumped 25% in its 2025 financial year, Vodacom finally got a four-year fibre deal over the line, and Cell C pulled off a JSE listing after a long restructuring. Those sound like telecom headlines—and they are. But if you run an online store, a delivery marketplace, a fintech product, or any kind of subscription service, these moves land much closer to home than most people think.

In this series on how AI is powering e-commerce and digital services in South Africa, I keep coming back to one point: AI outcomes follow infrastructure outcomes. Better networks, cheaper data, more fibre reach, stronger digital payment rails, and clearer policy rules don’t just “help tech”. They determine whether your AI customer support, personalisation, fraud detection, and marketing automation actually work reliably for South Africans.

TechCentral’s South African Newsmakers of 2025 list isn’t framed as an “AI list”, but it reads like a map of the forces that will decide which digital businesses win in 2026: consolidation, platform power, and leadership choices that shape connectivity, media distribution, and consumer trust.

Why “tech newsmakers” matter for AI in e-commerce

AI in e-commerce lives or dies on three basics: data, distribution, and trust. The leaders recognised this year influenced at least one of those.

• Data: AI needs clean, consented customer and transaction data, plus reliable event tracking across your site/app, payments, and logistics.
• Distribution: If customers can’t load your storefront quickly or stream product videos without burning their data budget, your conversion rate takes the hit.
• Trust: Fraud, account takeovers, SIM swaps, and social engineering reduce adoption—especially for digital financial services.

Here’s the stance I’ll take: South African e-commerce leaders should care less about “which model is trending” and more about which ecosystem shifts make AI practical at scale. 2025’s newsmakers point to those shifts.

The connectivity layer: where AI customer experience actually happens

The fastest way to improve AI-driven customer experience in South Africa is to reduce friction in connectivity and cost-to-serve. That’s why the telecom newsmakers matter.

Telkom’s “OneTelkom” play is an AI enabler (even if it’s not branded that way)

Serame Taukobong’s Telkom story is about integration: pulling mobile, fibre, and data centres into a more coordinated infrastructure business. The very practical e-commerce implication is this: integrated infrastructure tends to create more predictable performance and pricing, which makes it easier for online businesses to run data-heavy experiences.

From the RSS story:

• Telkom shifted from a fixed-line legacy into a data-led infrastructure player.
• The sale of Swiftnet for R6.75-billion strengthened the balance sheet.
• Telkom reported a 25% surge in EBITDA for the year ended 31 March 2025.

What this means for your AI roadmap:

• More stable networks support richer UX: AI search, product recommendations, and conversational commerce work best when latency and drop-offs are low.
• Data centre capability matters: If your AI workloads are hybrid (some cloud, some local), improved local infrastructure reduces costs and improves response times.

Vodacom’s fibre and regional expansion shapes the “addressable market” for AI

Shameel Joosub’s 2025 is dominated by two big moves: the Maziv fibre transaction and the plan to take control of Safaricom in a R36-billion deal. Whether you love consolidation or hate it, there’s no ignoring the second-order effect: fibre reach and mobile money ecosystems change consumer behaviour, and AI systems are trained on behaviour.

For South African e-commerce and digital services, Maziv matters because fibre scale typically translates into:

• higher household bandwidth,
• more video-first commerce (live shopping, richer product demos),
• more reliable work-from-home buying patterns, and
• a bigger audience for AI support channels (voice, chat, WhatsApp-style flows).

For cross-border digital services, Safaricom matters because it reinforces a model many South African businesses are pursuing: bundling connectivity + payments + digital services. And bundling is where AI performs well—because it sees more of the customer journey.

Cell C’s “asset-light” model is a preview of how AI operations will be run

Jorge Mendes led Cell C through a credible turnaround and a JSE listing. The core operational shift—network outsourcing and an asset-light operating model—mirrors what I’m seeing in mid-market digital businesses: fewer owned assets, more partner ecosystems, more APIs, more automation.

Cell C’s role as an MVNO host is also a quiet indicator of what’s coming: more niche, vertically focused digital propositions riding on top of big infrastructure. Expect the same pattern in AI for commerce:

• a few large model and platform providers,
• many specialised “AI wrappers” tuned for retail, logistics, support, and fraud.

If you’re building a commerce or fintech product, treat this as strategic permission to focus on what you can own:

• your customer relationship,
• your product differentiation,
• your first-party data quality,

…and integrate the rest.

The media layer: why content rights and streaming shape commerce AI

E-commerce is already a media business. Your ads, product videos, creator partnerships, and live selling aren’t side quests. They’re the funnel.

That’s why Maxime Saada’s Canal+ takeover of MultiChoice belongs in a conversation about AI in digital services.

A hybrid video future increases the value of AI personalisation

The deal thesis is about scale and content leverage across Africa, while the pressure is obvious: global streamers normalised on-demand viewing and different price expectations.

Here’s the commerce angle: as entertainment becomes more algorithmic, consumer expectations spill over.

People now expect:

• personalised feeds,
• “continue where I left off” experiences,
• recommendations that don’t waste their time,
• quick search that understands intent.

If your online store still treats every visitor the same, you’re competing against the UX norms set by streaming platforms.

Practical moves for South African retailers and digital services teams:

1. Fix your product data first. AI recommendations won’t rescue messy catalogues.
2. Use AI to segment by behaviour, not demographics. “Browsers”, “deal-hunters”, and “repeat replenishment” outperform age/gender segments.
3. Make video shoppable. Even basic tagging and “seen in video” product links can lift conversion.

The trust layer: IP battles, payout fights, and what they signal about data value

Nkosana Makate’s “please call me” saga is a reminder that ideas, data, and distribution have monetary gravity—even when it takes the legal system years to agree.

For AI in e-commerce and digital services, the parallel is blunt:

If you can’t prove ownership, consent, and value creation, you’ll struggle to defend your data and your AI outputs.

This isn’t academic. As more businesses deploy AI agents for support and sales, disputes will move from “who owns the idea?” to “who owns the training data, prompts, transcripts, and outcomes?”

A simple operating rule I’ve found useful:

• Treat customer conversations (calls, chat, WhatsApp-style support) as regulated assets.
• Store them with clear retention policies.
• Get consent language right.
• Lock down access and audit trails.

That’s how you protect your ability to improve models without creating compliance risk.

Policy and payments: the unglamorous work that makes AI profitable

AI is expensive when your environment is uncertain. Policy clarity lowers the cost of experimentation.

Solly Malatsi’s mention in the list is about momentum and pragmatism in digital policy. Whether you agree with every move or not, the business reality is that investment follows predictable rules. And AI adoption—especially in regulated sectors like payments, lending, and telecoms—depends on investment.

On the fintech side, Lincoln Mali and Lesaka’s acquisition of Bank Zero points to a broader direction: banking and fintech are merging into software-first financial services. AI sits in the middle of that shift:

• automated onboarding and document verification,
• transaction monitoring and fraud detection,
• personalised offers and next-best-action messaging.

For e-commerce, better fintech rails mean:

• higher approval rates,
• fewer false declines,
• less manual review,
• lower chargeback exposure.

And those are direct margin improvements.

What South African e-commerce teams should do next (a 30-day plan)

The best way to respond to these ecosystem shifts is to get your AI “boring basics” right. Here’s a practical 30-day sprint that works for many South African digital businesses.

Week 1: Instrument the journey properly

• Define 10–15 key events (view item, add to cart, checkout start, payment fail, delivery ETA viewed).
• Ensure events are consistent across app and web.
• Create a single source of truth for product and pricing fields.

Week 2: Deploy AI where it reduces cost-to-serve

Start with high-volume, repeatable work:

• AI email classification for support queues
• Suggested replies for agents (not fully autonomous)
• Returns and refund policy assistant

Measure:

• first response time,
• ticket deflection rate,
• escalation rate.

Week 3: Add personalisation that you can explain

• “Recently viewed” and “Frequently bought together” (rules + lightweight ML)
• Search autocomplete tuned to local terms and misspellings
• Dynamic onsite messaging for prepaid/data-constrained users (lighter pages, fewer videos)

Week 4: Build fraud and trust signals into the funnel

• Device and session risk scoring
• Login anomaly detection
• Payment failure analysis (reduce false declines)

If you only do one thing: reduce checkout friction. In South Africa, that’s where revenue is often lost.

The leaders to watch in 2026 (and why)

The next wave of AI growth in South African e-commerce won’t come from flashy pilots. It’ll come from leaders who make scale possible. That means:

• operators expanding fibre and lowering data friction,
• digital service providers bundling payments, identity, and support,
• policymakers reducing uncertainty so investment can move faster.

If you’re planning your 2026 roadmap, use the “newsmakers” list as a lens: not “who won the headlines”, but “who changed the constraints my customers live under”.

Want a useful gut-check for your own business? Ask this: If connectivity gets cheaper and more reliable next year, do we already have the data foundations to take advantage of it—or will we waste the moment cleaning spreadsheets?