Dailymaverick logo

Business Maverick

TECHNOLOGY

The tech challenges of 2026

Pricier devices, weirder ways of working and an exploitative data economy – all of these await us.

Young people are spending their days labelling images and moderating toxic content often for as little as R36 an hour. The cost of smart devices will rise by an estimated $70 per unit and monthly cloud bills are hitting the tens of millions. (Photos: iStock and Vecteezy) P41 lindsey future tech

Microsoft is betting on efficiency by moving processing power from the cloud to the device in your backpack.

John Press, Microsoft Surface business unit head at Core, frames the modern laptop enabled by artificial intelligence (AI) as a three-in-one value proposition. “It’s an education tool. It’s a business tool. And it’s quite frankly an entertainment tool all wrapped into one,” he says.

But although Press paints a picture of seamless productivity driven by a device’s neural processing unit (NPU), the reality of 2026 is biting back. For the average South African consumer and chief information officer (CIO), this dependency is arriving at the worst possible time, as the physical hardware required to run these AI tools is entering a crisis of scarcity.

A global shortage of memory chips, driven by manufacturers pivoting to build high-margin AI server chips, has sent prices skyrocketing. In late 2025, consumer solid state drive (SSD) prices rose 50% and RAM prices effectively doubled.

Samsung Electronics Co. Double-Data-Rate (DDR) memory modules are arranged for a photograph in Seoul, South Korea, on Friday, April 5, 2019. Samsung reported its worst operating-profit drop in more than four years, buffeted by falling memory-chip prices and slowing smartphone sales. Photographer: SeongJoon Cho/Bloomberg
A global shortage of memory chips has sent prices soaring. (Photographer: SeongJoon Cho / Bloomberg)

“You actually have guys who built gaming PCs this year stripping out and selling their GPUs and DDR5 memory for higher prices than they paid for it months ago,” said a store clerk at Computer Mania, referring to the graphics processing units and latest generation of high-speed computer memory.

For CIOs, this is a procurement nightmare, and they can expect a sticker shock of 20% to 30% on the next fleet of laptops. For the average consumer, the implications are dire too. The dream of the cheap but good smartphone is basically dead. The International Data Corporation estimates that the cost of smart devices will rise by an estimated $70 per unit because of the memory market crunch.

Instead of democratised 5G, 2026 is seeing a return to the 4GB RAM “dumb phone” tier – devices incapable of running the very on-device AI that Press and Microsoft are championing. The digital divide is being reinforced with silicon.

Deloitte’s Tech Trends 2026 report adds another layer to this squeeze: “inference economics”. Although the cost of AI tokens has dropped 280-fold in two years, enterprise usage has exploded so fast that monthly cloud bills are hitting the tens of millions.

The corporate survival strategy is to offload processing to “edge AI” – your laptop and phone. But this strategy relies entirely on you owning a device powerful enough to handle the load, precisely when these devices are becoming unaffordable.

But if you can afford the hardware, the office experience is undeniably shifting. McKinsey’s 2025 State of AI report notes that 62% of organisations are now experimenting with AI agents, and 2026 is shaping up to be the era of the “synthetic employee”.

Tools such as Microsoft’s Project Opal are designed to autonomously handle the corporate drudgery – employee on-boarding, procurement, invoice processing – with minimal human oversight. This has given rise to a phenomenon known as “vibe working”.

Professionals are now focusing on high-level goals while AI agents handle the execution. It sounds idyllic, but it introduces a massive governance gap.

Ria Pinto, general manager and technology lead at IBM South Africa, warns that although the capability is there, the strategy often isn’t. “Fragmentation, not capability, is the real barrier,” she says.

Hiding in the shadows

Although 67% of South African enterprises claim to have adopted generative AI, only 14% have a formal strategy. The result is lurking potential dangers.

About 32% of employees are now using unauthorised AI tools to do their jobs. In the era of the Protection of Personal Information Act, pasting customer data into an unvetted server to “just get the job done” is a compliance time bomb waiting to detonate.

And, of course, there is a darker side to AI-driven efficiency that rarely makes it into the glossy keynote presentations: behind the magic of every AI agent lies a human worker.

South Africa is increasingly becoming a hub for the invisible labour of data annotation. Young people are spending their days labelling images and moderating toxic content for global tech giants, often for as little as R36 an hour.

As Rennie Naidoo, research director at the Wits School of Business Sciences, put it in a recent blog: “Africa is not just participating in the AI revolution. It’s powering it ... with something far more indispensable: human labour.” It is a precarious, psychologically taxing economy that sits uncomfortably alongside South Africa’s 40% unemployment rate.

Perhaps the most chilling shift in 2026 is how technology is reshaping crime and our physical safety. Kidnappings in South Africa rose 6.8% in 2025, but the profile of the victim has changed. Criminals are no longer just targeting the ultra-wealthy. They are mining social media data – Strava runs, geotagged school photos, “checked in” selfies – to profile people’s routine and location.

“Kidnapping is no longer a crime driven by wealth; it’s now a fast way to make quick cash using digital footprints,” warns Richard Frost, product manager and consultant at Armata Cyber Security. “Oversharing is building a high-resolution profile that can be exploited in minutes.”

So be warned: leisure in 2026 is going to re-quire digital discipline – locking profiles and dis­abling geotags.

Small business opportunity

But, for all the gloom, there are genuine opportunities. The same hardware dependency that threatens the poor offers a lifeline to the agile.

Press sees an opening for small and medium enterprises (SMEs). “We definitely see a really big opportunity in SMEs... Putting these tools in the hands of that type of business user gives them an advantage.”

An SME using AI-enabled hardware to automate its accounts, marketing and customer service could punch well above its weight, bypassing the traditional scaling hurdles that strangle local start-ups.

And on-device AI – processing data on your laptop’s NPU rather than the cloud – could save South African businesses millions in data costs. It is an unsexy bene­fit, but it matters in an economy that is defined by high overheads.

If the AI wave crashes, we are looking at a future of expensive hardware, privacy breaches and a workforce displaced by synthetic colleagues that don’t even do their work properly. DM

This story first appeared in our weekly DM168 newspaper, available countrywide for R35.

Comments

Scroll down to load comments...