ITWeb contributor Phillip de Wet.
2026 is going to be quite something.
The good old CES in Las Vegas is going to show just how much artificial intelligence (AI) we now have at the edge. Memory shortages are going to bite everywhere. We’re going to see new chip approaches allowing cool new systems. There will be a lot of pressure to open up app stores and social media protocols, and every other closed ecosystem that acts as a platform.
But my bet for the big tech story of 2026 is mundane compared to all of that: AI regulation. Of the kind that gets users and vendors scrambling, and which upends entire business models, and brings projects to a screeching halt.
AI is a cross-cutting issue for just about every regulator you can think of, in South Africa including the powerful trio of ICASA, the Competition Commission, and the Financial Sector Conduct Authority.
In 2025, AI forced those regulators to ponder philosophical questions. In 2026, they’re going to have to deal with real-world products: automated decisions about everything, AI advice on everything, dynamic everything. They’re going to be snowed under by complaints from consumers denied access to credit, by companies having their intellectual property ripped off, and by those who suffer the consequences of foolish one-to-one swaps of humans for AI systems.
And regulators are, by design, work-shy. They won’t try to cut every head off the hydra, they’ll go for the heart, and seek to regulate the use of AI itself.
May God have mercy on their souls.
(That is not counting regulation at the international and mega-power level, by the way; America will likely remain confused and the European Union tentative, while the technocrats are forced into action.)
Related to that will be the rapid rise of vertical AI solutions. General-purpose LLMs were fun, but everyone wants domain-specific agents, ideally with a deterministic flavour to their workflow, that are auditable to some extent.
Also, let’s be honest, if you’re writing your own prompts, the buck stops with you. If you have a vendor claiming to provide a system for your industry, then you have someone to blame.
My bet for the second most important trend in 2026 is true cloud portability, of the kind AWS and Google are now selling as a product.
Sure, the hyperscalers want to lock you in with their amazing software or proprietary integration layer or case-specific pricing or whatever – but their competition isn’t each other anymore, it is hybrid and on-prem. They have to show that you can get out if they can’t satisfy national data sovereignty rules in your territory, and they have to offer easy fail-over when they keep falling over.
In 2026, for the first time, the cloud will become a true commodity, with the ability to swap providers for basic services without it becoming a multi-month project with dire risks and mammoth costs.
That will be great for innovation. Just ask the mobile industry how number portability and the resulting churn focused everyone’s minds and filled out the product ecosystem niches.
The impact of cloud standardisation and portability will build slowly at the start, but it is going to be one heck of a snowball before long.
The third trend I expect to see in 2026 is quantum computing.
Yes, really.
Most of the noise will still be around the transition to post-quantum crypto, which is starting to give off serious Y2K vibes, just without the firm deadline. We know there will be at least some trouble, if we fix everything in time there won’t be any trouble and lots of people will complain it was all a scare story, we can’t really test the fix until the whole system starts taking real-world strain, and it is bloody awful to have to dig into ancient code you’d rather pretend doesn’t exist.
But if you haven’t been paying attention recently, the actual quantum computing side of post-quantum is coming fast too. Hybrid solutions that plug quantum machines into standard HPC are in production testing, and the only question is whether chemistry or meteorology will be the first field to be disrupted.
Once one sector gets a magical computing breakthrough, everyone is going to want one – and the people who have been banging on about the encryption apocalypse might finally get some attention. At which point you’d better have a plan, or a plan for explaining why you don’t have a plan.
Or perhaps 2026 is the year the European Union loses patience and bans X and kicks off a platform war, or Russia starts cutting subsea cables on an industrial scale and makes the internet archipelagic.
But let’s hope not.
