Last month, I came back from a conference in San Francisco with a feeling I couldn’t shake.
I knew AI was moving fast; I’d been living that reality every day. But sitting in those sessions, watching the data points accumulate, the adoption curves, the compute power, the way it was already showing up in national GDPs, I realised the pace wasn’t just fast.
It was exponential. What had been a year earlier a debate about whether generative AI was hype had quickly become a conversation about how fundamentally it would change everything.
I’ve spent 2026 working at the intersection of enterprise and innovation in South Africa. The year taught me a few things. Here are five of them.
- The speed is real, but the believers matter more.
A year ago, leaders were asking, “Is this real?” This year, they stopped asking. They started putting money behind it, not always because the P&L had caught up, but because they genuinely believed AI would change how their organisations operate. That shift from skepticism to conviction matters more than any adoption metric.
- The real opportunity isn’t the model. It’s everything else.
I think of it as the 90% opportunity. AI is 10% technology and 90% everything else: the people who use it, the data that feeds it, the governance that frames it, the culture that questions it. The 90% is where the advantage will be built and where the most exciting work will happen.
- AI only matters when it improves how organisations work, how people experience services and how society benefits.
In 2026 the organisations that stood out were the ones asking how AI could improve customer experience, reach new audiences, build new products and ultimately make a difference to the people they serve.
- Data is the unsexy prerequisite nobody can skip.
We downplay it because it’s not exciting, but data is the single biggest predictor of whether AI will scale or fail. You can build a beautiful pilot with curated sample data and watch executives get excited. Then you try to scale, and you discover the data governance, the processes, the very things that got parked or didn’t get budget during the last hype cycle. Those are what stand between you and real adoption
- Trust is the currency. Guard it accordingly.
The lesson beneath all the others is this: AI is built on trust. If your solutions aren’t explainable, if they perpetuate the biases already present in your data and your teams, if you haven’t thought about the unintended consequences before, you’re spending that currency recklessly. And in the absence of a national AI policy, that responsibility falls to each organisation to self-impose. Risk-based, scaled to impact. A chatbot getting something wrong is one thing. A physical robot in manufacturing making an error is something else entirely. Responsibility isn’t one-size-fits-all. It’s proportional.
Looking back, 2026 was the year the conversation stopped being about whether AI would matter and started being about how we make it matter. Responsibly, inclusively, and together.
The technology is commoditising. The model is no longer the end goal. What separates the organisations that will thrive from the ones that won’t is everything else: the data, the people, the governance, the culture. The 90% where the real work happens. The problems we’re facing aren’t unique to any one organisation and the idea that we solve them alone is the most expensive myth this year dismantled.


