South African organisations have embraced AI almost universally. Far fewer can show that it pays.
The SA AI Index 2026, launched on Friday by research house World Wide Worx in partnership with Altron Digital Business, found that 93% of large South African organisations now use generative AI in some form. World Wide Worx’s earlier studies put that figure at 45% in 2024 and 67% in 2025.
The index is based on a telephonic survey of 100 decision-makers at organisations with more than 200 employees, conducted between September and November. World Wide Worx CEO Arthur Goldstuck argued that, because they include many of the country’s biggest employers, they represent a large share of the workforce. But figures for subgroups rest on small numbers.
Much of the use is happening outside official channels. The share of organisations using both approved and unofficial tools jumped from 20% to 54% in a year, and 75% now report some “shadow AI” use, which Goldstuck warned at the launch exposes organisations to attackers.
Only 51% of users rate AI’s impact on productivity as positive, and 37% and 38% say the same of competitiveness and customer service.
SA AI Index 2026
Yet only 51% of users rate AI’s impact on productivity as positive, and 37% and 38% say the same of competitiveness and customer service.
Goldstuck presented this as a collapse from 96% positive in 2024. But World Wide Worx’s earlier figures combined “positive” and “very positive” ratings. In its 2025 Generative AI Roadmap, produced for Dell Technologies and Intel, 83% rated AI’s effect on productivity as positive, but the “very positive” share alone was 51%. It is not yet clear whether the 2026 figures use the same basis.
Goldstuck puts the shift partly down to who is now using AI. “The early adopters were enthusiasts who went looking for gains,” he said in a statement. “Today almost everyone uses AI, often in a shallow way, and many organisations are discovering that a language model on every desk does not automatically translate into business results.”
Measure it or lose it
Only 27% of organisations can measure AI’s value at all. “This last number is the scary number for a lot of organisations,” Goldstuck said.
Measurement turns out to be the dividing line. Among organisations that can measure AI’s value, 78% rate its effect on productivity as positive, against 37% of those that cannot. On competitiveness, the split is 63% against 23%. “It’s no good just using it if you can’t measure the impact,” Goldstuck said.
Most cannot: 73% have no tools to assess whether AI delivers value, and 71% cannot put a figure on cost savings. Value is the weakest of the index’s five pillars, at 16.6 out of 100.
Getting AI into production matters as much. Among the 30% of organisations with AI running in a live business process, 80% report a positive effect on productivity and 60% on customer service. Among the 52% still experimenting, those figures fall to 44% and 31%.
Altron Digital Business MD Craig Stewart called the findings “very much a leadership and a cultural challenge that business leaders need to solve”. Altron itself generates “billions of token usage”, he said, “but the bigger challenge comes in when we look at operational AI and the impact on some of those key business metrics”.
Doug Woolley, Altron Digital Business’s head of sales, said AI was not always translating into “those three things that are critical: saving money, making money and reducing risk”. “People are cautious because one or two have been burnt,” he said.
Theo Spickett, Altron Digital Business’s executive for data and AI, said few customers still bother to measure gains in personal productivity. “When was the last time somebody asked you whether your word processor is making you more productive?” he said.
Banks and telcos lead
The average large organisation scores 34.3 on the index, which measures how deeply AI is built into core processes, governed, measured and absorbed by the workforce. Firms with more than a thousand employees average 42.6, against 28 for those with 201 to 500.
Financial services leads the sectors at 60, followed by telecoms at 46, which Goldstuck tied to their huge customer bases. “This doesn’t comment on the quality of their customer service, of course, for those who are fighting with their bank or their telco,” he added. The public sector fares poorly.
A companion SA AI Frontier Index, which measures how far organisations use AI to do new things, averages just 20.3. Only 1% of organisations actively use AI agents and 71% have no plans to, though more than a quarter intend to look into them.
There is “so much hype around it, and so much uncertainty about what it can deliver that organisations are steering clear of it”, Goldstuck said, describing South African firms as “traditionally very hard-nosed and sceptical”.
Accuracy and reliability top the list of barriers, cited by two-thirds of respondents, ahead of security, privacy and legal issues at 44%.
Rules, costs and jobs
Spickett said many customers were waiting for an “AI Act” to clarify the rules. South Africa has no such law: communications minister Solly Malatsi withdrew the draft national AI policy in April over fictitious sources, and a redraft is only due before cabinet by March 2027.
Cost is another obstacle. Stewart said that, in Altron’s own comparison, its local AI factory had run a large bank’s document-processing workload at a quarter of a hyperscaler’s per-query cost. But many customers were tied to “minimum commit agreements” with the likes of AWS and Microsoft, he said.
Stewart was blunt about jobs. “As hard and as callous as it does sound, there is going to be impact on employees, on certain job roles, and I don’t think we can turn away from that fact.”
As hard and as callous as it does sound, there is going to be impact on employees, on certain job roles…
Altron Digital Business MD Craig Stewart
An Oliver Wyman Forum survey published in October found that 63% of South African employees believe AI has improved their productivity, though only 44% think their organisation is ready to adopt AI at scale.
Globally, Boston Consulting Group classifies 7.5% of companies as “future-built”, generating substantial value from AI, with a further 41% actively scaling it. “The perception is that AI isn’t delivering,” said BCG global AI leader Nicolas De Bellefonds. “Our data says the picture is more nuanced: nearly half of companies are already creating real value from it, while many others are still struggling to translate investment into impact.”
The full SA AI Index report will be published on 23 November. – (c) 2026 NewsCentral Media
