On 19 March, CIO South Africa hosted an exclusive dinner at Aurum in Sandton, bringing together senior IT leaders for focused discussions on AI cost, control and value realisation.
Technology executives from multiple industries explored how AI is reshaping technology operating models while challenging the financial frameworks CIOs and CFOs have relied on for years.
The dinner, supported by CIO South Africa’s executive partner MagicOrange, created an environment for candid conversations on the realities of scaling AI, the associated costs and the measurable value delivered.
The evening opened with light icebreaker questions. When asked what they would title a book about their career, one attendee responded with, “The school of hard knocks,” noting that while the journey had not been easy, it had been fulfilling.
Another attendee shared a humorous account of the most unusual excuses he had encountered for missed meetings. “I’ve been working with young startup founders who have very little corporate experience, so almost every morning there is a new excuse for being late. The most ridiculous excuse was when they called to say they were still sleeping,” the attendee said with a laugh.
Attaching IT spend to business value
The central discussion focused on how CIOs are aligning IT spend with business value in an AI-driven environment.
“The biggest problem we’re solving in our organisation is cost. IT is a cost centre, and that means constantly justifying its existence by demonstrating value. That is the real challenge,” one attendee said.
There was broad agreement that demonstrating value requires IT leadership to develop a deep understanding of the business, including cost to serve and revenue drivers. This shift enables a more outward-looking approach, where technology decisions are driven by business outcomes rather than technical novelty.
“As CIOs, we need to ensure our IT leadership understands the business. Once they do, they begin to see how technology impacts revenue and cost, and the focus shifts away from technology for its own sake,” one CIO explained.
Another leader highlighted the need to eliminate silos between business and IT to ensure alignment on priorities and outcomes. Their organisation has recently transitioned from a project-based model to a product-based operating model to support this shift.
“This move is addressing the silo between business and IT. One of the most challenging aspects has been aligning on shared goals and being measured against them collectively,” the attendee noted.
Bursting the AI bubble
The group reflected on the early phase of AI adoption, where experimentation often preceded clear business cases. Many organisations invested in AI tools without defined use cases or measurable outcomes, leading to limited value realisation.
“Early in the AI adoption cycle, there was strong momentum to implement solutions, but many organisations invested in tools without clearly defined use cases or an understanding of the potential value. As a result, some were forced to discontinue these initiatives due to limited return. Now that the market has matured, organisations are re-evaluating how they quantify and measure value from their AI investments,” one attendee observed.
A contrasting perspective came from a CIO whose organisation is prioritising solutions that are cost-efficient and require minimal additional resources.
“I think I burst my own AI bubble. When AI emerged, we pursued as many use cases as possible. That changed once I fully understood the cost of computing power. It is not about what you can build, but what you are willing to spend to make it viable,” the attendee said matter-of-factly.
They added that AI initiatives in their organisation are divided into two categories. Automation and workplace productivity are managed by a central group technology function, while business IT focuses on solutions that directly drive measurable value.
“In business IT, the priority is to deliver solutions that contribute to the bottom line. With automation and productivity use cases, teams can spend months in research without certainty of value. That is not a trade-off we can afford,” they explained.
The cost of computing and infrastructure constraints
As the discussion turned to long-term strategy, participants highlighted the cost of computing as a major barrier to unlocking full AI value.
One IT leader argued that building proprietary models is essential for sustained competitive advantage, but remains financially challenging. “Consuming AI solutions alone is not enough. To realise real value, organisations need to build their own models, but the cost of computing is significant,” they said.
Energy costs emerged as a critical factor, particularly in the South African context. Beyond electricity, the infrastructure required to cool computing environments adds further complexity.
“The obvious cost is electricity, but the less visible challenge is cooling. South Africa faces constraints in both power and water, which makes large-scale AI infrastructure more difficult to sustain,” another attendee noted.
While the opportunities presented by AI are clear, the consensus among attendees was that success depends on disciplined, value-driven implementation. In the current economic climate, CIOs are prioritising use cases that deliver measurable business outcomes and justify the cost of investment.
Those in attendance were:
- Bramley Maetsa, head of digital and innovation enablement, Sasol
- Grace Pelo-Sithole, senior writer, CIO South Africa
- Greg Guye, co-founder and chief product officer, MagicOrange
- Makopi Nkopodi, executive of product design and innovation strategy and integration, Nedbank
- Oscar Stark, chief strategic enterprise architect, Liberty
- Pulana Ngwasheng, CIO, Zeda
- Stephen Coull, head of sales for Africa, Middle East and APAC, MagicOrange
- Thabo Modise, CIO, Hollard Insurance
