In the fourth part of the AI and the channel article, we asked what’s holding back those companies that are not getting full value.
Werner Herbst, MD of First Distribution
The biggest constraints are not access to AI, but organisational readiness.
Key challenges include skills shortages, data quality and availability, and the complexity of integrating AI into existing environments. Cost and regulatory considerations also play a role, particularly in the African context.
Importantly, many organisations are still thinking in terms of tools rather than transformation. Until there is a shift towards rethinking operating models, value will remain constrained.
Ravi Bhat, commercial solutions and AI officer at Microsoft South Africa
The biggest constraint is not the technology; it is organisational readiness. The Microsoft 2026 Work Trend Index shows that culture, manager support and talent practices account for more than twice the AI impact of individual behaviour alone. Many employees are ready to move faster; the operating model around them often is not.
Across Africa, structural barriers also matter – electricity, connectivity, digital skills, affordability, local-language access, data quality, governance and security. These foundations determine whether AI remains experimentation or scales into lasting value, underscoring that the question is no longer whether AI can create value. The question is whether organisations have the foundations to capture that value at scale.
Barry Buck, chief technology officer at Saucecode
At enterprise level, it’s governance and compliance red tape — full stop. I won’t pretend the caution is wrong; the stakes are high, and a badly-informed deployment can do real harm to customers and the business. But there’s a quieter brake too: vendor lock-in. Too many companies are tied to a supplier that never exposes them to the best frontier models — and you can’t get breakthrough returns from second-best tools.
Ebrahim Essop, senior automation solution manager at Nedbank CIA Robotics
Key constraints include infrastructure, data quality and accessibility, skills shortages, integration complexity with legacy systems, and evolving regulatory requirements. In many cases, organisations also struggle to move from pilot initiatives to scaled, enterprise-wide adoption.
Othelo Vieira, technical product manager lead at Cloud On Demand
Data quality is the number one constraint — and it’s consistently underestimated. Organisations often don’t know how poor their data is until they try to train a model on it. Behind that, the skills gap is severe: there aren’t enough data scientists, ML engineers, or AI-literate business leaders on the continent. Cost is a real barrier for mid-market and smaller companies — cloud compute for training and inference adds up quickly, and rand depreciation makes dollar-denominated cloud costs painful. Regulatory uncertainty around data privacy and cross-border data flows, and the absence of clear internal AI governance frameworks, round out the picture. The companies struggling most are those that treated AI as a technology project rather than a business transformation.
Andre Hogewoning, chief operating officer at Business AI
The technology itself is rarely the problem. The real challenges are human, organisational and methodological — and they are all solvable with the right approach and the right guidance.
The core challenges facing AI adopters are: fear and reliability concerns; understanding what AI actually is; data quality, data sovereignty and security; cost of implementation; applying traditional software development methods to AI projects; and regulatory and governance uncertainty.
Across all these challenges, education is the foundation upon which every successful AI journey is built.
Ryan Martyn, co-founder and chief marketing officer of Syntech Distribution
The main barriers are data security, compliance requirements, and the absence of consistent, accurately structured data to work from. On top of that, there’s a skills gap, both in using these tools properly and in adopting the processes needed to embed them into the business or legacy systems. Without addressing all of these, corporates will struggle to move past pilot projects.
Ziaad Suleman, senior vice-president and CEO at Cassava Technologies, SA and Botswana
The primary bottlenecks are a lack of high-performance localised compute infrastructure and the high cost of exporting African data to foreign clouds. This is further compounded by a critical specialised digital skills deficit and limited access to local GPU-as-a-service capabilities.
Beyond infrastructure, many organisations lack a clear AI strategy. Even where ambition exists, the experience and capability to execute are often missing. Cassava Technologies partners with customers to assess AI readiness, support implementation, and build local expertise. This partnership approach aims to foster a collective confidence that together, we can unlock Africa’s AI potential.