Unpacking the power and potential of choice in enterprise AI

By Johan Scheepers, Senior Manager: Solution Architects at Red Hat Sub-Saharan Africa

Enterprise CEOs and CIOs are asking their technology teams very valid questions. How are we embracing AI? How can AI enhance our organisation? What is the projected return on investment?

That last question has become very important for enterprises, and rightfully so. According to Bain & Company’s recent Automation and AI Pathfinder Survey, nearly 40% of companies worldwide report cost reductions from AI projects of less than 10%. This is despite targeting around 11% to 20%, with 90% of those companies then still increasing their budgets. As projects move from experimentation to production, justifying spend becomes all the more important for any enterprise, regardless of operating margins.

Many wonder whether budgets are spent on enhancing internal services and functions, or if transforming front-facing services like customer support, marketing and communications will yield higher value.

But before enterprises can unlock that value, they must consider the fundamentals and put the proper foundations in place. Enterprise AI is driven by choice: the models that organisations choose to deploy, how those models perform and align with their use cases, the level of control organisations have over their systems, and the freedom to direct their AI journeys.

Playing by future rules: Shadow AI and proactive governance
Africa’s regulatory environment, and by extension its IT operating environment, is quickly evolving and mandating how AI is to be implemented responsibly. Though legislation and policies may not yet be set in stone, the risks that enterprises face with non-compliance are. Enterprises find themselves facing cybersecurity threats, biased outcomes, and compliance fines. This can lead to hefty costs and lasting reputational damage.

Those risks are not limited to regulation alone. They are also influenced by how AI tools are adopted and governed within organisations. During the last few years, enterprise employees have adopted software and resources without the knowledge of infrastructure and network operators. Those resources are not always aligned with business and IT governance protocols. Today, the same thing is happening with AI tools, giving rise to what is referred to as ‘shadow AI’. Even though shadow AI is not always rooted in malice, employees deploying their favourite tools, often in the name of improving productivity and efficiency, can misalign with the objectives of the organisation.

The solution is to become proactive. As AI governance becomes more important and enterprises prepare for stricter regulations and frameworks, any strategy needs to reflect high levels of transparency, auditability, flexibility and control. Those levels then set the stage for sustainable innovation, which enterprises can leverage with the help of a platform that allows for the seamless integration of tools and can be customised to create a governance setup tailored to specific needs and industry regulations.

Any model, any accelerator, anywhere
The future of enterprise AI is not black-and-white. In other words, it’s not a question of whether organisations rely on large, foundational proprietary models, running on hardware from a single provider.

Instead, AI strategies will rely on a multi-vendor, interoperable architecture made up of layers that determine how vendor capabilities are best suited. This includes the top-end business layer, down to the data, application and technology layers. Backed by an ecosystem that incorporates major providers and more specialised architectures, organisations retain the option to switch resources and vendors as needed.

For any organisation to generate value from, or reduce costs using their AI investments, they need to be able to deploy any model, on any accelerator, across any environment, and deliver a consistent user and developer experience. Open source, and an inference platform built on its principles, helps them do that. It creates a standard for consistent and agile AI innovation. Whether the use case demands the use of a frontier model or a smaller, more specialised model, organisations are free to use whatever models they need to achieve the desired outcomes.

Model behaviour through open platforms
South African enterprises are increasing their investment in AI, with capital directed to high-impact areas such as agentic AI and workforce transformation. Central to that spending are the foundations enterprises put in place to sustain that transformation.

For many, that foundation will be built on open principles. Enterprises and their IT teams, looking to achieve operational consistency through the most cost-efficient means possible, should prioritise open platforms that simplify the adoption process and uphold the flexibility of hybrid cloud.

An open platform enables them to not only streamline model training and development, but also move workloads between environments with ease and without having to worry about interoperability. Platforms also let them expand their focus and explore new opportunities, such as agentic AI workflows. Additionally, they enable model observability and governance, meaning organisations can adhere to all governance requirements, which will become more critical as South Africa establishes and enforces AI sovereignty regulations.

At the end of the day, freedom of choice is how local enterprises generate value from their AI investments. It also lets them keep up with the pace of innovation, as new advancements in models, systems and infrastructure are announced regularly. By prioritising an open strategic approach, enterprises can remain agile and open to the possibilities of AI, both present and future.

END

READ ALSO : The Impact of Artificial Intelligence in Transforming Modern Homes in 2025