Locally hosted AI is emerging as an option for Kenyan businesses seeking to adopt artificial intelligence without investing in large computing facilities or sending every request to an external cloud service. Technology firms Dell and NTT Data are targeting smaller businesses with compact server systems that can run selected AI applications within company premises.
The approach could give businesses greater control over sensitive information while allowing them to introduce AI tools for specific operational tasks. Andrew Ngunjiri, general manager for technology solutions at NTT Data, says companies can begin with smaller deployments rather than building the extensive infrastructure traditionally associated with enterprise computing.
“You don’t necessarily need to have a whole floor or a building such as this to get the benefits of sovereign or enterprise AI,” Ngunjiri said.
He described the concept as “AI in a box”, in which businesses can run smaller models to support their applications using equipment installed closer to where their data is generated and processed.
How smaller AI systems work
Small language models (SLMs) are designed to perform language-related tasks with fewer computational resources than many large language models. Depending on their capabilities, they can support applications such as document summarisation, information extraction, internal knowledge search and customer-service assistance.
Their smaller computational footprint makes them suitable for deployment on a wider range of hardware, from capable laptops to dedicated servers. The equipment required depends on the model, the complexity of the tasks and the number of people using the system. A model running on an individual employee’s device, for example, serves a different purpose from a server supporting AI applications across an entire organisation.
For businesses, the attraction lies in being able to select the infrastructure that matches a particular workload. A company could use a smaller model to search internal documents or process routine enquiries while relying on larger cloud-based models for tasks that demand greater computing capacity.
The arrangement also allows businesses to combine local and cloud resources rather than committing every application to a single environment. NTT Data says companies are considering hybrid deployments that take account of workload requirements, costs and regulatory obligations.
Why Kenyan businesses may want AI on-site
Keeping AI processing within company-controlled infrastructure can help businesses manage how sensitive information is handled. Organisations working with confidential customer records, financial information, contracts or internal documents may prefer to process selected workloads locally instead of transmitting their contents to an external AI service.
Local deployment can also reduce dependence on a continuous connection to an external model provider for applications that run entirely on-site. Response times may improve where network latency is a constraint, although actual performance depends on the server hardware, model and workload.
Data protection is another consideration. Kenya’s Data Protection Act, 2019 regulates the processing of personal data and establishes conditions governing certain transfers outside the country. Some categories of processing face additional localisation requirements, but the law does not require all business data to remain in Kenya.
Locally hosted AI can help organisations implement their chosen data-handling controls, provided the systems are properly configured and managed. Businesses must still address access permissions, cybersecurity, retention, backups and the handling of information sent to connected services.
The operational question is therefore which applications benefit from local processing and whether the deployment provides adequate protection for the information involved. A server located inside an office does not automatically make an AI system secure or compliant.
The cost of bringing AI into business premises
Smaller systems offer businesses another route into AI adoption because they can begin with a defined application and add capacity as their needs develop. They may avoid some of the infrastructure commitments associated with operating large computing facilities, particularly where the workload can be handled by a modest server.
The financial comparison, however, extends beyond the purchase price of the equipment. Businesses must account for electricity, maintenance, software, security updates, technical support and the staff needed to manage the systems. Hardware with specialised processing capabilities may also require additional investment.
Cloud services have their own cost structure, including subscriptions and usage-based charges, but they allow businesses to access computing capacity without managing all the underlying equipment. For an organisation using AI occasionally, a cloud subscription may be more economical than purchasing and maintaining a dedicated server. A business processing large volumes of repetitive documents or internal records may find that local infrastructure offers better value over time.
The strongest use cases are likely to be those with a clear operational purpose and enough recurring activity to justify the investment. Processing invoices, retrieving information from internal records and assisting staff with routine documentation are examples of tasks that businesses could evaluate, depending on the capabilities of the selected model.
Kenya’s AI ambitions and the question of data sovereignty
The commercial push comes as Kenya implements its National Artificial Intelligence Strategy 2025–2030, which identifies digital infrastructure, data governance, research and innovation as priorities for developing the country’s AI ecosystem.
The strategy provides a policy framework for expanding AI capabilities, while businesses must decide how to translate those ambitions into practical applications. The availability of smaller systems could widen the range of organisations able to experiment with AI without first building extensive computing infrastructure.
Kenya’s wider technology market also illustrates the importance of matching infrastructure to demand. Large data-centre projects require capital, reliable electricity, cooling, connectivity and sufficient customers to justify their operating costs. Smaller on-premise deployments serve a different part of the market, allowing individual organisations to run selected workloads without constructing facilities of their own.
Data sovereignty adds another dimension to the decision. Control over where information is processed matters, but effective sovereignty also depends on who administers the systems, who can access the data, how security is maintained and which external services the applications depend on.
That distinction is relevant as businesses adopt AI tools through different routes. Some may use locally installed models for sensitive internal workflows, while others continue to rely on cloud platforms for applications that require more computing power or broader integration.
Where local servers fit into enterprise AI
The market for locally hosted AI sits alongside the development of AI-enabled devices and managed cloud services. Computer manufacturers such as ASUS have also promoted devices capable of running selected AI workloads locally, giving individual users access to some features without sending every request to a remote service.
Server-based deployments address a broader organisational need by allowing multiple users or applications to draw on shared computing resources. Their value depends on the workloads they support, the quality of integration with existing business systems and the ability of IT teams to maintain them.
For Kenyan SMEs, the practical decision will centre on the applications they need, the sensitivity of the information they process and the costs they can sustain. A carefully chosen local deployment could give a business more control over a specific AI workflow, while cloud services remain available for tasks that require greater capacity.
Dell and NTT Data’s approach places smaller infrastructure within that range of choices. As businesses assess how AI fits into their operations, the ability to start with a targeted application and expand according to measurable business needs could make local deployment a more practical consideration.
Real ESG impact doesn’t happen in panels alone, it happens in the rooms where financiers, operators, and policymakers actually align. Our GreenShift Forum 2026 cuts the noise, bringing together the people rewiring Africa’s sustainability and energy frameworks for one focused day in Nairobi. Secure your seat.
Go to TECHTRENDSKE.co.ke for more tech and business news from the African continent and across the world.
Follow us on WhatsApp, Telegram, Twitter, and Facebook, or subscribe to our weekly newsletter to ensure you don’t miss out on any future updates. Send tips to editorial@techtrendsmedia.co.ke
