A federated cloud grid has been launched to enable multi-sector AI and data services across East Africa, allowing organizations to share compute resources while keeping data sovereign. The grid, which is currently operational in Uganda, Kenya, Rwanda, and other East African countries, allows clinics, ministries, lenders, and universities to tap into shared compute resources while complying with local data-residency rules. This is achieved through a platform that links independently owned sites, enabling organizations to share resources while maintaining control over their data.
The federated cloud grid has several use cases across different sectors. For instance, a private clinic group operating in central Uganda has contracted with a local health-IT provider to run its workloads on capacity supplied by ABQ Cloud, the Ugandan Cloud Enabler. This arrangement allows the clinic to keep patient records within Uganda and back them up to a second local site. The clinic's clinical-decision-support tool runs on a small open medical model that shares a sliced GPU with other tenants, ensuring efficient use of resources.
Another use case is the East African revenue authority, which runs its sovereign tax-filing system on a dedicated grid location under a Build-Operate-Transfer arrangement. This arrangement allows the authority to embed its engineers for the early years and retain a copy of the platform's source code before handing full control to the public institution. The authority can keep classified workloads in the sovereign site, while non-sensitive services can burst to other federated locations during filing peaks.
A digital lender serving traders in Uganda, Kenya, and Rwanda is also using the federated cloud grid to deploy its services. The lender keeps a single contract with its Kampala-based provider, deploying to federated sites in each country to ensure that customer data never leaves national borders. The lender's credit-scoring and fraud-detection models run as inference workloads on shared accelerators next to the data, making efficient use of resources.
The federated cloud grid is also being used by a university in Kampala for its learning platform, which serves twenty thousand students. The university is also using the grid for research compute to fine-tune Luganda and Runyankole language models. If the university operates its own small site, it can return spare capacity to the federation during holidays, making efficient use of resources.
The grid's key properties, including data residency enforced by platform rules, resilience through geographic distribution, and capacity that follows demand, are already in operation in at least one Ugandan location. The grid also features backup and GPU slicing, ensuring that resources are used efficiently and effectively. Future work includes expanding the federation to more African sites and finalizing the rulebook governing admission, service levels, and dispute resolution.
The development of the federated cloud grid is a significant step towards enabling multi-sector AI and data services across East Africa. The grid has the potential to transform the way organizations operate, enabling them to share resources while maintaining control over their data. As the grid continues to expand, it is expected to have a major impact on the region's technology landscape.
Key points
- The federated cloud grid allows organizations to share compute resources while keeping data sovereign.
- The grid is currently operational in Uganda, Kenya, Rwanda, and other East African countries.
- The grid's key properties include data residency enforced by platform rules, resilience through geographic distribution, and capacity that follows demand.
