Expression of Interest For Provision Of Gpu/ Ai Compute Capacity For Smart Africa Member States And The African Public And Private Sectors

Expression of Interest for Provision of GPU/ AI Compute Capacity for Smart Africa Member States and the African Public and Private Sectors OBJECTIVE OF THIS REQUEST OF EXPRESSION OF INTEREST The objective of this Request for Expression of Interest is to identify and pre-qualify privatesector Service Providers with the technical, operational, and financial capability to provide GPUenabled AI compute capacity that can be made available to Smart Africa Member States and to public and private users across the continent. Specifically, through this REOI, Smart Africa seeks to: a) Facilitate GPU-enabled AI compute capacity across the continent, accessible on demand to Member State institutions and to public and private organizations; b) Facilitate Service Providers able to participate in a Proof-of-Value (PoV) demonstrating pooled, federated, and cross-border GPU capacity sharing (uberization of GPU capacity); c) Establish a shortlist of qualified Service Providers eligible to participate in the PoV and, subject to results, in subsequent scale-up phases facilitate access to market for the shortlisted organizations. This REOI is the first stage of a phased process. Qualified Applicants may be invited to a second stage (Request for Proposals and/or PoV participation). Submission of an Expression of Interest does not create any obligation on Smart Africa to contract, procure, or commit funding. 3. SCOPE OF THE INITIATIVE AND EXPECTED SERVICES Service Providers expressing interest should be capable of providing, or partnering to provide, all or part of the following GPU compute capabilities and services. The exact scope, volumes, and locations will be refined during the PoV design phase with shortlisted Applicants. 3.1. GPU Compute Capacity Provision Provision of GPU-enabled compute capacity (training, fine-tuning, and inference) suitable for AI and high-performance workloads, available to public and private users across Member States A differentiated offering combining high-performance GPU capacity for advanced workloads and distributed or CPU-augmented capacity for scalable inference and broad access. On-demand, elastic provisioning models (e.g. pay-as-you-go, reserved capacity, or subscription), with transparent pricing accessible to African users. Eligible Service Providers include but are not limited to: hyperscale and cloud computing companies, colocation/data-center operators, and specialized GPU-as-a-Service providers, operating individually or in consortia. 3.2. Federation, Interoperability, and Uberization of GPU Capacity Ability to participate in a federated GPU marketplace that pools capacity across multiple providers and jurisdictions and allocates it dynamically to where demand exists. Support for cross-border capacity sharing, for example, enabling a user in one region to access GPU capacity hosted in another region through interoperable platforms and APIs, with appropriate commercial settlement (e.g. commission or revenue-sharing). Adoption of open interoperability standards for identity, workload portability, scheduling, and billing to avoid vendor lock-in. 3.3. Hosting, Connectivity, and Energy Readiness Access to, or operation of, data-center facilities with adequate power availability, energyready locations, cooling, physical security, and connectivity to support GPU workloads. Demonstrable approach to reliable and, where feasible, sustainable energy sourcing for high-density compute. Network connectivity sufficient to support distributed and cross-border AI workloads. Applicants should indicate current or targeted Power Usage Effectiveness (PUE), access to renewable/green energy sourcing, and alignment with regional grid reliability and connectivity benchmarks. 3.4. Security, Data Sovereignty, and Governance Implementation of cybersecurity, identity and access management, encryption, monitoring, and incident-response controls appropriate for critical digital infrastructure. Ensure full compliance with Member States data sovereignty, data localization, and data protection requirements, including, where applicable, the AU Malabo Convention, national data protection legislation, and emerging continental frameworks such as the Data Embassy and Regional Sovereign Cloud model currently under consideration by the Africa AI Council. Support for workload isolation, multi-tenancy controls, and auditability. 10t h floor , Career Center , KG541St , P.O.Box 4913 Kigal i Website: http://www. smartafrica.org | Emai l : [email protected] · Page 4 3.5. Use-Case Enablement and Ecosystem Support Support for priority public-interest use cases (e.g. government services, health, agriculture, education, research, African-language AI, and innovation ecosystems). Provisions for research and start-up access, capacity building, and local skills development. Tender Link : https://www.developmentaid.org/#!/tenders/search