About
Subscribe
  • Home
  • /
  • Channel
  • /
  • Local AI cloud project puts over 400 Nvidia GPUs to work

Local AI cloud project puts over 400 Nvidia GPUs to work

Admire Moyo
By Admire Moyo, ITWeb news editor
Johannesburg, 27 Aug 2026
The infrastructure is expected to deliver 7.2 exaflops of computing performance. (Image generated using ChatGPT)
The infrastructure is expected to deliver 7.2 exaflops of computing performance. (Image generated using ChatGPT)

South African cloud provider Stratos Lab, AI infrastructure company Ecoblox and centre operator Parks Africa are partnering to deploy an artificial intelligence (AI) cloud infrastructure platform in South Africa, using more than 400 Nvidia graphic processing units (GPUs).

The companies announced the deployment yesterday, saying it will comprise more than 50 Nvidia B300 HGX servers supplied by Giga Computing, a Gigabyte Group company.

The infrastructure is expected to deliver 7.2 exaflops of computing performance and will target AI and high-performance computing workloads.

The deployment will be hosted at Digital Parks Africa’s data centre infrastructure, with Stratos Lab providing the cloud and compute services, and Ecoblox handling systems integration and infrastructure deployment.

The companies say the project is aimed at providing businesses and developers with access to high-performance AI computing capacity within Africa, reducing the need to send workloads to overseas infrastructure.

Leaping ahead

Chris Mostert, CEO of Stratos Lab, says the deployment will provide local access to high-performance computing for AI development.

“Deploying Africa’s most advanced AI cloud powered by Nvidia B300 GPUs represents a quantum leap for regional innovation. By integrating our specialised GPUaaS orchestration and managed PaaS capabilities with modular AI infrastructure, we are giving African enterprises and developers the raw, high-performance compute power needed to build world-class AI models locally.”

Ecoblox CEO Doug Makishima says the project is intended to address challenges involved in deploying high-density AI infrastructure.

“Solving the AI scaling bottleneck requires a fundamental shift in how compute environments are deployed,” he says. “Ecoblox is proud to contribute our system integration capabilities, from design and deployment through operations. Together with our AI infrastructure software stack, we are establishing a rapid, scalable and foundation for sovereign AI.”

Digital Parks Africa will provide the data centre facilities and power infrastructure required for the GPU deployment.

Ecoblox CEO Doug Makishima.
Ecoblox CEO Doug Makishima.

“Ultra-dense AI workloads demand an unyielding physical foundation, specialised power management and pristine environmental controls. Digital Parks Africa is proud to provide the carrier-neutral, highly-resilient data centre facility required to sustain these 50+ Nvidia B300 clusters, ensuring local data remains secure, accessible and truly sovereign,” says Menno Parsons, CEO of Digital Parks Africa.

Giga Computing is supplying the GPU servers for the project. Hsueh-Li Wang, senior sales director for the Middle East, Turkey and Africa at Giga Computing, says the company will support the deployment through its production and engineering resources.

“We are in a great position to support this strategic alliance with our R&D resources, cost optimisation through effective sourcing and priority production commitment. Every bit of Stratos Lab’s GPU compute power will be capitalised at absolute maximum effort,” says Wang.

AI computing demand

Stratos Lab will offer the resulting infrastructure through GPU-as-a-service, managed platform-as-a-service and bare-metal computing services. The platform is intended to support workloads such as AI model training, inference and AI-agent applications.

Menno Parsons, CEO at Digital Parks Africa.
Menno Parsons, CEO at Digital Parks Africa.

Ecoblox will be responsible for GPU cluster design, procurement, installation and optimisation, as well as the infrastructure software stack. Digital Parks Africa will provide the physical data centre environment and power management.

The companies did not disclose the investment value of the deployment, its expected operational date, or how much of the capacity has already been contracted by customers.

The announcement comes as demand for AI computing capacity increases, while African organisations face constraints around access to high-end GPUs, data centre power and cooling infrastructure.

The use of Nvida’s B300 platform also places the project among the newer generations of accelerated computing infrastructure.

Share