Launch Vector
- ✅ Private beta training circuits live in Johannesburg
- ⚙️ Provider onboarding mesh with automated hardening
- 🚀 Public launch ETA once latency audits finish
Mafutai — Africa’s GPU Supergrid
We are fusing bare-metal performance, remote GPU nodes, and generative ops pipelines into one local-first AI cloud. The control room is spinning up — expect neural-grade speed, sub-millisecond routing, and a marketplace for every serious model builder on the continent.
Go-live countdownTarget: 1 Jan 2026
68Days
12Hours
18Minutes
We’re testing and hardening the supergrid throughout 2025. Full public rollout kicks off in January 2026.
Tech Stack DNA
Hybrid Vultr + bare-metal rigs, containerized micro-schedulers, Celery pipelines, Stripe orchestration, and a rootless agent framework keeping GPUs humming. Observability is Grafana + OpenTelemetry, sealed with WireGuard lanes.
Why It Hits Different
Latency is local, power is priced in ZAR, and every node is vetted for AI inference, training, and RAG workloads. Get your workloads closer to your users, keep data sovereignty intact, and scale without the overseas tax.
Realtime GPU Telemetry1.2msaverage router hop inside SA backbone
Compute BackboneRTX Pro 6000Leadtek NVIDIA RTX Pro 6000 Blackwell cards with 96GB GDDR7 powering inference and training.
Edge AI Delivery12partner nodes online + scaling weekly
Provider MarketplaceQ4open listings with automated compliance