Owned capacity plus the control plane for everyone else's.
Nimbo Kinetics sells GPU-hours, tokens, Kubernetes and on-prem AI racks, and manages the GPUs customers already own. Each product pulls the next one through the same account, so contract value rises as customers move from tokens to hybrid.
Prices are rising and capital is concentrating on single-cloud models
Start with a credit card. End with a multi-year contract.
Developers land on tokens with no sales cycle. As usage becomes steady, the same account moves to dedicated endpoints, reserved GPUs, and finally hybrid: Fleet across the customer's estate, Nimbo Racks in their data center, and OpenShift AI services. Each step raises commitment and switching cost.
Six ways the same customer pays
| Line | Unit | Contract shape | Margin driver |
|---|---|---|---|
| Nimbo Compute (GPU) | GPU-hour | On-demand; 1 to 36-month reservations | Utilisation and cost of capital |
| Nimbo Turbine | Million tokens; GPU-hour for dedicated | Usage; committed spend | Tokens per GPU-hour from batching and caching |
| Nimbo Kubernetes and CPU | vCPU-hour, cluster-hour | Usage | Attaches to GPU workloads; low incremental cost |
| Nimbo Fleet | Managed GPU per month | Annual subscription | Software margin on GPUs we do not finance |
| Nimbo Rack | GPU-month | 36 to 60-month rack-as-a-service | Asset-backed, contracted, financeable |
| Services | Engagement, managed service | Fixed price; monthly | Opens enterprise accounts for the lines above |
Target gross margins and the revenue mix by year are in the data room model. Contact us for details →
No competitor covers all five
| Nimbo | GPU clouds (CoreWeave, Nebius, Lambda) | Control planes (Rafay, Spectro, Mirantis) | Hyperscalers | |
|---|---|---|---|---|
| Owns GPU capacity | Yes | Yes | No | Yes |
| Manages VMware, Proxmox, OpenShift estates | Yes | No | Partial | No |
| Managed AI racks on customer premises | Yes | No | No | Partial |
| Same token API on cloud and on-prem | Yes | No | Partial | No |
| NVIDIA and AMD under one scheduler | Yes | Partial | Partial | Partial |
"Partial" = some products in the group offer it, or it is limited to that vendor's own stack (for example AWS Outposts, Azure Local). Assessment from public product pages, Oct 2026.
The numbers diligence asks for first
ProgramsNimbo Kinetics is a member of NVIDIA Inception and AWS Startups.
Available under NDA
Contact us for details and data room access. Choose "Investor or partner" on the contact form.
- Financial modelGPU-hour unit economics, utilisation scenarios, payback and IRR by GPU generation.
- Capacity planSites, power contracts, GPU allocation and delivery schedule.
- Commercial pipelinePipeline, letters of intent and design-partner status.
- PlatformArchitecture, security model, integration status and live console demo.
- Capital planEquity raise, GPU-backed debt and vendor financing structure.