01PLANNED OFFERING
On-Demand Compute
RTX 5090 · 1 / 2 / 4 GPUs
For model experiments, fine-tuning and inference. Discuss the configuration and expected usage before provisioning.
32 GB per GPU · up to 128 GB aggregateExplore
ZaoSeq Compute
From RTX experiments to dedicated AI clusters. We help scope the GPU configuration, region and commercial requirements for your project.
Currently accepting requirements and partnership inquiries. Provisioning, availability, pricing and delivery remain subject to supplier confirmation. Self-service rental is not yet available.
02PLANNED OFFERING
Limited / Reserved Compute
8×5090 · 8×6000 Ada · RTX PRO 6000 · V100
Dedicated configurations and legacy workloads, scoped around your region, rental term and software requirements.
Configuration and term confirmed per projectExplore
03PLANNED OFFERING
AI Clusters
H100 / H200 / B200 / B300
Plan 8, 16, 32 or more GPUs with the interconnect, storage and deployment topology your workload needs.
8 / 16 / 32+ GPUs · project inquiryExplore
04PLANNED OFFERING
Privacy Compute
ZaoSeq Privacy Gateway → GPU Compute
A planned data preparation layer for de-identification and pseudonymization, with processing scoped and priced separately.
Add-on or standalone · separately quotedExplore
CONFIGURATION GUIDE
GPU configurations
| GPU | GPU count | Memory / GPU | Aggregate memory | Service direction |
|---|---|---|---|---|
| RTX 5090 | 1 / 2 / 4 | 32 GB | 32 / 64 / 128 GB | On-Demand |
| RTX 5090 | 8 | 32 GB | 256 GB | Reserved |
| RTX 6000 Ada | 8 | 48 GB ECC | 384 GB | Reserved |
| RTX PRO 6000 Blackwell | 1 / 2 / 4 | 96 GB ECC | 96 / 192 / 384 GB | Limited / Reserved |
| V100 Legacy | 1 / 2 / 4 / 8 | Variant to confirm | By configuration | Limited / Reserved |
Aggregate memory is the sum across GPUs, not a single unified memory pool. Multi-GPU use depends on software, topology and workload. RTX PRO 6000 edition and V100 variant must be confirmed with the provider.
Define the workload
Model, framework, precision, region, GPU count and expected duration.
Validate the configuration
Supplier availability, memory, networking, storage and software compatibility.
Agree the terms
A written quote will specify currency, taxes, per-GPU or per-node billing, minimum term and cancellation conditions. Storage, transfer and privacy processing are scoped separately.
LET’S BUILD THE CONNECTION
Talk to ZaoSeq