State-of-the-art GPUs
Zai Node's AI compute uses the best GPUs from NVIDIA along with the most efficient components on the market.
NVIDIA RTX Pro 6000 Blackwell Server Edition
Delivers a powerful combination of AI and visual computing capabilities to accelerate enterprise volume workloads. Unparalleled performance and flexibility for agentic AI, physical AI, and scientific computing to rendering, 3D graphics, and video.
96GB ultra fast memory
Air and liquid cooled
5th gen tensor cores
4th gen RT cores
NVIDIA CUDA cores
NVIDIA RTX Pro 6000 Blackwell Server Edition
Delivers a powerful combination of AI and visual computing capabilities to accelerate enterprise volume workloads. Unparalleled performance and flexibility for agentic AI, physical AI, and scientific computing to rendering, 3D graphics, and video.
96GB ultra fast memory
Air and liquid cooled
5th gen tensor cores
4th gen RT cores
NVIDIA CUDA cores
NVIDIA RTX 5090 Blackwell
Bringing game-changing capabilities to tackle the most advanced models and most challenging creative workloads with unprecedented AI horsepower.
32GB ultra fast memory
Air cooled
5th gen tensor cores
High speed image and video
NVIDIA RTX 5090 Blackwell
Bringing game-changing capabilities to tackle the most advanced models and most challenging creative workloads with unprecedented AI horsepower.
32GB ultra fast memory
Air cooled
5th gen tensor cores
High speed image and video
NVIDIA DGX Spark, Grace Blackwell Superchip
Complete platform for local autonomous agents. Large local memory, powerful AI compute and the NVIDIA AI software stack enable agents and large models to run locally.
GB10 Superchip
128GB memory
ConnectX networking
NVIDIA software
NVIDIA DGX Spark, Grace Blackwell Superchip
Complete platform for local autonomous agents. Large local memory, powerful AI compute and the NVIDIA AI software stack enable agents and large models to run locally.
GB10 Superchip
128GB memory
ConnectX networking
NVIDIA software
Low power use
The most energy efficient servers with advanced electrical design, intelligent distribution and redundant power systems.
Low power use
The most energy efficient servers with advanced electrical design, intelligent distribution and redundant power systems.
Novel cooling
Closed cooling loops eliminate evaporation and drastically reduce water demand.
Novel cooling
Closed cooling loops eliminate evaporation and drastically reduce water demand.
Heat reuse
100% of GPU heat offtake is reused on-prem.
Heat reuse
100% of GPU heat offtake is reused on-prem.
Networking
Segregated network with hardware, hypervisor and cluster management, inference serving and training networks with extreme E-W bandwidth for frontier training.
Networking
Segregated network with hardware, hypervisor and cluster management, inference serving and training networks with extreme E-W bandwidth for frontier training.
Compute / storage
Dense GPU servers with NVLink and PCIe Gen5. CPUs with high-core-count for orchestration, preprocessing and hybrid apps. Hot NVMe storage.
Compute / storage
Dense GPU servers with NVLink and PCIe Gen5. CPUs with high-core-count for orchestration, preprocessing and hybrid apps. Hot NVMe storage.
Cross connects
Fast and secure cross connects, ideal for connecting your AI to other devices, networks or DCs.
Cross connects
Fast and secure cross connects, ideal for connecting your AI to other devices, networks or DCs.
Manifest & attestation
Fully certified documentation, ensuring security protocols are met.
Manifest & attestation
Fully certified documentation, ensuring security protocols are met.
3rd party inspections
Our infra is open for physical and digital third party verifications.
3rd party inspections
Our infra is open for physical and digital third party verifications.
Origin certificate
Our servers are accompanied by a Certificate of Australian Origin.
Origin certificate
Our servers are accompanied by a Certificate of Australian Origin.
State-of-the-art AI compute and managed AI models
Scale your compute and models as fast as your ideas with Zai Node. Focus on your AI agents and products, not managing AI infrastructure or trying, in vain, to work with hyperscalers.