Find the right GPU server.

1. Reference system 01

   Team 32

   Workstation

   A selected-supplier RTX 5090 tower for controlled local chat, RAG, coding, speech, image and batch evaluation when the accepted workload fits one 32GB GPU.

   GPU configuration

   1 × NVIDIA GeForce RTX 5090 32GB

   GPU count

   1

   Per GPU

   32GB VRAM

   GPU memory

   32GB one GPU

   System memory

   64GB

   Primary storage

   2TB NVMe

   32GB is available on one GPU as one physical memory space.

   Power planning

   Exact PSU, connector and measured wall-power requirement pending

   Representative fit

   One team beginning a controlled local-AI deployment

   Check before selection

   32GB is one GPU's physical memory; context, cache, batching and concurrency reduce the usable model envelope.

2. Reference system 02

   Company 64

   Workstation

   A supplier-validated dual-GPU workstation for teams needing two local workers or qualified PCIe sharding, with no NVLink or pooled-memory claim.

   GPU configuration

   2 × NVIDIA GeForce RTX 5090 32GB

   GPU count

   2

   Per GPU

   32GB VRAM

   Physical total

   64GB not automatically pooled

   System memory

   128GB

   Primary storage

   2TB NVMe

   64GB is the physical total across 2 separate 32GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   2050W PSU; measured wall power and site circuit requirement pending

   Representative fit

   Two independent private AI or creative queues

   Check before selection

   64GB is aggregate capacity across two 32GB GPUs, not one universal memory pool.

3. Reference system 03

   Studio 96

   Workstation

   A selected-supplier RTX PRO 6000 Blackwell workstation for named models and professional workloads that need substantially more than 32GB on one GPU.

   GPU configuration

   1 × NVIDIA RTX PRO 6000 Blackwell 96GB

   GPU count

   1

   Per GPU

   96GB VRAM

   GPU memory

   96GB one GPU

   System memory

   128GB

   Primary storage

   2TB NVMe

   96GB is available on one GPU as one physical memory space.

   Power planning

   Exact GPU edition, PSU and measured wall power pending

   Representative fit

   A named model or creative workload that benefits from 96GB on one GPU

   Check before selection

   The exact RTX PRO 6000 Workstation or Max-Q selection remains a pre-quote decision.

4. Reference system 04

   Studio 192

   Workstation

   A supplier-validated dual RTX PRO tower for studios and technical teams, explicitly treated as 192GB aggregate capacity rather than one pooled memory space.

   GPU configuration

   2 × NVIDIA RTX PRO 6000 Blackwell 96GB

   GPU count

   2

   Per GPU

   96GB VRAM

   Physical total

   192GB not automatically pooled

   System memory

   128GB

   Primary storage

   2TB NVMe

   192GB is the physical total across 2 separate 96GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   Exact dual-GPU PSU, circuit and measured wall-power requirement pending

   Representative fit

   Two independent high-memory professional GPU services

   Check before selection

   192GB is aggregate capacity across two 96GB GPUs, not one universal memory pool.

* Aggregate VRAM is a hardware total, not a promise that every workload can use it as one memory pool. Confirm the exact model, runtime and parallelisation route.

1. Reference system 01

   Value Rack 128

   Rack server

   Four PCIe Gen5 GPU workers for batch, private inference and service queues. The written quotation confirms the final specification and GPU warranty.

   GPU configuration

   4 × NVIDIA RTX 5090 AI 32GB

   GPU count

   4

   Per GPU

   32GB VRAM

   Physical total

   128GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   128GB is the physical total across 4 separate 32GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   Exact four-GPU input, PSU population and circuit requirement pending

   Representative fit

   Four independent GPU services or queues

   Check before selection

   128GB is aggregate capacity across four 32GB GPUs.

2. Reference system 02

   Value Rack 256

   Rack server

   Eight PCIe Gen5 GPU workers for dense batch and service throughput. Selection depends on sustained utilisation, final topology, GPU warranty and facility acceptance.

   GPU configuration

   8 × NVIDIA RTX 5090 AI 32GB

   GPU count

   8

   Per GPU

   32GB VRAM

   Physical total

   256GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   256GB is the physical total across 8 separate 32GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   High-capacity dedicated power design required; exact measured system input pending

   Representative fit

   Eight independent services or parallel batch queues

   Check before selection

   256GB is aggregate capacity across eight 32GB GPUs.

3. Reference system 03

   Enterprise 384

   Rack server

   Four RTX PRO 6000 Blackwell Server Edition GPU workers for controlled enterprise services. The written quotation confirms the platform, cooling and warranty.

   GPU configuration

   4 × NVIDIA RTX PRO 6000 Blackwell 96GB

   GPU count

   4

   Per GPU

   96GB VRAM

   Physical total

   384GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   384GB is the physical total across 4 separate 96GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   Exact PSU population, input requirement and measured wall power pending

   Representative fit

   Several 96GB private inference endpoints

   Check before selection

   384GB is aggregate capacity across four 96GB GPUs.

4. Reference system 04

   Enterprise 768

   Rack server

   Eight RTX PRO 6000 Blackwell Server Edition GPU workers for large private platforms where independent services or tested sharding justify the facility requirements.

   GPU configuration

   8 × NVIDIA RTX PRO 6000 Blackwell 96GB

   GPU count

   8

   Per GPU

   96GB VRAM

   Physical total

   768GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   768GB is the physical total across 8 separate 96GB GPUs. It is not automatically pooled; supported software may divide a model or workload across them, subject to topology and testing.

   Power planning

   Dedicated high-capacity power required; exact input and measured wall power pending

   Representative fit

   Eight high-memory endpoints or batch workers

   Check before selection

   768GB is aggregate capacity across eight 96GB GPUs.

5. Reference system 05

   H200 1.1TB

   Rack server

   A quote-only eight-GPU PCIe H200 NVL system for qualified large-model work. Kimi K3 and other large models require an exact runtime profile and acceptance test.

   GPU configuration

   8 × NVIDIA H200 NVL 141GB

   GPU count

   8

   Per GPU

   141GB VRAM

   Physical total

   1128GB not automatically pooled

   System memory

   1024GB

   Primary storage

   11.68TB NVMe

   1,128GB is the physical total across eight H200 GPUs. It is not automatically one pooled memory space; the exact NVLink pairing, PCIe topology and supported parallel runtime must be confirmed in the quotation.

   Power planning

   At least 4.8kW GPU nameplate; exact full-system input pending

   Representative fit

   A named large model with a reproduced eight-GPU profile

   Check before selection

   1,128GB aggregate HBM is below the current 1.5609TB official Kimi K3 weight-file total.

* Aggregate VRAM is a hardware total, not a promise that every workload can use it as one memory pool. Confirm the exact model, runtime and parallelisation route.

1. Reference system 01

   Frontier Native 2.3TB

   Rack server

   A quote-only HGX B300 route using eight 288GB SXM GPUs and NVLink/NVSwitch architecture, subject to exact OEM configuration, software support and facility design.

   GPU configuration

   8 × NVIDIA B300 288GB

   GPU count

   8

   Per GPU

   288GB VRAM

   Physical total

   2304GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   2,304GB is distributed across eight B300 GPUs connected by HGX NVLink and NVSwitch. The fabric enables high-bandwidth multi-GPU work, but software support and the exact workload still determine usable capacity.

   Power planning

   Approximately 14kW system envelope; exact quoted input pending

   Representative fit

   A named model with a validated HGX B300 runtime

   Check before selection

   2.304TB HBM is a calculated capacity envelope, not proof that the official Kimi K3 checkpoint loads or serves one-million-token context.

2. Reference system 02

   Frontier Rack 20TB

   Rack server

   A quote-only GB300 NVL72 route with 72 Blackwell GPUs and 36 Grace CPUs. It is a complete rack-scale platform, not an ordinary GPU server.

   GPU configuration

   72 × NVIDIA GB300 288GB

   GPU count

   72

   Per GPU

   288GB VRAM

   Physical total

   20736GB not automatically pooled

   System memory

   Confirmed in quotation Exact system-memory capacity and population pending supplier BOM

   Primary storage

   Confirmed in quotation Exact NVMe capacity, model, endurance and layout pending supplier BOM

   20,736GB is distributed across a rack-scale NVLink and NVSwitch fabric. It is specialist shared infrastructure, not one ordinary GPU memory space.

   Power planning

   Up to 142kW full-rack power

   Representative fit

   A validated rack-scale model or training architecture

   Check before selection

   This is a complete 72-GPU liquid-cooled rack, not an the selected OEM 4U product or ordinary office delivery.

* Aggregate VRAM is a hardware total, not a promise that every workload can use it as one memory pool. Confirm the exact model, runtime and parallelisation route.