ABS AI Workstation - 192GB Aggregate VRAM (2x 96GB RTX PRO 6000 Blackwell Max-Q), AMD Ryzen Threadripper PRO 7975WX, 128GB ECC DDR5, 2TB M.2 + 3.84TB SATA SSD, Ubuntu (Zaurion ZRP7975WX-RP60002M)
- Compliance Documents Required for Purchase
- Assembled in USA
- TAA Compliant
- Operating System: Ubuntu
- CPU: AMD Ryzen Threadripper PRO 7975WX
- GPU: 2x RTX Pro 6000 Blackwell MaxQ (96 GB GDDR7)
- Motherboard: Gigabyte MH53-G40
- Memory: 128GB DDR5 Server RAM - ECC, Registered
- OS SSD: 2TB M.2 NVMe (Installed)
- Data SSD: 3.84TB SSD (Installed)
- LAN: 2 x 10GB/s LAN, 1 x 10/100/1000 Mbps Management LAN
- PSU: Single 2000W ATX 80 PLUS Gold power supply
- Barebone: Gigabyte W773-H5D-AA01
- AI Agent Ready: Optional pre-installed OpenClaw (enterprise AI agent framework)
- Local LLM Support: Optional Qwen 3.5 27B or higher models for on-prem AI inference
Any questions? Our AI mode will help you find out quickly.
Built for Two.
Same server-pedigree sTR5/WRX90 CPU platform, same 128 GB of DDR5 and 5.84 TB of installed SSD storage as this chassis lineup's single-GPU configuration — but with a second GPU. Two independent NVIDIA RTX PRO 6000 Blackwell Max-Q cards give you 192 GB of aggregate VRAM for multi-GPU training, distributed rendering, and large-scale data modeling.
What This Pairing Actually Buys You
A server-pedigree CPU platform feeding two independent GPUs, built for workloads that need more than one card:
192 GB Aggregate VRAM, Two Ways to Use It
- Two independent 96 GB GPUs — 192 GB aggregate, not a single unified pool. Split a model across both, or run each independently.
- Max-Q's 300W-per-card design1 trades peak single-GPU performance for the efficiency and density that make a two-GPU tower possible on one 2000 W supply.
- Up to four isolated MIG instances per GPU1 — concurrent workloads or users, each with dedicated resources and guaranteed quality of service.
- Blackwell 5th-gen Tensor Cores with FP4 accelerate agentic and generative AI on each card at architecture-native speed.
One Socket, Two GPUs Fed
- sTR5 socket on the WRX90 chipset2 — a server-derived workstation platform with the PCIe lanes to run two dual-slot GPUs at full bandwidth.
- 6x PCIe 5.0 x16 slots mean both GPUs run at full Gen5 bandwidth, with lanes to spare for future expansion.
- 8-channel DDR5 memory keeps both GPUs supplied with data without the CPU becoming the bottleneck.
- ECC memory support and workstation-grade reliability bring server-class RAS features to long, multi-GPU training runs.
Two Ways to Scale
The listing's own use cases point to two different ways to put a second GPU to work:
Split One GPU, Many Ways (MIG)
Partition either GPU into up to four fully isolated instances, each with its own memory, cache, and compute — concurrent workloads or users on one card, with guaranteed quality of service.
Combine Both GPUs, One Big Job
Split a single large model or dataset across both GPUs for multi-GPU training, multi-instance inference, or distributed rendering — workloads too large for one 96 GB card alone.
One GPU or Two?
This chassis lineup offers the identical CPU, memory, and storage with either one or two GPUs — here's what the second GPU buys:
Same AMD Ryzen Threadripper PRO 7975WX, same memory, same storage, same Gigabyte MH53-G40 platform — the second GPU is the only variable. Compare options on this listing.
Two Slots Used, Four Available
This configuration populates 2 of the platform's 6 PCIe 5.0 x16 slots — there's room for more:
Front: 2× USB 3.2 Gen1 Type-A + 1× USB 3.2 Gen1 Type-C, power button, audio/mic jack, and 4× 3.5″/2.5″ SATA hot-swap bays installed (4 more optional — 8 total possible). Rear: dual 10Gb/s LAN (Broadcom BCM57416) plus a dedicated MLAN management port, USB and audio connectivity, and the 2000 W power inlet.
8× DIMM slots on 8-channel DDR5 (2 populated), 6× PCIe 5.0 x16 slots rated for up to 4× dual-slot GPUs (this config populates 2 of them), and 4× M.2 SSD slots with only one populated by the 2 TB OS drive.3 Two GPUs in, two more slots of headroom left.
Five Jobs, Two GPUs
The listing's own AI Features, translated into workloads that lean on multi-GPU scale:
AI Development
Accelerate AI development and inference workloads, and build agentic AI applications.
Data Science
Accelerate your data pipelines.
AI-Driven Rendering & Graphics
Deliver stunningly realistic designs and simulations with speed and precision.
HPC
Accelerate scientific breakthroughs with enhanced accuracy.
Video Content & Streaming
Boost performance for live media and video pipelines.
Scene photos are AI-generated illustrations of typical deployments, not photographs of this exact product.
Fine-Tune Even Larger Models with Phison aiDAPTIV+
An optional add-on that offloads GPU memory to flash, reducing the VRAM full fine-tuning actually requires:
| Target Model Size | VRAM Needed (Without aiDAPTIV+) | VRAM Needed (With aiDAPTIV+) |
|---|---|---|
| ~13B | 120–200 GB | 32 GB |
| ~34B | 250–350 GB | 96 GB — either of this workstation's GPUs |
| ~70B | 500–700 GB | 192 GB — this workstation's two GPUs, combined |
Phison aiDAPTIV+ SSD is an optional add-on and is not included by default. VRAM figures are vendor-published estimates for full fine-tuning.
As Configured
| ZRP7975WX-RP60002M | |
|---|---|
| CPU | AMD Ryzen Threadripper PRO 7975WX — sTR5 socket, WRX90 chipset2 |
| GPU | 2x NVIDIA RTX PRO 6000 Blackwell Max-Q — 96 GB GDDR7 each, 192 GB aggregate1 |
| Memory | 128 GB DDR5, ECC Registered, 8-channel |
| Storage | 2 TB M.2 NVMe (OS) + 3.84 TB SSD (Data) |
| Networking | 2× 10 Gb/s LAN (Broadcom BCM57416) + MLAN management port |
| Motherboard | Gigabyte MH53-G40 (Gigabyte W773-H5D-AA01 barebone) |
| Power | Single 2000 W ATX 80 PLUS Gold |
| Expansion | 8× DIMM (8-channel DDR5), 6× PCIe 5.0 x16 (2 populated), 4× M.2 SSD, 4× hot-swap bays installed |
| OS | Ubuntu |
| Optional | Phison aiDAPTIV+ SSD; OpenClaw agent framework; Qwen 3.5 27B+ local LLM package |
Quick Answers Before You Buy
What's different between the Max-Q GPUs here and the standard RTX PRO 6000 Blackwell on other configs?
Max-Q is a lower-power, higher-density version of the same GPU family — about 300 W per card. Its own published figures (3511 TOPS AI performance, 333 TFLOPS RT, 110 TFLOPS single-precision) are lower per-GPU than the standard card, but this configuration runs two of them.
Is the 192 GB of VRAM one big pool or two separate 96 GB GPUs?
Two independent 96 GB GPUs — 192 GB aggregate, not a single unified pool. Split a model across both, or run each independently, including MIG-partitioning either card.
What's different between this config and the single-GPU tier?
The CPU, 128 GB DDR5 memory, and 2 TB+3.84 TB storage are identical to this chassis lineup's single-GPU tier. This configuration adds a second Max-Q GPU for 192 GB aggregate VRAM instead of 96 GB.
How much room is there to add more GPUs?
6× PCIe 5.0 x16 slots rated for up to 4× dual-slot GPUs; this configuration populates 2 of them.
Does it come with an AI agent framework or local LLM pre-installed?
Not by default. OpenClaw and a Qwen 3.5 27B+ local LLM package are optional add-ons configured at order time.
Is it eligible for government / public-sector purchasing?
It is TAA compliant and assembled in the USA, with compliance documents available for purchase review.
Warranty & Returns
Warranty, Returns, And Additional Information
Warranty
- Limited Warranty period (parts): 1 year
- Limited Warranty period (labor): 1 year
- Read full details
Return Policies
- Contact Manufacturer for Assistance
- This item is covered by Newegg.com's Manufacturer Only Return Policy.
- Read full return policy for details.
Manufacturer Contact Info
- Manufacturer Product Page
- Manufacturer Website
- Support Phone: 1-800-685-3471
- Support Email: [email protected]
- Support Website
- View other products from ABS














