Dual RTX PRO 6000 Blackwell workstation, 192GB of VRAM.
Two RTX PRO 6000 Blackwell Workstation Edition cards provide 192GB of GDDR7 and 48,128 CUDA cores. We build the pair on two platforms: the JNCS AI PRO on Threadripper 9000, and the JNCS AI MAX on Threadripper PRO 9000, which has the slots and power cables for four GPUs.
- 192GBGDDR7, 2× 96GB
- 48,128CUDA cores
- ~1,650Westimated full-load draw
- 3000Wpower supply
Two platforms for the pair
Both run each GPU on a dedicated PCIe 5.0 x16 slot.
| JNCS AI PRO | JNCS AI MAX | |
|---|---|---|
| Processor | Threadripper 9000: 24, 32, or 64 cores | Threadripper PRO 9000: 16 to 96 cores |
| PCIe 5.0 lanes | 80 | 128 |
| Memory | 64GB to 512GB registered ECC DDR5, four channels | 128GB to 1TB registered ECC DDR5, eight channels |
| Motherboard | ASUS Pro WS TRX50-SAGE WIFI | ASUS Pro WS WRX90E-SAGE SE |
| NVMe drives | Up to two | Up to four |
| Case | Fractal Design Define 7 | SilverStone WS380-E, eight hot-swap 3.5-inch bays |
| GPU count offered | 1 or 2 | 1, 2, or 4 |
What 192GB of VRAM runs
Total VRAM sets the largest model that can be loaded.
Model weights occupy roughly parameters × bits per weight ÷ 8. A 70-billion-parameter model at 4-bit quantization is about 35GB of weights; with context and runtime overhead it requires approximately 48GB of VRAM. Context length and batch size add to every figure below.
A 70B model at 16-bit
A 70B model at 16-bit precision is about 140GB of weights and fits across the two cards with no quantization.
Larger quantized models
At 4-bit quantization, models of roughly 200 to 300 billion parameters fit, depending on context length.
Several users at once
Each active conversation needs its own context memory on the GPU. The headroom above the weights decides how many users, and how long a context, a model can serve at once.
LoRA on a 70B model
LoRA fine-tuning on a 16-bit 70B base needs about 140GB for the frozen weights, which fits across the pair.
Fine-tuning guide →Power and circuit requirements
Two RTX PRO 6000 cards with a 350W Threadripper or Threadripper PRO processor draw approximately 1,650W at full load. Both builds use the SilverStone HELA 3000Rz, a 3000W 80 Plus Platinum supply with a 12V-2x6 cable for each GPU, for this configuration. The system requires a dedicated 20A, 110–120V circuit, and a 30A, 220–240V circuit is recommended. A standard 15A, 120V circuit supplies 1,440W continuously, which is not enough.
One RTX PRO 6000, or two RTX PRO 4500 or 5000 cards, draws 650W to 1,050W and runs on a standard circuit.
Assembly, burn-in, and delivery
Each system is assembled by hand at our shop in Fairport, NY, by a shop that has built computers for more than thirty years.
Fit and power validated
The configurator checks physical fit, memory, PCIe slots, and power on every selection. The power supply is sized automatically to the GPU count and estimated draw, with a 12V-2x6 cable for every GPU.
Burn-in under load
Every system is stress-tested under sustained full load and firmware-updated before it leaves the bench, so the first long job is not the first test.
Windows 11 Pro installed
Windows 11 Pro is installed and updated before shipment.
Warranty and support
Every build carries a 3-year warranty and lifetime technical support from the shop that built it.
Quoting and payment
The configurator shows a price for every part and an estimated total, including assembly, burn-in, and testing, as the build is configured. GPU and memory pricing changes frequently, so a technician confirms GPU availability and final pricing by email, with lead time, typically within 1-2 business days. There is no cost and no obligation to request a quote.
Payment for AI workstations is by ACH bank transfer; credit cards are not accepted for these systems. An invoice with bank details is issued once the quote is approved. Systems ship across the United States from Fairport, NY, with local pickup available.
Dual RTX PRO 6000 questions
How much VRAM do two RTX PRO 6000 cards have?
What power does a dual RTX PRO 6000 workstation need?
AI PRO or AI MAX for two RTX PRO 6000 cards?
Can a 70B model run unquantized on two RTX PRO 6000 cards?
Is two RTX PRO 6000 cards better than four RTX PRO 5000 cards?
How is an AI workstation ordered and paid for?
Related AI workstation pages
RTX PRO 6000 Blackwell workstation
One or two RTX PRO Blackwell GPUs, up to 192GB VRAM, Threadripper 9000.
Read →Multi-GPU AI workstation
Up to four RTX PRO Blackwell GPUs, up to 384GB VRAM, Threadripper PRO 9000.
Read →Local LLM workstation guide
How much VRAM a model needs, and which build provides it.
Read →LLM fine-tuning workstation
VRAM for full fine-tuning, LoRA, and QLoRA, by model size.
Read →Configure a dual RTX PRO 6000 workstation.
Select the RTX PRO 6000 and two GPUs in the AI PRO or AI MAX. The configurator sizes the power supply and shows the circuit requirement.
Have an exact parts list or a department spec sheet? Send the whole list. For more than four GPUs or rackmount systems, see AI servers and larger systems.