Buy a workstation with 2× NVIDIA RTX PRO 6000 Blackwell Max-Q in Europe

Who the two-card Max-Q workstation is for

You can buy a workstation with two NVIDIA RTX PRO 6000 Blackwell Max-Q cards from us, built to order and delivered anywhere in the EU. It gives a team 192 GB of GPU memory under a desk or in a small technical room: enough for Llama 3.3 70B in FP8 for dozens of colleagues at once, for gpt-oss-120b as two independent copies, or for a model of up to about 150 GiB of weights split over both cards. Our team sizes the configuration to your workload and budget and replies within one business day.

Companies buy this configuration for a few recurring jobs. An ML team uses it as a shared development machine for fine-tuning experiments and evaluation runs. A department runs its first private assistant or a RAG search over its own documents before deciding on a server. A research or engineering lab keeps data on site and splits the cards with MIG into isolated 24 GB GPUs for several people. A service that runs continuously for several departments or for customers belongs on a rack server instead; the last section compares the two.

The Max-Q is the edition for this: NVIDIA's 300 W RTX PRO 6000 with a blower cooler, which its product page says "enables up to four GPUs in a single system". The 600 W Workstation Edition has the same 96 GB and 1,792 GB/s, but its coolers release the heat inside the case, and our product page advises one card comfortably, two at most. Our comparison of the three editions shows the differences.

What the configuration contains

The table shows how we configure a two-card Max-Q workstation and what we change per project.

Our AI servers page (platforms, components, installation); memory and cores by NVIDIA's configuration guide for NVIDIA-Certified Systems (30 September 2026), written for servers and applied here by the same arithmetic.

NVIDIA's configuration guide covers servers only, and asks for at least six physical cores per GPU and system memory of at least twice the total GPU memory. We apply the same rule to workstations: 12 cores and 384 GB for two cards. In practice the processor is chosen for the work around the GPUs. A 32-core Threadripper PRO 9975WX, with a default TDP of 350 W, 128 PCIe 5.0 lanes and eight DDR5 channels, leaves room for data loading, tokenisation and a vector database next to the models.

Processor, PCIe lanes and slots

Two cards at full width take 32 PCIe lanes, four take 64. Both platforms we build on supply that: AMD lists the Threadripper PRO 9975WX with 128 PCIe 5.0 lanes, and Intel lists its Xeon 698X workstation processor with 128 PCIe 5.0 lanes and eight memory channels. How the board wires its slots matters just as much. In a review of 23 June 2026, StorageReview ran two Max-Q cards in an HP Z8 Fury G6i, which wires every GPU slot at PCIe 5.0 x16, and in a Dell Precision 7875 whose second card ran at PCIe 4.0 x16; serving Llama 3.1 8B in FP8 with vLLM at batch 256, the HP reached 23,004 tokens per second and the Dell 16,833. Both cards belong in x16 slots on the same processor, with a free slot between them.

The free slot follows Puget Systems' advice for the Max-Q: the memory on the back of the board needs air. It also keeps the two cards on a short peer-to-peer path, which a model split over both cards uses after every layer. The Linux settings that decide whether that path works, IOMMU and ACS, are worth checking in the acceptance test; our guide to RTX PRO 6000 P2P on Linux explains them.

Built to order, or cards for a tower you run

There are two ways to buy the configuration from us.

• Built to order. We assemble the workstation on a Threadripper PRO or Xeon W platform, test it before delivery and install the software on request.
• Cards for your tower. If your company standardises on a maker's workstation, we supply the Max-Q cards for it. The makers list the Max-Q for these towers in their own documents: Lenovo's ThinkStation P5 Gen 2 with up to two cards, and HP's Z8 Fury G6i, Lenovo's ThinkStation PX and Dell's Precision 7960 with up to four. We check the power supply, cables and slots of your model before we quote.

HP Z8 Fury G6i specifications page, Lenovo PSREF specifications for the ThinkStation P5 Gen 2 (2 September 2026) and ThinkStation PX (9 October 2026), Dell Precision 7960 configurator; all read on 24 October 2026.

We supply the RTX PRO 6000 Max-Q in both ways; send us the tower model or the platform you prefer, and our team replies with a configuration and a written quote.

What it runs

By our users-per-card rule, 0.9 × 95.6 GiB − 3 GiB per card, the two cards hold 166.1 GiB of weights and KV cache. The table gives four examples; our guide to what a workstation with two Max-Q runs has the full model list, both layouts, published results and the arithmetic.

Our estimates by the users-per-card rule: concurrent conversations at full context length, FP8 KV cache for Llama 3.3 70B, 16-bit for gpt-oss-120b.

These are requests in flight at the same moment; the logs of your gateway show the peak to size against. With MIG, each card can instead be split into up to four isolated 24 GB GPUs, eight per workstation, on Linux with driver 575.51.03 or later and the card in compute mode, which turns its display outputs off. A MIG workstation therefore runs without a monitor or with a small extra card for the screen, as our RTX PRO 6000 MIG profiles explain.

Power, heat and noise on 230 V

Two cards at 300 W and a 350 W processor make 950 W. With 15 per cent for memory, drives, fans and power-supply losses, the workstation draws about 1.1 kW at the wall under full load by our estimate.

Our estimate from NVIDIA's 300 W board power, AMD's 350 W default TDP for the Threadripper PRO 9975WX and Intel's 350 W base power for the Xeon 698X, plus 15 per cent; 1 W = 3.412 BTU/h.

• Circuit. An ordinary 230 V 16 A office circuit carries the workstation; check what else shares it. Our article on Max-Q power on 230 V covers circuits for two, four and eight cards.
• Cord. At 1.1 kW the load stays well inside the 10 A rating of a C13 appliance cord, which is 2.3 kW at 230 V. The large supplies of four-card towers use C19 cords; StorageReview's review of the Dell Precision 7960 lists a C19 cord for its 2,200 W unit, rated at 181 to 240 V.
• Heat. About 1.1 kW of heat, for as long as jobs run, noticeably warms a closed office over a day.
• Noise. No maker publishes acoustic figures for a Max-Q tower under GPU load. Put the rear of the workstation, where the blowers exhaust, towards open space; our guide to a GPU server in an office covers rooms shared with people.

Software delivered with the workstation

On request the workstation arrives with the operating system, NVIDIA driver, CUDA and a container runtime installed, so the team can start a serving engine or a notebook on the first day.

Our AI servers page; MIG prerequisites from NVIDIA's MIG user guide as cited in our MIG profiles article.

As of October 2026, R580 is NVIDIA's long-term branch with support to June 2028, and NVIDIA's security bulletin of 30 September 2026 names 580.178.04 as the first fixed R580 driver; our article on NVIDIA driver branches and CUDA versions explains the choice. Open-source serving engines on bare metal need no NVIDIA licence. The Max-Q is not on NVIDIA's vGPU list; for virtual machines with vGPU, a server with the Server Edition is the right platform, as our page on the Max-Q in virtual machines explains.

Ready for four cards from day one

A team can start with two cards and add two when a larger model or more users arrive. Four Max-Q give 384 GB, which holds models such as Qwen3-235B-A22B-2507 in FP8 that two cards cannot. The step is simple if four cards were planned at the start:

1. A tower or board with four x16 slots at double-slot spacing, and a processor with at least 64 PCIe lanes for the cards.
2. A power supply for about 1.8 to 2 kW at the wall. HP's dual supply for the Z8 Fury G6i has to be configured at purchase, according to HP's specifications page.
3. Memory planned for 768 GB, with modules that let the channels be filled evenly later.
4. Room cooling for about 2 kW of heat.

Our page to buy a 4× RTX PRO 6000 Max-Q workstation covers the four-card build.

Workstation or rack server with Server Edition cards

A rack server is the better buy when the service runs continuously for several departments or for customers, when nobody should sit next to the machine, or when virtual machines need vGPU. Our entry server has two RTX PRO 6000 Server Edition cards, two AMD EPYC processors, 384 GB of DDR5 ECC and two NVMe drives, in a chassis with redundant power supplies and out-of-band management. The Server Edition is passive, runs at up to 600 W and supports vGPU from release 19.0; its 1,597 GB/s are about 11 per cent below the Max-Q's 1,792 GB/s for decoding. The configuration is on our page to buy a 2× RTX PRO 6000 Server Edition server, and AI workstation vs AI server compares the form factors.

How ordering works with us

The order runs in the same steps for one workstation or several for different teams.

1. You describe the workload: models, users, data, the room and the timeline. Sizing is free of charge.
2. We propose the configuration and check that cards, platform, power supply and cooling fit the room and the circuit.
3. You receive a written quote within one business day, with price, lead time and warranty terms per component. Quantities and lead time are confirmed in writing with the quote.
4. You order on one EU contract with an EU invoice from Eurokommerz in Vienna.
5. We assemble the workstation and test it before delivery (power-on and basic load test).
6. We deliver anywhere in the EU, with commissioning on site on request; other destinations on request through our logistics partner.
7. Every component carries the manufacturer warranty, handled through us; DOA units are replaced, and the assembly is covered by us.

What to send us for a quote

A short description is enough to start; these points let us quote the exact configuration in the first reply.

• Models you plan to run, with quantisation and context length
• How many people will use the machine, and whether they share it through one service or through MIG instances
• Other work on the same machine: fine-tuning, embedding, data preparation, graphics
• Operating system and, if you have one, the tower model your company standardises on
• The room: circuit rating, cooling, and where the machine will stand
• Whether you plan to go to four cards, and when
• Delivery site, invoicing entity and timeline

Describe these points in the form below, and our team replies with a configuration and a written quote within one business day.

What we supply

We supply the RTX PRO 6000 Blackwell Max-Q as cards for towers whose makers list them, and build two-card and four-card Max-Q workstations to order, together with Server Edition rack servers for continuous operation. Everything comes on one EU contract and invoice with manufacturer warranty, handled through us; our guide to buying RTX PRO 6000 in the EU covers editions and warranty.

FAQ