Buy a workstation with 1 or 2 NVIDIA RTX PRO 6000 Blackwell Workstation Edition

Who a Workstation Edition PC is for

You can buy a workstation with one or two NVIDIA RTX PRO 6000 Blackwell Workstation Edition cards from us, built to order and delivered anywhere in the EU. The Workstation Edition is the full 600 W card with 96 GB of GDDR7 at 1,792 GB/s, which NVIDIA's family page describes as "engineered for maximum performance in single-GPU workstations", and the makers of current towers list one or at most two per machine. Our team sizes the configuration to your workload and budget and replies within one business day.

The configuration suits work where one person or a small team sits at the machine and wants the most from each card. An ML engineer runs fine-tuning experiments and evaluations on a model that needs the whole 96 GB. A design studio renders scenes too large for a 48 GB card. An engineering department runs simulation and a local language model next to its CAD work. A research lab keeps sensitive data on site. In each case the card also drives the monitors, unless MIG is switched on, which the software section explains. For three or four cards in one tower, or for a service that runs continuously for several departments, another edition fits better, and the last sections say which.

One card or two

The RTX PRO 6000 Blackwell comes in three editions with the same GPU and 96 GB per card. The table puts the Workstation Edition builds next to the alternatives a buyer usually weighs.

NVIDIA RTX PRO 6000 family page and Max-Q product page, read on 24 October 2026; tower counts from the makers' specifications in the next table.

One card is the easy case: a 600 W card in a tower whose maker lists it, with no neighbour to heat. Two cards double the memory to 192 GB, enough for a model of up to about 150 GiB of weights split over both cards or two independent copies of a mid-sized model. The second card is where the design gets harder, because the first card's exhaust feeds the second. Two Max-Q cards give the same 192 GB at half the card power, with blowers that push the air out through the back of the case. In a review of July 2025, Puget Systems measured the Max-Q 5 to 14 per cent behind the Workstation Edition in content-creation applications; for LLM token generation, which follows memory bandwidth, both editions have 1,792 GB/s.

Why makers stop at one or two cards per tower

NVIDIA's family page lists the Workstation Edition with a "double flow-through" cooler at 600 W and 5.4 × 12 inches, against the Max-Q's active cooler at 300 W and 4.4 × 10.5 inches. Puget Systems explains what the flow-through design means in a multi-card tower: air leaves through the top of the card, and "the lower card will push hot air into the intake of the upper card". Its parts page for the Workstation Edition calls the card "NOT suitable for most multi-GPU configurations" and allows two only in some specially designed systems. The makers' own specifications show the same limit.

Lenovo PSREF specifications for the ThinkStation P8 (15 September 2026) and PX (9 October 2026), HP Z8 Fury specifications page, Dell configurators for the Pro Precision 9 T6 and Precision 7960, all read on 24 October 2026; the T6 power supply from StorageReview's launch report of 16 March 2026.

Of the towers that take four Max-Q cards, only Dell's Pro Precision 9 T6 offers two Workstation Edition cards, and StorageReview gives it a 2,400 W supply; the others list one or none. HP's specifications page mentions an optional Max Side Panel for "600W graphics"; its current table for the Z8 Fury G6i names only the Max-Q, while the archived table for the earlier Z8 Fury G5 lists one Workstation Edition. Integrators who design their own systems reach the same limit. Puget Systems builds its dual-card RTX PRO 6000 Workstation systems in a 5U rackmount chassis with server-grade fans and 2,800 W supplies on 200 to 240 V circuits, for server rooms (October 2025). VRLA Tech tested four 600 W cards in towers in September 2026, measured 89 to 90 °C on the hottest card with the side panels on, and does not ship that configuration. Our own product page puts it as one card comfortably, two at most. The detail of two cards side by side is in our article on two Workstation Edition cards: heat and power.

We build one-card and two-card Workstation Edition workstations to order and supply the cards for towers whose makers list them; send us the tower model or the workload, and our team replies with a configuration and a written quote.

What the configuration contains

The table shows how we configure the two builds 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 asks servers 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, so 192 GB for one card and 384 GB for two. In practice the processor is chosen for the work around the GPU. AMD lists the Threadripper PRO 9975WX with 32 cores, a default TDP of 350 W, 128 PCIe 5.0 lanes and eight DDR5 channels, which leaves room for data loading, tokenisation and a vector database. Each card belongs in a slot wired at PCIe 5.0 x16 on the processor. In a review of 23 June 2026, StorageReview served Llama 3.1 8B in FP8 with vLLM on two Max-Q cards and reached 23,004 tokens per second in a tower with every GPU slot at Gen5 x16, against 16,833 in one whose second card ran at Gen4 x16.

Each Workstation Edition has one 12V-2×6 power connector, according to Puget's parts page, so the power supply needs one native 600 W cable per card.

What it runs

By our users-per-card rule, 0.9 × 95.6 GiB − 3 GiB per card, one card holds 83.0 GiB of weights and KV cache and two cards 166.1 GiB. The memory is the same as on two Max-Q cards, so our guide to what a workstation with two Max-Q runs applies model by model.

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 the others.

These are requests in flight at the same moment, not the number of staff who use the service. For a single user, the ceiling on generation speed is the bandwidth divided by the bytes read per token: 1,792 GB/s over 70.6 GB gives about 25 tokens per second for Llama 3.3 70B in FP8 on one card, by our arithmetic. The cards have no NVLink, so a model split over two cards exchanges data over PCIe, as our article on RTX PRO 6000 without NVLink explains; our guide to how many users one RTX PRO 6000 serves shows the arithmetic for one card.

Power and room on 230 V

Our estimate from NVIDIA's 600 W board power and AMD's 350 W default TDP for the Threadripper PRO 9975WX, plus 15 per cent for memory, drives, fans and power-supply losses; 1 W = 3.412 BTU/h.

We plan a 230 V 16 A circuit at 80 per cent, 2.94 kW. One card leaves most of that for other equipment; a two-card tower at full load takes 1.8 kW of it, so check what else shares the circuit. Workstation power supplies give their full rating only at 200 V or more: according to StorageReview, Dell's 2,400 W unit for the Pro Precision 9 T6 delivers 1,500 W at 115 to 199 V, and HP rates its Z8 Fury supply at 1,700 W from 200 V and 1,350 W at 115 V. On a European 230 V socket the full rating is available.

A power limit lowers the figures. At 450 W per card, set with nvidia-smi, a two-card tower comes to about 1.4 kW at the wall by the same estimate; StorageReview measured the cost of a 450 W cap on one Workstation Edition at 2 to 5 per cent of throughput at short context and 4 to 9 per cent at long context (October 2026). Our guide to the RTX PRO 6000 power limit shows the commands and how to keep the limit across reboots.

All of the power becomes heat in the room. For one card, GamersNexus measured about 32.5 dBA in September 2025, as our comparison of the three editions reports; no maker publishes an acoustic figure for a two-card tower under GPU load. 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 notebook, a renderer or a serving engine on the first day.

Our AI servers page; MIG prerequisites from NVIDIA's MIG documentation as cited in our MIG profiles and edition comparison articles.

MIG splits each card into up to four isolated 24 GB GPUs. On the Workstation Edition it needs Linux, driver 575.51.03 or later, a vBIOS of 98.02.55.00.00 or later and the card switched to compute mode with NVIDIA's DisplayModeSelector, which turns its display outputs off; our RTX PRO 6000 MIG profiles explain the steps. 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. The Workstation Edition is not on NVIDIA's vGPU list; virtual machines with vGPU need the Server Edition in a server.

When a Max-Q workstation or a server is the better buy

• Three or four cards. The Max-Q is the edition for that. NVIDIA states up to four per workstation, and HP, Lenovo and Dell list four for their large towers; our page to buy a 4× RTX PRO 6000 Max-Q workstation covers the build.
• Two cards in an office. Two Max-Q give the same 192 GB at 600 W of card power and about 1.1 kW at the wall by our estimate, and leave the tower ready for four; see our page to buy a 2× RTX PRO 6000 Max-Q workstation.
• A shared service. When the models run continuously for several departments or for customers, need redundant power and remote management, or must run virtual machines with vGPU, a rack server is the platform. Our entry server has two RTX PRO 6000 Server Edition cards, two AMD EPYC processors, 384 GB of DDR5 ECC and two NVMe drives; its 1,597 GB/s give roughly 11 per cent fewer tokens per second on generation than the Workstation Edition. The configuration is on our page to buy a 2× RTX PRO 6000 Server Edition server.
• The Server Edition in a tower. It is a passive card that depends on ducted server airflow, which a tower does not provide; our article on the Server Edition in a workstation explains why we do not recommend it.

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: applications or models, users, data, the room and the timeline. Sizing is free of charge.
2. We propose the configuration and check that card, 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.

If your company standardises on a maker's tower, we supply the card for it and check the power supply, cables and slots of your model first. The checks in our GPU server acceptance testing guide also serve for your own acceptance test after delivery.

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.

• The applications or models, with precision and context length for language models
• One card or two, and whether three or four may follow later
• How many people use the machine, and whether through one service or MIG instances
• 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
• 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 Workstation Edition as cards for towers whose makers list them and build one-card and two-card workstations with it to order, together with Max-Q workstations for three or four cards and Server Edition rack servers for continuous operation. Our guide to buying RTX PRO 6000 in the EU covers the editions and warranty.

FAQ