NVIDIA brings Blackwell acceleration to compact professional workstations

NVIDIA´s new RTX PRO 4000 Blackwell SFF and RTX PRO 2000 deliver Blackwell architecture and Artificial Intelligence acceleration in compact, energy-efficient workstations for engineers, designers and creators.

NVIDIA is packing the company´s Blackwell architecture into smaller, lower-power professional graphics cards aimed at users who need workstation-class compute in tight form factors. The new NVIDIA RTX PRO 4000 Blackwell SFF Edition and NVIDIA RTX PRO 2000 Blackwell put fourth-generation rt cores and fifth-generation tensor cores into half the size of a traditional GPU while maintaining aggressive power limits. The design targets mainstream engineering, content creation, 3D visualization and edge inference, where space and thermal budgets often limit hardware choices.

Performance claims underline the generational jump. The RTX PRO 4000 SFF is rated for up to 2.5x higher artificial intelligence performance, 1.7x higher ray-tracing throughput and 1.5x more memory bandwidth compared with the prior generation, while holding the same 70-watt max power envelope. The RTX PRO 2000 is tuned for everyday professional workloads and offers up to 1.6x faster 3D modeling, 1.4x faster CAD, 1.6x quicker rendering, plus a 1.4x boost in image generation and a 2.3x leap in text generation for model fine-tuning and on-device inferencing. These figures are positioned to accelerate iteration cycles and shorten time to insight on single-station builds.

Real-world users across public works, geospatial teams, architecture studios and engineering firms are already testing the new cards. Organizations including a regional flood district, the government geospatial office of Cantabria, design studio Studio Tim Fu, engineering consultancy Thornton Tomasetti and product startup Glüxkind reported faster model runs, quicker fine-tuning and improved responsiveness for complex simulations and generative design tasks. Benchmarks cited range from near doubling of GIS and hydrologic workloads to multifold speedups for finite element analysis versus a CPU.

The ecosystem angle is central to the pitch. NVIDIA maps the new hardware to the NVIDIA AI Enterprise suite, NVIDIA Cosmos foundation models and omniverse visualization and simulation tools, plus CUDA-X libraries and thousands of supported applications. The Cosmos‑Reason1‑7B model is noted as runnable on the RTX PRO 4000 SFF for physical reasoning at the edge. Availability is planned later this year through distribution and system partners, including PNY, TD SYNNEX, BOXX, Dell, HP and Lenovo, with broad manufacturer and channel support expected for the small form factor and full-length cards.

68

Impact Score

This AI won’t drain your battery

Google DeepMind´s Gemma 3 270M promises on-device Artificial Intelligence that uses almost no phone battery, while Sam Altman lays out OpenAI´s post-GPT-5 strategy.

###CFCACHE###

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.

Please check your email for a Verification Code sent to . Didn't get a code? Click here to resend