NVIDIA Launches Blackwell-Powered RTX PRO GPUs and Servers, Delivering Major AI and Graphics Performance Boosts for Enterprise Workloads

NVIDIA has unveiled a new generation of professional GPUs and servers powered by its Blackwell architecture, marking a significant leap in AI acceleration for enterprise and creative workflows. The NVIDIA RTX PRO 4000 Blackwell SFF Edition and RTX PRO 2000 Blackwell GPUs are designed for compact, energy-efficient workstations, delivering high performance in half the size of traditional GPUs with just 70 watts of power. These new models feature fourth-generation RT Cores and fifth-generation Tensor Cores, enabling up to 2.5x higher AI performance, 1.7x faster ray tracing, and 1.5x greater bandwidth compared to the previous generation—all while maintaining the same power envelope. The RTX PRO 2000 Blackwell offers up to 1.6x faster 3D modeling, 1.4x improved CAD performance, and 1.6x quicker rendering speeds. It also delivers a 1.4x boost in image generation and a 2.3x increase in text generation, accelerating design iteration and AI-driven workflows. Real-world applications are already demonstrating the impact: the Mile High Flood District uses the GPU for complex flood simulations and 3D visualizations; the Government of Cantabria’s geospatial office sees twice the speed in AI model fine-tuning; Studio Tim Fu leverages it for real-time text-to-3D urban design; and Thornton Tomasetti reports nearly 3x faster structural analysis using its in-house solver, outperforming CPUs by 27x. In the enterprise space, NVIDIA is expanding access to the RTX PRO 6000 Blackwell Server Edition, now available in 2U rack-mounted servers from global partners including Cisco, Dell Technologies, HPE, Lenovo, and Supermicro. These servers bring powerful GPU acceleration to mainstream data centers, enabling breakthrough performance across AI, data analytics, 3D visualization, and scientific simulation. With up to 45x better performance and 18x higher energy efficiency than CPU-only systems, they represent a major shift toward accelerated computing in on-premises environments. The new servers support NVIDIA’s AI Data Platform, a customizable reference design for building modern AI infrastructure. Dell, for example, announced the PowerEdge R7725 2U server with two RTX PRO 6000 GPUs, integrated with the AI Data Platform and NVIDIA AI Enterprise software. These systems are ideal for agentic AI, digital twins, robotics, and physical AI applications, with simulation and synthetic data generation running up to 4x faster than with previous-generation L40S GPUs. NVIDIA’s software ecosystem further enhances the platform’s capabilities. AI Enterprise provides enterprise-grade tools for building and deploying AI at scale, while the Cosmos platform delivers optimized foundation models like Cosmos-Reason1-7B for edge and workstation deployment. The Omniverse platform enables collaborative 3D design and simulation, and CUDA-X libraries support thousands of optimized applications across industries. The RTX PRO 2000 Blackwell and RTX PRO 4000 Blackwell SFF Edition are expected to be available later this year through PNY, TD SYNNEX, and system builders like BOXX, Dell, HP, and Lenovo. The 2U RTX PRO Server Editions will follow later in the year, with configurations supporting up to eight GPUs in 4U form factors already available. With these advancements, NVIDIA is accelerating the transition from traditional CPU-based computing to GPU-accelerated, AI-powered workloads—making high-performance AI accessible across industries, from engineering and architecture to healthcare and robotics.