NVIDIA unveils RTX PRO 4500 Blackwell server edition GPU

NVIDIA has introduced a passively cooled, single-slot RTX PRO 4500 Blackwell Server Edition aimed at compute-dense server deployments. The card closely matches the standard RTX PRO 4500 Blackwell while lowering power and memory speed to fit hyper-dense configurations.

NVIDIA has introduced the passively cooled, single-slot RTX PRO 4500 Blackwell Server Edition for compute-dense server deployments. It comes with 10,496 CUDA cores, 82 Ray Tracing cores, and 32 GB of GDDR7 memory running on a 256-bit bus, providing 800 GB/s of memory bandwidth, all within a total graphics power of 165 W.

The configuration is similar to the current RTX PRO 4500 Blackwell with an active dual-slot cooler, but it lowers power use, as the actively-cooled edition has a 200 W TGP. NVIDIA attributes the difference in TGP to higher-binned ‘Blackwell’ GB203 dies with better frequency tuning, while targeting similar performance. This server edition SKU also reduces memory bandwidth, running the 32 GB GDDR7 modules at 25 Gbps effective, while the regular blower-style RTX PRO 4500 Blackwell uses full 28 Gbps modules.

The design is intended for server configurations that require hyper-dense setups, where a single-slot solution is cooled by high-RPM server fans. Server farms could install a dozen of these GPUs in parallel within a single system, stacking them as long as there are available PCIe slots. High airflow chassis would push air through the passively cooled GPU shroud, cooling the 165 W TGP.

NVIDIA also indicates that this is not the most efficient GB203 bin with 10,496 CUDA cores, noting that the GeForce RTX 5090 Mobile GPU SKU has only a 95 W TDP. That lower power envelope comes with trade-offs in clock speeds, and clock details for the RTX PRO 4500 Blackwell Server Edition remain undisclosed.

52

Impact Score

Designing carbon materials with Artificial Intelligence at exascale

Argonne researchers are using supercomputers and Artificial Intelligence to predict how carbon changes under extreme heat and pressure. The work could help design nanocarbon materials for medicine, energy, and national security before they are built in the lab.

Snap speeds Snapchat A/B testing with NVIDIA data libraries

Snap has moved key Snapchat experimentation workloads to NVIDIA-accelerated Apache Spark on Google Cloud, aiming to process large daily data volumes faster and at lower cost. The shift supports broader feature testing across engagement, performance and monetization metrics.

NVIDIA expands local agent computing with RTX PCs and DGX Spark

NVIDIA used GTC to highlight new open models, local agent software, and fine-tuning tools aimed at running agentic Artificial Intelligence workloads on RTX PCs and DGX Spark. The announcements focus on privacy, lower operating costs, and better local performance for personal assistants and creative applications.

NVIDIA RTX systems connect to Apple Vision Pro

NVIDIA and Apple are bringing native integration of NVIDIA CloudXR 6.0 to visionOS. The move enables secure streaming of NVIDIA RTX-powered simulators and professional 3D graphics applications to Apple Vision Pro.

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.