NVIDIA introduces RTX PRO 5500 Blackwell with 84GB GDDR7 and 600W power limit
NVIDIA has expanded its professional workstation lineup with the addition of the RTX PRO 5500 based on the Blackwell architecture. This new model sits between the RTX PRO 5000 and the flagship RTX PRO 6000, providing a significant boost in memory capacity and a much higher power ceiling than its predecessor.
Technical Specifications and Features
Detailed specifications provided by PNY reveal that the RTX PRO 5500 features 21,760 CUDA cores and 84GB of ECC GDDR7 memory. The memory is connected via a 416-bit interface, delivering a bandwidth of 1,398 GB/s. The card's maximum board power (TDP) is rated at 600W, matching the power budget of the top-tier models. For display connectivity, it offers up to four DisplayPort 2.1b outputs.
The GPU includes three ninth-generation NVENC engines and three sixth-generation NVDEC engines. A key feature for data center and workstation deployments is Multi-Instance GPU (MIG) support, which allows the card to operate as a single 84GB instance or be split into two isolated 42GB instances. The card is expected to be available in both active air-cooled and RXM liquid-cooled configurations.
Market Positioning
The RTX PRO 5500 is positioned much closer to the flagship RTX PRO 6000 than its numbering might suggest. Compared to the RTX PRO 6000 (which features 24,064 CUDA cores, 96GB GDDR7, and a 600W limit), the 5500 has approximately 9.6% fewer CUDA cores and 12.5% less memory. However, it shares the same 600W power limit.
This new model addresses a substantial gap in the lineup. The current RTX PRO 5000 series operates at a 300W limit and offers either 48GB or 72GB of GDDR7. By offering 84GB of high-speed memory and full power headroom, the RTX PRO 5500 targets high-end rack-mounted workstation environments. NVIDIA has listed the product as “Coming Soon” on its regional sites, with official pricing yet to be announced.
Гика
