NVIDIA’s latest professional workstation GPU, the RTX Pro 6000, has arrived with a spec sheet that firmly cements it as a Titan-class card. With its high core count, extensive memory capacity, and a power budget that pushes the limits of PCIe 5.0, the RTX Pro 6000 stands as a serious competitor to the RTX 5090 – targeting a different audience. However, when comparing the two, it’s clear that the Pro 6000 is essentially a full-die version of the RTX 5090, with some notable differences.
RTX Pro 6000 vs. RTX 5090: Key Specs
At the heart of the RTX Pro 6000 is the GB202 GPU, the same silicon used in the RTX 5090. However, instead of the 21,760 CUDA cores found in the 5090, the Pro 6000 is equipped with 24,064 cores, making it one of the most powerful implementations of the GB202 die outside of the data center-focused GB200 chips.
Similar to the modded 47GB VRAM RTX 4090, A major differentiator is the memory configuration. The RTX Pro 6000 boasts an impressive 96GB of GDDR7 memory with error correction (ECC), running on a 512-bit bus that provides a massive 1,792 GB/s of memory bandwidth – identical to the RTX 5090. This is a significant increase over the 5090’s 32GB of GDDR7, making the Pro 6000 far more suited for AI workloads, like running locally 70B LLM models.
A Look at Power and Cooling
Another key difference between the two GPUs is power consumption. While the RTX 5090 is rated for 575W of board power, the Pro 6000 pushes this further to a full 600W – the maximum that the 12V-2×6 connector is officially rated to handle. This represents a 25W increase, and given that the RTX 5090 already pushed thermals to the limit, managing additional heat output will be a critical factor for professional users deploying these cards in dense workstation environments.
Interestingly, NVIDIA appears to have planned for the additional thermal load well in advance. The 5090 Founders Edition already featured a heavily thinned backplate, likely anticipating the need for additional memory modules in future iterations. With the Pro 6000, we now see a dual-sided memory configuration using higher-density modules, allowing NVIDIA to achieve 96GB of VRAM without significantly increasing PCB size.
PCIe 5.0 and AI-Focused Design
Both the RTX Pro 6000 and the RTX 5090 leverage PCIe Gen 5, a feature that offers theoretical improvements in bandwidth but remains to be fully realized in real-world workloads. For professional applications, however, the added bandwidth could be more beneficial than in gaming scenarios, particularly for high-speed data transfer in AI computing.
NVIDIA is positioning the RTX Pro 6000 as a workstation and AI processing powerhouse. While the Pro 6000 is unlikely to be found in consumer gaming rigs, its presence sets the stage for what we can expect from the next generation of high-performance GPUs.
Final Thoughts
The RTX Pro 6000 is effectively a full-die GB202 card, offering more CUDA cores, significantly higher memory capacity, and a slightly higher power target than the RTX 5090. While both GPUs share a similar foundation, the Pro 6000 is clearly optimized for AI and professional workloads rather than gaming. The increased memory and nearly full utilization of GB202 make it one of the most powerful workstation GPUs NVIDIA has released to date. However, with a 600W power draw and a dual-sided memory configuration, cooling and efficiency will be critical factors in its real-world deployment.
As we get closer to availability, we’re eager to see real-world benchmarks and thermal data to determine just how well this card performs in demanding professional environments.