NVIDIA has introduced a new version of its professional graphics card, the RTX Pro 5000 Blackwell, which now comes with 72GB of video memory (VRAM). This is a substantial increase over the standard 48GB model released earlier. The card is built on the “Blackwell” architecture, which is the same technology used in NVIDIA’s latest high-end data center chips. This hardware is not intended for gaming; instead, it is designed for engineers, scientists, and AI researchers who need to process massive amounts of data at their desks. The extra memory allows these professionals to run larger AI models and more detailed 3D simulations without the computer slowing down.
The “sarcasm” mentioned in some tech reports refers to the extreme price difference between professional graphics memory and standard computer memory. While you can buy 64GB of DDR5 memory for a desktop computer for around $200, the RTX Pro 5000 Blackwell 72GB is expected to cost approximately $5,000 to $9,000 depending on the retailer. This high price is because the GPU uses a much faster and more expensive type of memory called GDDR7. Additionally, professional cards come with specialized software drivers and direct technical support from NVIDIA, which adds to the total cost.
Tech Specs and Performance figures
The RTX Pro 5000 72GB uses the GB202 graphics processor, which contains 14,080 CUDA cores. These cores are the “engines” that perform the actual calculations. While the number of cores is the same as the 48GB version, the 72GB model has a wider “memory bus” and higher bandwidth, reaching up to 1.3 terabytes per second. This means data can move in and out of the memory much faster, which is critical for tasks like training artificial intelligence or rendering photorealistic movies.
Here are the key features of the Blackwell architecture –
- Fifth-Gen Tensor Cores: These are specialized for AI tasks and offer up to 3x the performance of the previous generation.
- Fourth-Gen RT Cores: These handle “ray tracing,” which is the technology used to create realistic lighting and shadows in 3D designs.
- Efficiency: Despite its power, the card is designed to fit into a standard workstation and uses about 300 watts of power, which is manageable for most professional desktop setups.
Who is the target market for this graphics card?
This card is a “middle ground” option for businesses. Previously, if a professional needed more than 48GB of memory, they had to upgrade to the top-tier RTX Pro 6000, which costs nearly $10,000 and has 96GB of memory. The 72GB version of the RTX Pro 5000 provides a way for researchers to work on large-scale projects, such as Large Language Models (LLMs), without having to pay for the most expensive card on the market. It is also useful for video editors working with 8K or 12K footage, as those files require massive amounts of VRAM to play back smoothly.
For the average user or gamer, this card is not a practical purchase. The specialized drivers are not optimized for games, and you would be paying thousands of dollars for memory that a video game cannot use. However, for a company developing new AI tools, the 72GB capacity can significantly speed up their work. It allows them to keep more data “locally” on the card rather than having to send it back and forth to a slower cloud server, which saves time and improves security.
Release date and pricing information
The NVIDIA RTX Pro 5000 72GB Blackwell was officially launched in late October 2025 and began appearing at specialized professional retailers in early 2026. While NVIDIA has not set a single official price for the public, early listings from distributors like Scan and SHI show the card priced between $5,000 and $9,300.
The price can vary significantly based on whether the card is sold as a standalone part or as part of a pre-built workstation from companies like HP, Dell, or Lenovo. Because these are “enterprise” products, they often include a three-year warranty and long-term software support, which is reflected in the high cost. For businesses that rely on fast AI processing, the high price is often seen as a necessary expense to stay competitive in the rapidly changing tech industry.
