Nvidia RTX Pro 6000 GPU Price Nearly Doubles to $16,000 Amid AI Hardware Demand

How Much?! Nvidia RTX Pro 6000 Will Now Set You Back $16,000

Hey kid, you want to buy a professional graphics card? It's $16,000. That's Nvidia's pitch for anyone who wants a top-tier Nvidia prosumer GPU in 2026. Its flagship RTX Pro 6000 -- the full-fat version of the RTX 5090 built for AI and design work -- is now priced at a staggering $16,000. That's almost double the $8,565 launch price from March 2025, VideoCardz reports. Prices of Nvidia's consumer graphics cards have skyrocketed in recent months, and they weren't cheap when they launched, either. The RTX 50-series debuted in 2025 with a flagship that was an at-the-time record-breaking $2,000, while more modest cards like the RTX 5070 were still $550. But now, just over a year and a half later, the RTX 5090 is closing in on $5,000, and the RTX 5070 has just reached an average price of $900 on Newegg. But perhaps we can take comfort in the fact that Nvidia's professional GPU pricing is equally ridiculous. Its flagship workstation GPU, the RTX Pro 6000, is now $16,000. That's a near doubling from last year and a 21% increase since June, when the price was a mere $13,250. There hasn't been any kind of announcement of these price hikes; Nvidia just sells the card at that price now. It has the same GPU, the same cooler, and the same memory configuration, but a massively inflated price tag. The price increase must come at least partially from the equally crazy RAM prices in 2026. The RTX Pro 6000 comes with 96GB of high-speed GDDR7, which will make up a sizeable portion of the bill of materials. The GB202 GPU, at its heart (and the same one used in the RTX 5090), is also complete with the full complement of 24,064 CUDA cores, so it will be especially binned for this purpose. But even then, it's hard to imagine the card costing anywhere near what it's being sold for. TechPowerUp estimates that the 3GB GDDR7 chips Nvidia uses for these cards cost around $60-$70 each. So even if those doubled in cost since June (which they haven't), that would still be only an extra $2,240, not the near-$3,000 Nvidia is charging. Still, it helps to be the only real game in town for high-end GPUs with CUDA support. Although AMD's options are credible and have their own uses for AI developers and 3D designers, Nvidia's CUDA platform is a hell of a moat that it takes full advantage of. Professional GPU users who want the top cards will simply have to pay the price or make other arrangements.

NVIDIA's RTX PRO 6000 Blackwell GPU price almost doubles, reaching $16,000

The NVIDIA RTX PRO 6000 Blackwell Workstation Edition graphics card launched last year, and it's basically a souped-up dual-slot GeForce RTX 5090 Founders Edition built for AI. In addition to an expanded CUDA Core count of 24,064 compared to the RTX 5090's 21,760, the RTX PRO 6000 Blackwell also features 96GB of GDDR7 memory, with a similar 512-bit interface as the RTX 5090 offering up a massive 1,792 GB/s of bandwidth. Even when it launched last year, the AI boom and memory crisis were still a thing, albeit not quite as insane as it is right now. An RTX 5090-class workstation GPU with 96GB of fast GDDR7 memory meant it sold for $8,565 at launch, making it NVIDIA's most expensive desktop-class graphics card for workstations. Cut to 2026, and that's still the case, except that it's now almost double the price. On NVIDIA's own marketplace and storefront, the RTX PRO 6000 Blackwell is now priced at $16,000 USD (via Videocardz), an 87% increase over the launch MSRP and a 20% increase over the $13,250 price point in June. The main reason for the price increase, which we safely assume, is the sharp increase in component costs, specifically for memory. For the RTX PRO 6000 Blackwell GPU, which features 32 x 3GB GDDR7 modules, the memory price increases we've seen in recent months are multiplied by 32. Also, as we've heard from multiple sources, 3GB GDDR7 memory modules have recently become substantially more expensive than 2GB modules; this would also factor into the price. Although a workstation product, rumors have suggested that NVIDIA was planning to use 3GB modules for its GeForce RTX 50 SUPER Series - plans which are now on hold. The current RTX Blackwell-powered GeForce RTX 50 Series GPUs all feature 2GB GDDR7 memory modules. Of course, there is some irony in the NVIDIA RTX PRO 6000 Blackwell's massive price increase. With demand for AI hardware driving memory capacity to the brink and prices increasing exponentially, NVIDIA's success and AI hardware leadership are key reasons prices are becoming untenable. So much so that NVIDIA is now nearly doubling the price of its flagship GPU hardware.