NVIDIA RTX Blackwell Workstation GPU Leaked Online; Equipped with 96GB GDDR7 Memory and 512-Bit Interface

NVIDIA appears to be gearing up for the launch of its next-generation RTX Blackwell “workstation”GPUs, with a new 96GB model recently identified in shipping documents. This suggests a significant advancement in their professional GPU lineup, likely focused on high-performance computing (HPC) and artificial intelligence (AI) workloads.

NVIDIA’s Ambitious Blackwell Workstation GPU Series: A Closer Look

As NVIDIA continues to unveil its upcoming consumer graphics cards, there has been limited information regarding the workstation models. However, recent shipping manifests, as reported by ComputerBase, indicate a promising new model that features 96GB of GDDR7 memory. This new addition suggests the company is making significant strides to bolster its workstation offerings with the anticipated RTX Blackwell series.

Intriguingly, the forthcoming GPU seems to operate on a 512-bit memory interface—a feat that NVIDIA can achieve by utilizing a “clamshell”configuration comprising 3GB GDDR7 modules. This setup is a departure from the 2GB modules employed in the GeForce RTX 5090, hinting that NVIDIA will harness the potential of larger memory configurations in their Blackwell workstation GPUs. The laptop variant of the GeForce RTX 5090 already utilizes 3GB modules, further validating the feasibility of this 96GB model.

Additionally, emerging reports suggest an exclusive PG153 board will power this new GPU, which has not been seen in prior RTX Blackwell models. The introduction of 96GB GDDR7 memory marks an almost twofold increase over the previous RTX 6000 “ADA” model’s VRAM. This strategic increase is likely designed to enhance performance for demanding applications, particularly in AI and HPC scenarios.

While NVIDIA has yet to officially announce the RTX Blackwell workstation series, the indications point to active development. Industry experts speculate that a formal presentation could coincide with the GTC 2025 event, which begins in March, showcasing NVIDIA’s innovations tailored for professional applications.

Current NVIDIA Workstation Graphics Card Lineup

Graphics Card	RTX 6000	RTX A6000	Quadro RTX 8000	Quadro RTX 6000	GV100 Frame
GPU	Ada Lovelace GPU	Ampere GPU	Turing GPU	Turing GPU	GPU Vault
GPU Process	5nm	8nm	12nm	12nm	12nm
Die Size	608mm²	628mm²	754mm²	754mm²	815mm²
GPU Cores	18,176 Cores	10,752 Cores	4,608 Cores	4,608 Cores	5,120 Cores
Tensor Cores	568 Cores	656 Cores	576 Cores	576 Cores	640 Cores
Boost Clock	2.50 GHz	1.80 GHz	1.77 GHz	1.77 GHz	1.62 GHz
Single Precision	91.1 TFLOPs	38.7 TFLOPs	16.31 TFLOPs	16.31 TFLOPs	16.66 TFLOPs
Ray Tracing Spec	210.6 TFLOPs	75.4 TFLOPs	10 GigaRays/Sec	10 GigaRays/Sec	N/A
VRAM	48 GB GDDR6X	48GB GDDR6	48GB GDDR6	24GB GDDR6	32 GB HBM2
NVLINK VRAM	N/A	96 GB With NVLINK	96 GB With NVLINK	48 GB With NVLINK	N/A
Memory Bus	384-bit	384-bit	384-bit	384-bit	4096-bit
Memory Bandwidth	960GB/s	768GB/s	672 GB/s	672 GB/s	870GB/s
TDP	300W	300W	~225W	~200W	250W
Launch Price	$6800	$4650	$10000	$6300	$9000
Launch Date	Q1 2023	Q4 2020	Q4 2018	Q4 2018	2018

With these developments, NVIDIA is clearly setting the stage for its future in the professional GPU market, pushing the envelope on what performance and memory capacity can achieve in real-world applications. Keep an eye on the upcoming months as more updates about the RTX Blackwell series are likely to emerge.

Source & Images