NVIDIA's BoM for its upcoming Vera Rubin "NVL72" rack shows a massive surge in memory prices that now make up 26% of the total system cost.
Rising Memory Prices & Demand Pushes Memory Costs Up 26% For NVIDIA's Vera Rubin Versus 9% on Grace Blackwell Racks
Vera Rubin is in production and is confirmed for first shipments in the third quarter of 2026, followed by volume ramp in the fourth quarter. While NVIDIA cooks up its grandest AI platform to date, the pricing is also going to be grand.
Morgan Stanley Research has shared its estimated BoM for NVIDIA's Vera Rubin "VR200" NVL72 rack, which will feature 72 Vera Rubin GPUs, and each tray will house four Rubin GPUs with two Vera CPUs. We'll need to get into the details, which will give you a better understanding of the platform & its cost breakdown.
Let's start with the basics: the NVIDIA NVL72 rack is called Oberon and makes use of 72 GPUs. A single Vera Rubin tray houses 4 "Rubin" GPUs and 2 "Vera" CPUs, as mentioned above. Two GPUs and a single CPU are housed on a motherboard, which is called Superchip. There are 36 Superchips on the NVL 72 rack. So that's a total of 72 GPUs and 36 CPUs.
Each Rubin GPU houses 288 GB of HBM4 memory, and each Vera CPU comes with 1.5 TB of LPDDR5X memory. For an NVL72 rack, that's 20.7 TB of HBM4 memory and 54 TB of LPDDR5X memory. There is a lot more that goes into Vera Rubin NVL72 racks, such as networking, cooling, power, interconnects, etc.
From Morgan Stanley's data, which highlights an "estimated" BoM (Bill of Materials), we can get a better understanding.
Starting with the main cost, the Rubin GPU. The Rubin GPUs are expected to cost almost $4 Million, making it the single-most expensive cost within the NVL72 VR200 rack. That's a 57% bump over the Blackwell NVL72 B300 rack, which had the GPU price around $2.5 million. That's $55,000 US per GPU.
The second biggest cost is for the memory, which shouldn't be surprising given how constrained the supply is for LPDDR and HBM technologies right now. With Rubin, the memory alone sees a bump of 435%, jumping from $373,939 in Grace Blackwell to over $2 Million on the Vera Rubin platform. The memory price is combined for both LPDDR5X and HBM4.
The Vera CPUs amount to $180,000 of the total cost, which puts each chip at roughly $5000 US. All three combined, the total cost of the rack ends up at $6.14 million. The remaining ~$2 million costs include the NVLink Switches, Networking chips, cooling, power supply, PCB, ABF substrates, MLCC, and additional components. The PCB sees the second-highest bump of 233%, going up from $35,100 in Blackwell to $116,730 on Rubin.
NVIDIA’s upcoming Vera Rubin NVL72 rack marks a powerful new chapter in AI infrastructure, with a total estimated BOM of $7.8 million, driven largely by a 57% jump in GPU costs and a dramatic 435% surge in memory expenses.
Memory now accounts for 26% of the entire system cost, highlighting the intense supply constraints and soaring demand for HBM4 and LPDDR5X. As production ramps for Q3 2026 shipments, Vera Rubin is poised to deliver unprecedented performance at a grand price, reinforcing NVIDIA’s dominance while underscoring the rising cost of pushing the frontiers of AI.
About the author: A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as Wccftech's Senior Editor for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking.
Follow Wccftech on Google to get more of our news coverage in your feeds.
