NVIDIA GB200 Rack-Mounted Solution Optimization

Investing.com — The NVIDIA (NASDAQ:NVDA) GB200 rack-mounted solution requires further optimization and adjustments in its supply chain, according to recent research by TrendForce.

Supply Chain and Design Challenges

The complex design specifications of the GB200 rack, including high-speed interconnect interfaces and thermal design power (TDP) requirements that exceed market norms, have led to this need for optimization. As a result, TrendForce predicts that mass production and peak shipments will likely occur between Q2 and Q3 of 2025.

Rack Series Overview

The NVIDIA GB rack series, which includes the GB200 and GB300 models, features complex technology and higher production costs. This positions it as a preferred solution for large Cloud Service Providers (CSPs) and potential users such as Tier-2 data centers, national sovereign cloud providers, and academic research institutions working on High-Performance Computing (HPC) and Artificial Intelligence (AI) applications. The GB200 NVL72 model is anticipated to be the most popular in 2025, possibly representing up to 80% of total deployments as NVIDIA intensifies its market efforts.

NVLink Technology

NVIDIA's proprietary NVLink technology is crucial for enhancing the computational performance of AI and HPC server systems, allowing high-speed connections between GPU chips. The GB200 utilizes the fifth-generation NVLink, providing total bandwidth that significantly exceeds the current industry standard, PCIe 5.0.

Thermal Design Power (TDP)

The TDP of the HGX AI server, which dominated in 2024, typically ranges from 60 kW to 80 kW per rack. However, the GB200 NVL72's TDP reaches 140 kW per rack, doubling the power requirements. This has prompted manufacturers to accelerate the adoption of liquid cooling solutions, as traditional air cooling methods are inadequate for handling such high thermal loads.

Concerns About Shipments

The advanced design requirements of the GB200 raise concerns regarding potential delays in component availability and system shipments. TrendForce notes that production of Blackwell GPU chips is progressing mostly as planned, with only limited shipments expected in 4Q24. Production volume is expected to gradually increase from 1Q25 onward. However, due to supply chain adjustments for AI server system components, shipments at the end of 2024 are projected to be lower than industry expectations. Consequently, TrendForce predicts the peak shipment period for the GB200 full-rack system to be delayed to between Q2 and Q3 of 2025.

Liquid Cooling Necessity

The GB200 NVL72's TDP of 140 kW makes liquid cooling indispensable, outpacing traditional air-cooled solutions. The adaptation of liquid-cooling components is accelerating, with leading industry players investing significantly in research and development for these technologies.

Notably, suppliers of coolant distribution units are working to enhance cooling efficiency by increasing rack sizes and developing more efficient cold plate designs. Current sidecar CDUs can dissipate between 60 kW and 80 kW, but future designs are anticipated to double or triple this cooling capacity. The emergence of liquid-to-liquid in-row CDU systems has enabled cooling performance to exceed 1.3 mW, with further improvements expected as computational power demands continue to increase.

This article was generated with the support of AI and reviewed by an editor. For more information, see our T&C.