NVIDIA's Blackwell Platform Set to Transform Liquid Cooling
NVIDIA's upcoming Blackwell platform is gearing up to make a notable difference in liquid cooling solutions. Expected to launch in the fourth quarter of 2024, the platform is likely to spark a considerable spike in liquid cooling adoption. Current estimates indicate a rise in its penetration from around 10% in 2024 to over 20% by 2025.
Shift Driven by ESG Awareness and AI Servers
With a growing global consciousness about environmental, social, and governance (ESG) factors, many organizations are shifting from traditional air cooling methods to more effective liquid cooling systems. Additionally, the increasing deployment of AI servers by cloud service providers (CSPs) is fueling this transition. NVIDIA holds a strong position in the AI server market, with its GPUs accounting for nearly 90% of the total market share.
Current Market Landscape
While demand for NVIDIA's Blackwell platform is expected to rise sharply, initial shipments this year will be limited. This shortfall is due to manufacturers finalizing essential product testing and validation processes, including high-speed data transmission and optimizing cooling designs. The Blackwell platform's significant energy consumption, especially with the GB200 full-rack solution, underscores the need for more efficient cooling technologies.
Challenges and Opportunities for ODMs
Despite the anticipated increase in liquid cooling adoption, the current implementation rate within existing server ecosystems remains relatively low. Original Design Manufacturers (ODMs) face several hurdles, such as the risk of leakage and subpar cooling performance as they begin this transition.
Looking Ahead: Projections for the Blackwell Platform
As we look toward 2025, forecasts suggest that NVIDIA's Blackwell platform could dominate over 80% of the high-end GPU market. This dominance is likely to attract numerous manufacturers of power supplies and cooling solutions into the AI liquid cooling sector, initiating a new competitive environment in the industry.
Liquid Cooling Trends Among Major CSPs
In light of the rising demands from AI technology, leading CSPs such as Google, AWS, and Microsoft are rapidly enhancing their AI server deployment. They primarily rely on NVIDIA GPUs in conjunction with custom-designed Application-Specific Integrated Circuits (ASICs). The thermal design power (TDP) for the GB200 NVL72 racks is approximately 140 kW, illustrating a pressing need for efficient liquid cooling solutions, with Liquid-to-Air (L2A) technology becoming a favored approach.
Supplier Landscape for Liquid Cooling Components
Google is leading the way in liquid cooling among CSPs, utilizing a blend of air and liquid cooling systems for its Tensor Processing Unit (TPU) chips. The cold plates crucial to these systems come mainly from suppliers like BOYD and Cooler Master. Meanwhile, Alibaba's expansion of liquid-cooled data centers in China stands out, even as other providers stick with traditional air cooling.
Key Suppliers and Innovations
TrendForce has pinpointed key suppliers for liquid cooling components that are essential for NVIDIA's GB200 racks. Companies such as Asia Vital Components and Cooler Master are at the forefront of cold plate manufacturing, while others like Vertiv and Delta Electronics supply coolant distribution systems. International firms dominate the quick disconnect components, which are vital for preventing cooling leaks. By 2025, it's expected that Taiwanese companies like LOTES and Fositek will also join the supply chain, helping to ease supply shortages.
About TrendForce
TrendForce is well-regarded as a global authority in technology sector insights and analysis. Founded in 2000 in Taipei, Taiwan, the company has built a substantial client base of over 500,000, offering precise analysis across five key research areas including Semiconductor and Energy Research. TrendForce has also broadened its reach into China, strengthening its influence in the Asian market.
Frequently Asked Questions
What is the significance of NVIDIA's Blackwell platform?
NVIDIA's Blackwell platform is anticipated to significantly boost liquid cooling adoption, increasing from 10% in 2024 to over 20% in 2025.
How does liquid cooling technology benefit AI servers?
Liquid cooling technology effectively manages heat generated by high-performance AI servers, enhancing both efficiency and overall system performance.
What challenges do ODMs face in adopting liquid cooling?
ODMs need to tackle issues like potential leaks, performance concerns, and transitioning from traditional air cooling systems.
Who are the main suppliers for liquid cooling solutions?
Key suppliers include Cooler Master, Asia Vital Components, and Vertiv, among others, providing critical components for liquid cooling systems.
What impact will Blackwell have on the GPU market?
The Blackwell platform is projected to exceed 80% market share in high-end GPUs by 2025, thus influencing competitive dynamics within the industry.