Liquid leads the future, intelligent chain for win-win! The 2026 AI Chip Thermal Management Thousand-Person Conference Unlocks New Opportunities for the Synergy of Storage and Heat Dissipation
In 2026, the 4th Data Center Liquid Cooling and AI Chip Thermal Management Supply Chain Thousand-Person Conference is about to kick off. With the core theme of "Liquid Leading the Future, Intelligent Chain Win-win", we invite you to join this industry peak appointment, deeply decode the technological trends and market opportunities of AI chip thermal management, and explore the new possibilities of the collaboration between storage chips and thermal management!
When the power of NVIDIA's GB300 GPU soared to 1200W, the TDP of the next-generation VR200 chip exceeded 2300W, and the power consumption of a single AI server cabinet entered the 150kW+ era, the thermal management of AI chips has transformed from a "technical supplement" to a "lifeline" for the explosive growth of computing power - and as the core carrier of data, the storage chip Its heat dissipation adaptation capability has become a key variable determining the stability of AI systems. Traditional air cooling can no longer bear the weight of high power density. Liquid cooling technology is reshaping the industrial pattern of the synergy between thermal management and storage of AI chips with an irreversible trend, driving the market worth hundreds of billions to explode at an accelerated pace.
Market Outlook: AI chip thermal management Ignites a blue ocean worth hundreds of billions, storage heat dissipation becomes a key growth factor
The global AI chip thermal management market is experiencing explosive growth: the global market size is expected to approach 100 billion yuan by 2026, with the domestic market size reaching 15 billion yuan, and the compound annual growth rate from 2024 to 2026 will exceed 300%. Goldman Sachs predicts that the global server cooling market size will reach 14 billion US dollars in 2026, with AI chip thermal management contributing the core increment, increasing by 114% year-on-year to 11.6 billion US dollars.
The leapfrog increase in penetration rate has become the core signal: The penetration rate of liquid cooling in AI training servers will rise to 74% by 2026, and 100% of AI servers on all racks will be equipped with liquid cooling as standard. The "dual drive" from the policy end has been continuously strengthened. The three major domestic operators require that over 50% of their projects apply liquid cooling by 2025. In Shanghai, it is mandatory that the proportion of liquid-cooled cabinets in newly built intelligent computing centers exceed 50%, and the energy efficiency red line of PUE≤1.25 makes liquid cooling the core carrier of green computing power. It is worth noting that the heat dissipation demand for storage chips has soared simultaneously - high-density storage architectures such as 3D NAND and HBM, due to massive data reading and writing in the processing of trillions of parameters in AI large models, have led to a sharp increase in power consumption. Liquid-cooled SSDS and thermal adaptation of storage chips have become industry essentials, driving the market demand for collaborative solutions of storage and thermal management to increase by 400% annually. From NVIDIA's GB300 all-cold plate solution to TSMC's CoWoS packaging and liquid cooling integration technology, the localization of AI chip thermal management and the synergy of storage and heat dissipation are becoming the two core growth points in the market.
Technological iteration: AI chip thermal management enters the "precise heat dissipation" era, and storage collaboration technology becomes the key to breakthrough
The exponential increase in computing power density has driven the thermal management technology of AI chips to evolve from "passive cooling" to "active adaptation". Currently, cold plate liquid cooling dominates the large-scale application with over 90% of the market share. With the single-cabinet power exceeding 150kW, immersion liquid cooling and bidirectional cold plates have become the core choices for high-power scenarios. By 2026, they will form a "multi-dimensional heat dissipation" pattern together with direct silicon liquid cooling - TSMC's IMC-Si cooler has achieved a cooling power of 3.4kW. It directly acts on the silicon-based back of the chip, and its thermal efficiency is three times higher than that of traditional cold plates.
In the field of storage collaboration, technological innovation continues to break through: Solidigm's liquid-cooled SSD achieves single-sided heat dissipation upgrade through a cold plate kit, eliminating the reliance on fans, and is compatible with 1U high-density AI servers. The hot-swswapping function solves the pain points of data center maintenance. HBM heat dissipation has become a technical focus. Through a liquid cooling design co-packaged with the GPU, the temperature of the memory chip is controlled within the safety threshold of 85℃, ensuring the stability of high-bandwidth data transmission. Thermal simulation technology is widely applied in the thermal design of memory chips. Tools such as Simdroid-EC can accurately simulate the temperature distribution under different working conditions, with an error controlled within ±1℃, providing a scientific basis for the thermal adaptation of memory chips. In terms of core materials, electronic fluorinated liquid, with its high insulation and low viscosity characteristics, has become the "blood" of liquid cooling. The domestic substitution rate has risen from less than 20% in 2022 to over 80% expected in 2026. The thermal conductivity of liquid metal TIM material exceeds 80W/m·K, effectively reducing the thermal resistance between the memory chip and the cold plate, making it an essential choice for high-power scenarios.
Intelligent Chain Win-win: Storage and Thermal Management Synergy, Linking the core forces of the industrial chain
The explosion of AI chip thermal management has never been a victory of a single breakthrough point, but rather the result of the coordinated efforts of the entire chain from "computing power - storage - heat dissipation". From the upstream coolant (fluorinated liquid, coal-based dielectric liquid), cold plates (toothed type, buried pipe type), quick couplings, to the midstream CDU cooling units, thermal management system integration, and then to the downstream liquid-cooled SSDS, AI server storage modules, an ecological network covering the entire chain is taking shape, and storage chip agents are at the key hub of this ecosystem.
This thousand-person conference, with the original intention of "Intelligent Chain Win-win", focuses on the core of AI chip thermal management and gathers the core players of the global industrial chain: not only do technology leaders such as NVIDIA and TSMC share their cutting-edge layouts in CoWoS packaging heat dissipation and GPU-storage co-cooling, but also domestic cold plate and coolant enterprises showcase their breakthroughs in domestic production. It not only delves deeply into the disputes over the implementation of technical routes such as bidirectional cold plates and immersion liquid cooling, but also focuses on practical issues like the heat dissipation adaptation of storage chips and the large-scale application of liquid-cooled SSDS. As a storage chip agent, you will connect with the storage and thermal management support demands of hundreds of billions of AI servers here, discuss the compatibility optimization of storage products and liquid cooling systems, link up with core customers such as cloud service providers, AI enterprises, and data center operators, and achieve ecological win-win in the era of explosive computing power.
In 2026, liquid cooling has arrived, and a win-win situation is within reach!
The 4th Thousand-Person Conference on Data Center Liquid Cooling and AI Chip Thermal Management Supply Chain
We invite you to join a thousand industry elites to explore the trends of AI chip thermal management technology, discuss the collaborative opportunities of storage and heat dissipation, and jointly build an industrial ecosystem.
Let's take liquid cooling as the bridge, storage as the core, and innovation as the wing
In the vast ocean of AI computing power, unlock the infinite possibilities of the collaboration between thermal management and storage!
