On August 6, Envision Group commissioned the Ulanqab Star River Base, the world's largest single-body AI computing supercomplex, covering 120,000 square meters with million-level parallel computing cards and millions of PetaFLOPS, setting new global AI infrastructure density records. With 2GW planned capacity, the flagship of the Gobi Mission targets China's largest AI c...
Anthropic plans to secure at least 1.4GW of data center computing resources in Australia with $15 billion in total infrastructure investment, launching no less than 1GW by the end of 2027. The Southern Hemisphere strategy responds to US bottlenecks, where Blackstone-owned QTS canceled its 850-hectare Virginia project amid local opposition. Backed by annualized revenue gr...
Liquid cooling has become an irreversible industry trend in the AI era, with cold plate cooling remaining mainstream for the next 3-5 years. Micro-channel MLCP integrates heat spreaders and cold plates, shortening heat transfer paths and cutting multi-layer TIM resistance. 3D-printed cold plates create TPMS lattice and turbulent flow structures impossible with traditiona...
Zhejiang Xinhan Intelligent Technology has completed its Pre-A round financing with proceeds exceeding 100 million RMB, backing the industrialization of phase change liquid cooling, which absorbs and transfers heat through liquid-gas phase transition for far higher efficiency than single-phase cooling. The funds target R&D of core phase change materials and system archit...
Liquid cooling has evolved from an optional upgrade to mandatory infrastructure for kilowatt-scale AI computing. Traditional air systems face dual dilemmas of near-half facility electricity consumption and insufficient heat dissipation for 120-155kW high-density racks. Rack-integrated air-liquid CDUs enable zero-downtime phased upgrades, lowering PUE and long-term energy...
Liquid cooling has become mandatory infrastructure as AI cabinet power surges 6-8 times above traditional racks, with NVIDIA full-rack shipments driving penetration from zero in 2025 to explosive growth in 2026 as NVIDIA and ASIC platforms adopt the technology. The global supply chain is being reshaped as Chinese enterprises including Yutong Technology, Jinfu Technology ...
Liquid cooling is transforming from an auxiliary heat dissipation method into core AI infrastructure, with NVIDIA GB200 NVL72 racks requiring 120kW cooling capacity making thermal management decisive for chip operation and cabinet layout. Cold plate systems attach heat exchangers close to heat sources while cutting fan and refrigeration energy. The industry chain value l...
Liquid cooling has evolved from optional upgrade to standard infrastructure for green AI data centers, replacing fan and precision air conditioning systems that consume over 40% of facility power and struggle beyond 30-120kW rack loads. Liquid enables efficient direct-contact cooling through two mature routes: cold plate systems installing customized plates on CPUs and G...
Policy and technology are jointly making liquid cooling standard AI infrastructure: April policies mandated efficient cooling for new intelligent computing centers, while NVIDIA's June 21 Vera Rubin launch adopted 100% full liquid cooling with fan-free design and rigid metal manifolds replacing rubber hoses. Cold plates demand micron-level microchannel precision, led by ...
Envicool's two decades of thermal expertise paved its way into the NVIDIA MGX ecosystem, after NVIDIA's global audit team visited its Shenzhen Longhua facility in spring 2025 and selected its UQD quick connectors and MQD manifolds, marking a leading position among Chinese liquid cooling enterprises. Growing from telecom base station temperature control into AI data cente...
Jiangxi Zero-Carbon Future developed the Bingnengbao low-temperature phase change cold storage unit, using dedicated -18°C phase change materials with high cold storage density and stable cold release. Following Valley Power logic, it stores cold during off-peak night hours and shuts down main compressors during daytime peak-price periods, with melting materials sustaini...
The liquid cooling manifold serves as the core vascular system of AI data centers, a precision multi-branch distribution hub that delivers coolant evenly to server cold plates and balances thermal circulation. Paired with CDUs, it forms a closed-loop heat exchange system solving uneven cooling across multi-server clusters. High-precision manifolds control branch flow, pr...
As AI cabinet power density surpasses 100kW, the cooling distribution manifold has become the indispensable vascular link between CDU and server cold plates, evenly distributing coolant and collecting heated fluid while setting the reliability baseline for the entire loop. With dozens of welding points and quick-connect interfaces, leakage, uneven flow or trapped gas can...
NVIDIA's Rubin architecture completely cancels traditional cabinet hoses, driving a 90% surge in per-rack pipeline value and upgrading the liquid cooling manifold from a marginal auxiliary part into high-value core hardware dominating fluid distribution in full liquid cooling racks. The manifold precisely distributes low-temperature coolant to every server cold plate and...
Within liquid cooling systems, the CDU governs total cooling capacity and fluid temperature while the server liquid coolant manifold delivers cooling power precisely to every thermal terminal, and their coordination determines whether AI servers run steadily at full-load 120kW. Manifolds adopt 304 or 316L stainless steel with 0.8-1.5mm walls sealed by laser welding or va...
Laser welding is reshaping manufacturing reliability for liquid cooling manifolds, where weld quality directly decides sealing safety and long-term system stability for high-density AI racks. Traditional argon arc welding suffers thermal deformation over 0.3mm, high leakage risk and low productivity, while vacuum brazing demands high capital investment, strict material c...
Shielded server cabinets combine waveguide ventilation, liquid cooling, vortex tube and heat pipe technologies for hybrid passive-active thermal management. Waveguide windows use honeycomb metal grids where apertures below half the electromagnetic wavelength block signals while allowing airflow, delivering over 100dB shielding with explosion-proof fans. When single-cabin...
AI data center thermal management has evolved beyond chip-level cooling into full-system liquid cooling, treating chip dissipation as merely the first link of a six-stage chain: chip heat generation, cold plate absorption, manifold distribution, distributed pump circulation, rack-level exchange and facility water loop discharge. NVIDIA's GB300 NVL72 marks a new era with ...
Manifolds and quick-disconnect fittings form the thermal last mile of liquid cooling, handling reliable coolant delivery and return between CDUs and server cold plates. Manifolds split a single CDU pipeline into multiple branches—8/12/16/24 channels at 5-15 L/min each with pressure drop below 30kPa—in top-of-rack, in-rack or row-level CDM forms, with flow balance critica...
Cold chain warehouse management relies on temperature and humidity control hardware to protect commodity quality, with constant-temperature and controlled-atmosphere cold rooms as two mainstream forms—the latter adjusting oxygen and carbon dioxide levels to put produce into dormancy. Complete workflows span insulation storage, warehousing, sorting and packaging, loading-...
NVIDIA's GTC 2026 landmark move takes direct control of the liquid cooling supply chain from OEMs, establishing a certified supplier roster across cold plates, quick connectors, manifolds and CDUs. The Vera Rubin NVL72 supercomputing platform—72 Rubin GPUs, 36 Vera CPUs, 2300W per GPU and 190-230kW racks—adopts 100% full liquid cooling as factory standard with 45°C warm ...
Two pipeline routes reshape AI data center liquid cooling: mature stainless steel and innovative polymer systems. 304/316L steel dominates with corrosion resistance, over 25MPa burst pressure and 15-20 year life, though heavy weight, on-site welding and water treatment raise costs. PPH and PVDF polymers solve these pains and weigh 75% less with factory prefabrication and...
A Variable Frequency Drive (VFD) is the core power control device modernizing cold room chiller units through AC-DC-AC conversion and PWM technology. Traditional compressors run at fixed speed with on-off control, causing temperature instability, energy waste and mechanical wear. VFD inverters enable stepless compressor speed regulation, intelligently boosting output for...
Advanced CDU technology is evolving to empower ultra-high-density AI liquid cooling, with leading suppliers launching 2.6MW and 3MW units (5MW in R&D) for 100kW+ racks and 80-200kW cabinets. Turbulent plate designs lift heat transfer 25-40% with 2.0-2.5MPa pressure resistance and 316L or titanium composite plates. Two routes have taken shape: cold-plate CDUs circulate de...
NVIDIA is redefining AI data center liquid cooling from Blackwell to Rubin, shifting from standalone GPUs to standardized full-stack AI Factory racks. The VR200 Vera Rubin platform integrates cooling, power supply and renovation, with GB200, GB300 and Rubin racks peaking at 142kW. NVIDIA confirms cold plate cooling with 45°C warm-water single-phase direct chip cooling as...
The Middle East data center market is booming, projected to grow at 23.91% CAGR through 2030 to exceed $6.98 billion, as the UAE and Saudi Arabia invest over $300 billion in AI infrastructure. The UAE operates more than 30 data centers while Saudi Arabia plans 6.6GW of computing facilities by 2034, driving massive renovation demand across legacy Gulf facilities. Geopolit...
Liquid cooling has become the standard thermal solution for high-power AI racks, with NVIDIA GB200 adopting full cold plate cooling and GB200 super chips reaching 2700W TDP; industry forecasts predict rack power density breaking 1MW by 2029, rendering air cooling obsolete as it wastes nearly forty percent of power on air conditioners. ZTE and Schneider data confirm liqui...
Rack level liquid cooling treats each server cabinet as an independent thermal management unit, integrating CDU, cold plates, sealed pipelines, heat exchangers and intelligent control to boost cooling performance and cut power consumption. Three architectures dominate: direct-to-chip cold plates, full immersion and rear door heat exchange, with rear door solutions enabli...
Three liquid cooling technologies—cold plate, immersion and spray—cover AI data center thermal management, cutting PUE below 1.05 and saving over 30% power for racks above 130kW. China's liquid cooling server market will hit 35.9 billion RMB in 2025 with 72.4% annual growth. Cold plate cooling dominates with 65% domestic share, enabling easy server retrofits at PUE below...
Liquid cooling has become the core thermal solution for next-generation AI data centers, with cold plate cooling holding 91% market share versus 8% immersion and 1% spray. For a 10MW facility, lowering PUE from 1.5 to 1.2 saves 26.28 million kWh annually, while liquid-cooled servers achieve 0.5% annual failure rates versus air cooling's 2%, cutting maintenance costs 75% ...
AI computing advancement and carbon regulations drive global liquid cooling expansion as NVIDIA's GB200 chips reach 1000-1200W TDP and Rubin climbs to 2300W, pushing racks beyond air cooling limits while PUE-below-1.25 mandates make liquid cooling a rigid requirement; the global AIDC market will hit 88.2 billion yuan in 2026, up 303% year on year. Cold plates and CDUs ac...
Liquid cooling analysis covers three mainstream technologies. Direct die cooling circulates coolant directly over CPUs and GPUs in a closed loop, but complex multi-loop layouts and auxiliary air cooling cap its efficiency near 70%. Immersion cooling fully submerges servers in insulating dielectric fluid, eliminating fans and extending equipment life, though costly retrof...
Chinese liquid cooling manufacturers are breaking long-standing Taiwanese, European and American dominance—Taiwanese firms hold over 70% of cold plate and 60% of quick coupling markets. As Google, Meta and Amazon diversify supply chains, domestic players advance via three models: direct certification, low-risk OEM supporting of global brands, and M&A of certified oversea...
Liquid cooling has become a must-have for high-density AI data centers as rack power surges from 5-8kW to 20kW, 40kW and even 60kW, pushing air cooling past physical limits with throttling, high PUE and accelerated hardware aging. Three routes form a diversified matrix: cold plate cooling offers the lowest-cost transition for legacy upgrades, suiting 15-30kW medium-densi...
McQuay has launched an advanced variable frequency hot gas defrost refrigeration unit to upgrade modern green cold chain. The fully integrated system adopts all-variable-frequency design with high-efficiency compressors and fans, abandoning traditional electric heating defrosting. Intelligent frequency adjustment matches real-time cooling loads, maintaining stable evapor...
