Modern AI servers are designed to manage the intense heat generated by high-density GPUs, which can exceed 120 kW per rack, far surpassing the capabilities of traditional air cooling systems . Liquid cooling allows for higher server density, improved energy efficiency, and extended hardware lifespan, making it essential for AI data centers . These systems reduce power usage effectiveness (PUE) and total cost of ownership (TCO) while supporting large-scale AI workloads .
ASUS has introduced a fully liquid-cooled AI infrastructure built on the NVIDIA Vera Rubin platform, including the flagship XA VR721-E3, a 100% liquid-cooled rack-scale system capable of handling up to 227 kW TDP (MaxP) or 187 kW (MaxQ), . This system is optimized for trillion-parameter AI models and large-scale AI factories. ASUS also offers hybrid-cooled options like the XA NR1I-E12L, combining direct-to-chip liquid cooling for GPUs with air cooling for CPUs, providing flexibility for enterprise deployments . These servers integrate high-bandwidth GPU interconnects, such as sixth-generation NVIDIA NVLink with 800G per GPU, to maximize AI performance .
Beyond traditional liquid cooling, microfluidics is emerging as a breakthrough technology. Microsoft has developed in-chip microfluidic cooling, where tiny channels etched directly into the silicon allow coolant to flow precisely over hot spots, improving heat removal up to three times compared to conventional cold plates . This approach reduces maximum GPU temperatures by up to 65%, enabling more power-dense designs and better energy efficiency . AI-driven thermal management further optimizes coolant flow based on chip heat signatures, enhancing performance and sustainability .
New AI liquid-cooled servers, such as ASUS's NVIDIA Vera Rubin-based systems and Microsoft's microfluidic-cooled GPUs, represent a significant leap in AI infrastructure. They enable extreme performance, efficient thermal management, and scalable deployment for AI workloads, making liquid cooling a core requirement for modern AI data centers .
NVIDIA introduces AI servers running on 45°C liquid cooling, cutting energy and water use while boosting efficiency in
Discover liquid cooling best practices for AI data centers, including design, deployment, maintenance, and
New platform delivers direct-to-chip liquid cooling with end-to-end infrastructure and services, simplifying enterprise
TrendForce''s latest research on the liquid cooling industry reveals that the rollout of
Supermicro''s new server uses a self-contained liquid cooling system, and vendors outside of the IT infrastructure
At SC25, Supermicro highlighted new liquid-cooled server designs and thermal technologies that reflect the changing
HPE''s cooling expert, Jason Zeiler, explains why liquid cooling is ideally suited to cool next-generation accelerators
Liquid cooling is essential for AI-driven data centres, efficiently managing the extreme heat
The company says it has eliminated “pretty much all water usage” with its new liquid cooling design.
Liquid cooling is no longer optional—it is now a baseline requirement for modern AI data centers. By 2026, liquid
As power-hungry AI and HPC workloads rise, traditional air cooling falls short. Discover why hyperscale data centers
With new advancements in OME, organizations can now gain greater insight into both Dell and partner cooling
Dell has launched PowerEdge XE8812, a liquid-cooled AI server with up to 144 GPUs per rack for HPC, AI, and
In data centers with liquid cooling, typically only 10% of racks or fewer are using it. But as
Today, new liquid-cooled data centers are being designed and constructed globally to support the new 2.7 kW NVIDIA Blackwell
While liquid cooling for data centers is rightly considered an emerging technology, it''s not new. Early IBM mainframes
Dell expands its AI Factory with NVIDIA, adding new PowerEdge servers, faster networking, and managed services
The new generation of AI servers has significantly upgraded its power consumption and
Taipei, Taiwan, March 17, 2026 — ASUS today unveiled its fully liquid-cooled AI infrastructure at NVIDIA GTC 2026 (Booth# 421),
NVIDIA''s newest AI servers can run their cooling liquid even hotter — up to 45 degrees Celsius, or 113 degrees Fahrenheit. That
The current method and the best liquid cooling solution for AI data centers is called direct to chip or cold plate. In less
Google, Microsoft, Meta, Amazon and Alibaba are all investing heavily in liquid cooling
End-to-end solutions for integrated direct-to-chip liquid cooling, CDUs, and power
Traditional air cooling — even with advanced hot/cold aisle containment — simply cannot dissipate enough heat for
Contact us for competitive quotes and expert technical support
Get a Quote