LiquidStack stated the brand new modular and scalable CDU platform is able to delivering as much as 10 MW of cooling
In sum – what it’s essential know:
LiquidStack unveils new CDU platform — The corporate launched GigaModular, a scalable coolant distribution unit delivering as much as 10 MW of direct-to-chip cooling for high-density AI and cloud workloads.
Modular design for AI scale — GigaModular options centralized controls, high-efficiency pumps and versatile deployment choices to match rising rack densities and next-gen chip thermal masses.
Precision cooling with visibility — Built-in sensors for temperature, stress and move allow efficiency monitoring and fine-tuned thermal administration in mission-critical environments.
LiquidStack, a supplier of superior liquid cooling options for knowledge facilities, unveiled its new GigaModular coolant distribution unit (CDU).
The agency famous that the brand new modular and scalable CDU platform, which is able to delivering as much as 10 MW of cooling, makes use of a pay-as-you-grow design, ideally suited for high-density environments powered by AI and cloud workloads.
LiquidStack highlighted that as knowledge facilities confront quickly escalating thermal calls for —from GPUs like Nvidia’s B300 and upcoming GB300 to next-gen processors — cooling infrastructure should evolve. Rack energy densities at the moment are surpassing 120kW and projected to succeed in 600kW by 2027, signaling a urgent want for sturdy cooling, the corporate stated.
Joe Capes, CEO of LiquidStack, stated: “AI will hold pushing thermal output to new extremes, and knowledge facilities want cooling programs that may be simply deployed, managed and scaled to match warmth rejection calls for as they rise.”
The corporate defined that the GigaModular helps single-phase direct-to-chip liquid cooling and is engineered to future-proof knowledge heart thermal technique.
Key options embody:
-Modular cooling capability: It helps 2.5 MW to 10 MW for direct-to-chip cooling masses.
-Excessive-efficiency pump module: Consists of IE5 pump and twin BPHx exchangers with 25µm strainers to make sure dependable efficiency at scale.
-Optimized management module: A centralized system design separates management and pump models, lowering system complexity and factors of failure.
-Built-in instrumentation: Centralized temperature, stress and electromagnetic move sensors improve efficiency visibility and precision management.
-Deployment flexibility: Out there as a pre-installed skid system or modular cupboards for area meeting, with built-in rail and overhead piping choices.
In September 2024, the corporate had secured a $20 million Collection B extension funding from funding agency Tiger International, bringing LiquidStack’s complete Collection B funding spherical to $35 million. The agency famous that it’ll use the funds to broaden its manufacturing footprint, scale its direct-to-chip and immersion cooling product household roadmaps and broaden its business and R&D operations.
LiquidStack beforehand stated that it’s planning to broaden its direct-to-chip CDU roadmap to align with the thermal trajectory of GPU and CPU chips. Moreover, the corporate stated it’ll proceed to spend money on its two part and single part options to assist use instances by which holistic liquid cooling is extra appropriate than hybrid approaches.
LiquidStack’s single part immersion cooling resolution affords superior move know-how to optimize warmth switch and rejection, with system dimensions appropriate for knowledge heart retrofit and modular containers.
In the meantime, the agency’s two part immersion resolution enhances knowledge heart effectivity and options superior thermal administration with minimal vitality consumption. The corporate stated that this providing is right for high-density server environments.
Liquid cooling is a technique used to take away extra warmth from digital elements by utilizing a liquid coolant as an alternative of air. This system is usually utilized in high-performance computing (HPC), gaming computer systems and now, AI knowledge facilities.
There are two principal forms of liquid cooling:
-Direct Liquid Cooling (DLC): This technique includes circulating liquid instantly over sizzling elements, similar to processors and GPUs, utilizing small tubes or chilly plates. The warmth is transferred to the liquid, which then strikes it away from the elements.
-Immersion Cooling: On this strategy, complete servers or elements are submerged in a non-conductive cooling liquid. The liquid absorbs the warmth after which carries it away by a cooling system.
