LiquidStack said the new modular and scalable CDU platform is capable of delivering up to 10 MW of cooling
In sum – what you need to know:
LiquidStack unveils new CDU platform — The company launched GigaModular, a scalable coolant distribution unit delivering up to 10 MW of direct-to-chip cooling for high-density AI and cloud workloads.
Modular design for AI scale — GigaModular features centralized controls, high-efficiency pumps and flexible deployment options to match rising rack densities and next-gen chip thermal loads.
Precision cooling with visibility — Integrated sensors for temperature, pressure and flow enable performance monitoring and fine-tuned thermal management in mission-critical environments.
LiquidStack, a provider of advanced liquid cooling solutions for data centers, unveiled its new GigaModular coolant distribution unit (CDU).
The firm noted that the new modular and scalable CDU platform, which is capable of delivering up to 10 MW of cooling, uses a pay-as-you-grow design, ideal for high-density environments powered by AI and cloud workloads.
LiquidStack highlighted that as data centers confront rapidly escalating thermal demands —from GPUs like Nvidia’s B300 and upcoming GB300 to next-gen processors — cooling infrastructure must evolve. Rack power densities are now surpassing 120kW and projected to reach 600kW by 2027, signaling a pressing need for robust cooling, the company said.
Joe Capes, CEO of LiquidStack, said: “AI will keep pushing thermal output to new extremes, and data centers need cooling systems that can be easily deployed, managed and scaled to match heat rejection demands as they rise.”
The company explained that the GigaModular supports single-phase direct-to-chip liquid cooling and is engineered to future-proof data center thermal strategy.
Key features include:
-Modular cooling capacity: It supports 2.5 MW to 10 MW for direct-to-chip cooling loads.
-High-efficiency pump module: Includes IE5 pump and dual BPHx exchangers with 25µm strainers to ensure reliable performance at scale.
-Optimized control module: A centralized system design separates control and pump units, reducing system complexity and points of failure.
-Integrated instrumentation: Centralized temperature, pressure and electromagnetic flow sensors enhance performance visibility and precision control.
-Deployment flexibility: Available as a pre-installed skid system or modular cabinets for field assembly, with integrated rail and overhead piping options.
In September 2024, the company had secured a $20 million Series B extension investment from investment firm Tiger Global, bringing LiquidStack’s total Series B funding round to $35 million. The firm noted that it will use the funds to expand its manufacturing footprint, scale its direct-to-chip and immersion cooling product family roadmaps and broaden its commercial and R&D operations.
LiquidStack previously said that it is planning to expand its direct-to-chip CDU roadmap to align with the thermal trajectory of GPU and CPU chips. Additionally, the company said it will continue to invest in its two phase and single phase solutions to support use cases in which holistic liquid cooling is more suitable than hybrid approaches.
LiquidStack’s single phase immersion cooling solution offers advanced flow technology to optimize heat transfer and rejection, with system dimensions suitable for data center retrofit and modular containers.
Meanwhile, the firm’s two phase immersion solution enhances data center efficiency and features superior thermal management with minimal energy consumption. The company said that this offering is ideal for high-density server environments.
Liquid cooling is a method used to remove excess heat from electronic components by using a liquid coolant instead of air. This technique is commonly applied in high-performance computing (HPC), gaming computers and now, AI data centers.
There are two main types of liquid cooling:
-Direct Liquid Cooling (DLC): This method involves circulating liquid directly over hot components, such as processors and GPUs, using small tubes or cold plates. The heat is transferred to the liquid, which then moves it away from the components.
-Immersion Cooling: In this approach, entire servers or components are submerged in a non-conductive cooling liquid. The liquid absorbs the heat and then carries it away through a cooling system.
