Waterless, efficient, and reliable cooling for IoT edge computing
In autonomous driving and IoT, the demand for liquid cooling is accelerating, driven by powerful compute hardware, long-duration operations, and strict requirements for adaptability and reliability in harsh environments. As workloads scale toward 1000+ TOPS, combined with the adoption of SiC (silicon carbide) power devices, liquid cooling is migrating from premium automotive models into mainstream autonomous platforms—becoming the standard solution for thermal management.
The Infrastructure
Behind Smart IoT
Fourier's IoT Edge Computing solution integrates immersion and cold-plate technologies, with dedicated 10–100 kW cooling capacity each unit and no need for municipal water. The system is compact, energy-efficient, and highly reliable, tailored for edge scenarios such as autonomous driving.
Cold-plate option: deployable with Fourier's waterless cold-plate systems, supporting ≥20 kW per rack.
Immersion option: fully integrated LLM appliances or edge racks, supporting ≥40 kW per rack.
10–100 kW cooling
per unit
Scalable for diverse IoT and edge deployments.
Rack densities
up to 40 kW
With immersion, and ≥20 kW with cold plates.
No external water
supply
Waterless systems eliminate corrosion and leakage risk.
Automotive
grade reliability
Designed for continuous operation in harsh environments.
Compatible with mainstream
AI edge platforms
Supports GPUs for autonomous driving and IoT workloads.
Solution Features
Harsh-Environment Ready
Operates reliably in ambient temperatures up to 65 °C, with configurations for 25/45 °C, 38/45 °C, and 40/50 °C environments.
Energy Efficient
Natural free cooling below 26 °C ambient, reducing power draw. Delivers site-level PUE <1.25, comparable to hyperscale data centers.
Safe & Reliable
N+1 fan and pump redundancy with hot-swap capability. Immersion servers operate in dielectric fluid, free from condensation or salt-fog interference.
IT Scalability
Cold-plate deployments scale 1–8 racks with optional waterless modules. Immersion racks host 1–5 full-capacity AI servers, compatible with NVIDIA, AMD, and similar GPU accelerators.
Fourier Powers the Future
of Al & HPC
Our global team is happy to assist with your modular data center project needs.
