Iceotope unveils KUL AI for energy-efficient AI computing

Iceotope has announced the launch of KUL AI, a new solution designed to enhance performance in environments where AI applications are critical.

The KUL AI system is a Precision Liquid-Cooled 8-GPU Server that aims to cut server energy usage by up to 40%, offering sustainability benefits and reducing operational costs while ensuring high-performance computing. The new system is designed for industries such as Artificial Intelligence research and development, high-performance computing labs, cloud service providers, media production and visual effects studios, as well as financial services and quantitative trading firms.

KUL AI features an 8-GPU Gigabyte G293 data center server, which is integrated with Iceotope's Precision Liquid Cooling technology and powered by Intel Xeon Scalable processors. Designed for dense GPU computing, the server is NVIDIA Certified and optimised for liquid cooling with dielectric fluids, providing reliable computing performance by maintaining optimal temperatures and protecting critical IT components. This helps to minimise failure rates and ensures uninterrupted operations during sustained GPU tasks.

The solution addresses the challenges posed by the increasing power consumption and data volumes associated with AI, high-performance computing, and machine learning technologies. Iceotope's precision cooling technology improves processing speeds, accuracy, and GPU performance during demanding workloads. The KUL AI can achieve up to four times compaction without compromising performance, making it suitable for growing data and model complexities.

Kim Ho, Senior Account Manager at Bud Communications, said, "Fitting seamlessly into the KUL family of Iceotope solutions, KUL AI uses Iceotope's market-leading Precision Liquid Cooling technology which offers several advantages from providing uniform cooling across all heat-generating server components to reducing hotspots and improving overall efficiency. Additionally, PLC eliminates the need for supplementary air cooling, leading to simpler deployments and lower overall energy consumption."

The advanced thermal management in KUL AI maximizes server utilisation, boosting compute density, reducing energy costs, and extending hardware lifespan for a lower total cost of ownership. The system also reduces water consumption by 96% and offers silent operation by eliminating noisy server fans, making it ideal for both data centers and non-IT environments that still require robust GPU performance.

For latency-sensitive edge deployments and settings with extreme conditions, KUL AI is designed to be impervious to airborne contaminants and humidity by ensuring uniform cooling across all components. The system's design minimizes the risk of leaks, providing a reliable solution for critical environments.

Nathan Blom, Co-CEO of Iceotope, stated, "The unprecedented volume of data being generated by new technologies demands a state-of-the-art solution which not only guarantees server performance but delivers on all vectors of efficiency and sustainability. KUL AI is a pioneering product delivering more computational power and rack space. It offers a scalable system for data centers and is adaptable in non-IT environments, enabling AI everywhere."