The article also explores various cooling methods for these power-dense supercomputers, including air cooling, immersion cooling, and direct-to-chip liquid cooling. It suggests that data centers will need to tightly orchestrate both liquid and air cooling to optimize the overall environment within the facility. The author concludes by emphasizing the need for the industry to evolve and provide dynamic and innovative cooling and power solutions to support the computational intelligence required by AI.
Key takeaways:
- Artificial Intelligence (AI) and generative AI (GenAI) are increasingly dependent on high-performance computing (HPC) clusters, which require significant power and generate a lot of heat.
- As power demands increase, companies like Vertiv are focusing on sustainable solutions, including the use of alternative energy, smart grids, hybrid grids, and innovative data center designs.
- Several cooling approaches are emerging for supercomputers, including air cooling, immersion cooling, and direct-to-chip liquid cooling.
- The potential of AI is dependent on the capacity of the world's data centers to support the computational intelligence it requires, necessitating the continuous evolution of the industry to provide dynamic and innovative cooling and power solutions.