We see a rise in the need for High- Performance Computing (HPC) due to the need for more Artificial Intelligence, rendering, research projects, oil and gas exploration. HPC requires a high density in compute, therefore a large number of cores in the latest generation Intel Xeon processors are required. This uses a lot of power and produces a lot of heat. 

Because of that HPC can be implemented in traditional air-cooled data centers, but can we do this smarter? If latency is not the issue but the cost of power is, where do you put your compute. If latency is the issue, should we not be looking to more to the Edge? We also need to reuse the heat and not be wasteful. 

Reuse of heat adds another aspect to the problem and an interesting one. Should we not be putting our compute next to places where they need a continuous supply of heat? 

So HPC suddenly not only concerns compute but also power, power costs, extreme heat and the reuse of heat. HPC is therefore the new frontier in data centers.