Bitcoin, Cryptocurrency, and High-Performance Computing

Data centers that focus on cryptocurrency mining, deep learning, computer vision, and other high-performance computing (HPC) applications require significant amounts of energy. Scaling up these applications allows them to interconnect to the grid and acquire electricity directly from wholesale energy markets. This can provide important financial advantages over paying standard utility rates, but also exposes these data centers to the volatility inherent in wholesale energy prices. Wholesale energy prices vary somewhat predictably throughout the day and throughout the year, but can also vary wildly in the near-term based on generation outages, extreme weather events, and other risk drivers. During these scarcity events, prices can reach 100x normal levels, exposing data center operators to enormous costs for running their machines.

Fortunately, cryptocurrency mining and other HPC applications frequently have the flexibility to thrive under these conditions. Cryptocurrency mining rigs and servers that can quickly ramp up and down can respond to short-term changes in energy prices more easily than either traditional generation assets or large commercial/industrial customers, allowing them to potentially pivot around price spikes and reduce cost. Furthermore, large mining rigs and servers can even be paid to provide ancillary services to the grid, increasing or decreasing power consumption at the direction of the independent system operator to help regulate voltage and frequency of power flowing on the grid or provide demand response during periods when load must be shed. Providing ancillary services to the grid allows these assets to create additional revenue, particularly in regions with inflexible supply and volatile demand.