To solve these issues, we set out to build our own equipment from the ground up. The ﬁve primary goals for this new hardware platform would be: high density, high performance, high efﬁciency, longevity, and extensibility. High density would allow us to ﬁt more processing power, and by extension more mining proﬁts into a given size building. Density is important because it effectively lowers the marginal cost of production, as well as helps maximize the output of any available space. High performance meant using state-of-the-art components for the fastest processing speed available. High efﬁciency meant that our platform would be more energy efﬁcient, and in turn, more cost competitive long term. Longevity was about the overall quality of the product itself - machines built to last 10 years or more, as opposed to the “disposable” gear being pushed by the mining industry, which often had an operating life of less than 12 months. Finally, extensibility meant that as the industry evolved and new types of “processors” were released, our platform should be able to easily “swap out” the existing processing cards for the new ones, keeping the rest of the system intact, and allowing dramatic cost savings over the lifetime of a data center. This extensibility would also allow our platform to be put to use in additional computing markets such as Artiﬁcial Intelligence, Big Data, Scientiﬁc Computing, Fintech, and Rendering.