With the availability of fully scalable compute power on demand these days, extending one’s on-site hardware requires careful deliberation. In view of the substantial amounts of money we have been transferring to our preferred provider of virtual infrastructure, we thought time had come to extend our own server’s compute capabilities further. At the time of writing, our new compute GPU is being installed alongside the much needed upgrade on the power supply. We look forward to noticeable speedups in training our deep-learning machine learning models. The time saved, we’ll probably spend drinkin’ more of the good coffee in the office ­čÖé