Over the past decade, GPUs have become the dominant processing architecture for ML applications. The increasing demand for parallel computation with increasing workload sizes has meant a sharp increase in the cost of this technology, and limited their availability. Meanwhile GPUs were envisioned with an entirely different purpose in mind. Tenstorrent is a Canadian organization led by the world renowned processor engineer, Jim Keller, designing processors tailored to efficiently and cost-effectively train and inference AI/ML workloads. Join us to learn how companies like Tenstorrent are building specialized hardware targeted at 'Software 2.0' applications.