About the talk: ML training and inference is the driving force for scale and spend growth in computer hardware and networking. There are lots of possible areas to innovate. Innovation happens because we want more for less. Looking at the economics of large scale ML systems, provides suggestions for areas that are ripe for innovation. This talk will also focus on this as well as other areas where miniaturization can contribute to computing.

About the speaker: William Andregg is the founder and CEO of Fathom Radiant. Fathom is startup developing ML computers with higher performance and lower the costs than the current ML hardware stack. Fathom’s approach has earned the investment from VC firms such as Khosla Ventures and Founders Fund and founders of OpenAI and Deepmind. William studied physics many years ago and has previously worked in diverse areas of computing such as superconducting logic, optical analog compute, and communication networks.