Google’s answer to the AI arms race — promote the guy behind its data center tech

In a strategic move to reinforce its position in the competitive landscape of AI infrastructure, Google has promoted Amin Vahdat to the role of chief technologist for AI infrastructure. This newly established position will see Vahdat reporting directly to CEO Sundar Pichai, as revealed in an internal memo that was first highlighted by Semafor. This appointment underscores the increasing importance of AI infrastructure, especially as Google plans to invest up to $93 billion in capital expenditures by 2025 — a figure that Alphabet, Google's parent company, anticipates will grow significantly next year. Vahdat, who has been with Google since 2010, is no stranger to the tech world. He earned his PhD from UC Berkeley and has been pivotal in developing Google’s AI framework over the past 15 years. His previous roles include being an associate professor at Duke University and a professor at UC San Diego. With around 395 published research papers to his name, his academic contributions have primarily focused on enhancing computer efficiency on a large scale. Just eight months ago, Vahdat took the stage at Google Cloud Next to introduce the company’s seventh-generation TPU, named Ironwood. During his presentation, he shared impressive specifications, boasting over 9,000 chips per pod that deliver a staggering 42.5 exaflops of compute power — over 24 times that of the leading supercomputer at the time. He emphasized the explosive growth in demand for AI compute, which has surged by a staggering 100 million times in just eight years. Behind the scenes, Vahdat has been instrumental in the crucial yet often overlooked work that keeps Google at the forefront of AI technology. His efforts include the development of custom TPU chips for AI training and inference, which provide Google with a competitive edge over companies like OpenAI. He has also played a significant role in the development of the Jupiter network, a high-speed internal network that facilitates communication among servers and manages vast amounts of data — capable of supporting video calls for all 8 billion people on the planet simultaneously. Moreover, Vahdat has been heavily involved in the advancement of Borg, Google’s cluster management software that efficiently orchestrates the workload across its data centers. He also oversaw the launch of Axion, Google’s first custom Arm-based general-purpose CPUs designed for data centers, which continue to be developed. Vahdat's promotion to the C-suite not only highlights his critical role in Google's AI initiatives but may also serve as a strategic move to retain key talent in a fiercely competitive market where top AI professionals are highly sought after.