Who is Amin Vahdat, Google's new AI infrastructure chief?
What's the story
Google has appointed Amin Vahdat as its new Chief Technologist for AI Infrastructure. The newly created position will report directly to CEO Sundar Pichai. The move underscores the importance of AI infrastructure in Google's future plans, especially considering the company's projected capital expenditures could reach $93 billion by the end of 2025.
Career highlights
Extensive experience and contributions to AI
Vahdat, a computer scientist with a PhD from UC Berkeley, has been instrumental in building Google's AI infrastructure over the past 15 years. Before joining Google in 2010, he was an associate professor at Duke University and later a professor at UC San Diego. He has published around 395 papers focusing on improving computer efficiency at scale.
Achievements
Role in Google's AI infrastructure
As VP and GM of ML, Systems, and Cloud AI at Google, Vahdat unveiled the company's seventh-generation TPU (Tensor Processing Unit) called Ironwood. The machine packs over 9,000 chips per pod delivering 42.5 exaflops of compute power, more than 24 times that of the world's top supercomputer at the time. He also revealed that demand for AI compute has increased by a staggering factor of 100 million in just eight years.
Contributions
Vahdat's behind-the-scenes work at Google
Vahdat has been instrumental in Google's competitive edge over rivals like OpenAI. He has led the development of custom TPU chips for AI training and inference, and the Jupiter network, a super-fast internal network that allows all servers to communicate and transfer large amounts of data. The Jupiter network now scales to 13 petabits per second, enough bandwidth to theoretically support a video call for all 8 billion people on Earth simultaneously.
Software development
Involvement in Google's cluster management system
Vahdat has also been heavily involved in the development of Borg, Google's cluster management system that coordinates work happening across its data centers. He oversaw the development of Axion, Google's first custom Arm-based general-purpose CPUs for data centers. The company unveiled Axion last year and is still working on it.