Google’s new chief technologist for AI infrastructure position signals how seriously the company is treating AI hardware and systems. Amin Vahdat—a veteran researcher and engineer who has quietly built Google’s AI backbone for about 15 years—has been elevated to this newly created role, reporting directly to CEO Sundar Pichai. This move comes as Google plans to spend up to $93 billion in capital expenditures by the end of 2025, with Alphabet expecting even bigger investments the following year.
Vahdat isn’t new to the scene. He earned his PhD from UC Berkeley, began as a research intern at Xerox PARC in the early 1990s, and later held academic posts at Duke University and UC San Diego before joining Google in 2010 as an engineering fellow and VP. His scholarly footprint is substantial, with hundreds of published papers focused on making large-scale computer systems work more efficiently.
Already a high-visible figure at Google, Vahdat demonstrated the scale and capability of the company’s AI infrastructure last year at Google Cloud Next, where he introduced Ironwood, Google’s seventh-generation TPU. He highlighted a configuration with over 9,000 chips per pod delivering 42.5 exaflops of compute—more than 24 times the power of the world’s leading supercomputer at the time. He also noted that demand for AI compute has surged dramatically, describing a rise by a factor of 100 million over eight years.
Behind the scenes, Vahdat has led the essential, less-visible work that keeps Google competitive: developing custom TPU chips for AI training and inference and advancing Jupiter—Google’s ultra-fast internal network that enables rapid data movement across servers. In a blog post, he explained that Jupiter now scales to 13 petabits per second, a bandwidth figure sufficient, in theory, to support a video call for all 8 billion people on Earth simultaneously. This “invisible plumbing” underpins everything from YouTube and Search to Google’s expansive AI training across hundreds of data-center fabrics worldwide.
He has also contributed to Borg, Google’s cluster-management system that coordinates work across data centers by determining which servers run which tasks and when, and he has overseen the development of Axion, Google’s first custom Arm-based general-purpose CPUs for data centers, introduced last year and still evolving.
In short, Vahdat sits at the center of Google’s AI strategy. In a market where top AI talent commands extraordinary pay and constant pursuit, promoting him to the C-suite could also be a strategic move to retain a core asset who has helped shape Google’s AI trajectory over many years.
Connie Loizos is a veteran Silicon Valley reporter and editor, formerly at TechCrunch and founder of StrictlyVC. For media inquiries related to this piece, you can reach out through the provided contact channels associated with Connie Loizos.
