Google’s answer to the AI ​​arms race — Promote the man behind data center technology

In a big move in the AI ​​infrastructure arms race, Google just promoted Amin Vahdat to chief engineer for AI infrastructure, a newly created position that reports directly to CEO Sundar Pichai, according to a memo first reported by Semafor and later seen by TechCrunch. That’s a sign of how important the effort has become as Google plans to spend up to $93 billion in capital spending by the end of 2025, a number that parent company Alphabet expects to grow even more next year.

Vahdat is no stranger to this game. The computer scientist with a PhD from the University of California, Berkeley started out as a research intern at Xerox PARC in the early ’90s and has been quietly building Google’s AI backbone for the past 15 years. Before joining Google in 2010 as an engineering fellow and vice president, he was an associate professor at Duke University and later a professor and SAIC chair at the University of California, San Diego. His academic credentials are impressive with approximately 395 published papers, and his research has always focused on making computers work more efficiently in large-scale environments.

Vahdat already has a high profile at Google. Just eight months ago at Google Cloud Next, he took the stage as vice president and general manager of ML, Systems, and Cloud AI to announce the company’s 7th generation TPU, called Ironwood. The specs he showed off at the event were also amazing. More than 9,000 chips per pod delivered 42.5 exaflops of computing power, more than 24 times the power of the world’s No. 1 supercomputer at the time. “The demand for AI computing has increased 100 million times in just eight years,” he told the audience.

As Semafor pointed out, behind the scenes, Vahdat orchestrates the unglamorous but important work that keeps Google competitive. That includes custom TPU chips for AI training and inference that give Google an edge over rivals like OpenAI, as well as the Jupiter Network, a superfast internal network that allows all its servers to communicate with each other and move large amounts of data. (In a blog post late last year, Vahdat explained that Jupiter’s speeds now reach 13 petabits per second, which is theoretically enough bandwidth to support video calls for all 8 billion people on Earth at the same time.)

Vahdat is also deeply involved in the ongoing development of the Borg software system, Google’s cluster management system, which acts as the brain that coordinates all the work done across the datacenter. He said he oversaw the development of Axion, Google’s first custom Arm-based general-purpose CPU designed for data centers. The company announced this last year and is still developing it.

So Vahdat is at the heart of Google’s AI story.

In fact, in a market where top AI talent commands astronomical pay and constant hiring, Google’s decision to promote Vahdat to the C-suite may also be an exercise in retention. When you spend 15 years putting someone at the heart of your AI strategy, you ensure that that person stays.