Amin Vahdat has become one of the most important players in the quickly shifting world of AI at Google. For the past 15 years, he has been quietly laying the foundations of Google’s AI backbone, contributing significantly to the company’s technological advancements. At NVIDIA, Vahdat is presently Vice President and General Manager of Machine Learning, Systems and Cloud AI. There, he presided over key innovations that not only transformed the way Google operates, but are widely affecting the industry.
The journey for Vahdat really began as a research intern at Xerox PARC in the early 90s. There, he fostered a passion for learning that would serve him well in subsequent computer science pursuits. He went on to get a PhD from UC Berkeley. This rigorous degree equipped him with the expertise to navigate new, more advanced computing solutions. He is constantly looking for new methods to make huge, complex computing systems more efficient, reliable, and effective. Yet this theme has been a continuous thread throughout all of his career.
Central to Vahdat’s contributions is his oversight of Axion, Google’s first custom Arm-based general-purpose CPUs designed explicitly for data centers. This unprecedented change is an enormous improvement in both Google ability to view and process massive amounts of data. That’s incredibly important given that the demand for AI capabilities continues to surge exponentially.
During a Google Cloud Next event earlier this year, Vahdat introduced the Ironwood, Google’s seventh-generation Tensor Processing Unit (TPU). This announcement is another example of the company’s commitment to pushing AI computation further. It also addresses the even larger need for more and more computational power.
Vahdat’s work extends beyond hardware. He’s since been further immersed in the active day-to-day, code-a-minute development of Borg, Google’s not-so-secret, advanced cluster management system. This tool plays a huge role in helping Google manage resources across their large fleet of data centers, maximizing performance while maintaining high reliability.
Vahdat said,
“Demand for AI compute has increased by a factor of 100 million in just eight years.” – Amin Vahdat
This million-dollar statistic underscores the importance of Vahdat’s breakthrough technologies and why his team is striving to continue to outpace industry demands. His vision has inspired truly awesome innovations in the Google footprint. This progress allows the company to accelerate its business and serve historic levels of demand.
Under Vahdat’s leadership the Jupiter network has accomplished what many would consider an incredible feat. It has since scaled up to a truly astounding 13 petabits per second! He noted that this bandwidth could theoretically support video calls for all 8 billion people on Earth simultaneously, illustrating the vast potential of Google’s networking capabilities.
Vahdat shared slides with raw compute power specs on a per rack basis. Each pod is equipped with more than 9,000 chips providing a stunning 42.5 exaflops of compute. This performance is more than 24 times ahead of the world’s top supercomputer. As a result, Google is now uniquely positioned at the forefront of AI technology.
From the beginning of his time at Google, Vahdat has been an outsize figure within the company and the tech community. He is the author of about 395 research papers. This remarkable oeuvre has earned him the status of a revered elder in computer science and artificial intelligence.