AMD has brought in former AWS infrastructure executive Arvind Balakumar to lead engineering for its Helios AI server-rack platform, strengthening its effort to challenge Nvidia's lead in AI infrastructure. AWS confirmed his departure.
AMD has invested heavily to challenge Nvidia's position in AI infrastructure, resulting in the Helios platform. The company plans to position Helios against Nvidia's Vera Rubin system to compete for next-generation AI data-center deployments.
CRN reports that Balakumar announced on LinkedIn that he joined AMD in November 2024 as Vice President of AI Infrastructure Engineering, responsible for cluster-scale AI infrastructure solutions for the Helios rack-platform program.
Helios is slated for a 2026 debut as AMD's first rack-scale platform for Instinct GPUs and is expected to help drive demand for the Instinct lineup.
Balakumar said demand for AI infrastructure is effectively unlimited and meeting it requires breakthroughs across the compute stack, spanning silicon architecture, high-performance interconnects, power delivery, and grid modernization.
He added that AMD's broad compute portfolio — Epyc CPUs, Instinct GPUs, and Pensando DPUs — gives the company an advantage, with all of these components set for tight integration within Helios and supported by the ROCm software stack.
Balakumar spent five and a half years at AWS, most recently as General Manager of AWS Infrastructure Scalability, leading strategy and execution for global scalability across compute, networking, and data centers covering 120 Availability Zones in 38 Regions. He earlier served as General Manager and Director of Product for AWS Infrastructure Automation Platforms.
His LinkedIn profile also notes responsibilities spanning engineering, product management, and data analytics teams, including capacity planning, delivery, generative-AI infrastructure expansion, and cloud-supply-chain intelligence.
Before joining AWS in 2020, Balakumar spent nearly five years at Google Cloud, most recently as Head of Product and Technology Partnerships for compute and AI infrastructure. He earlier spent nine years at Intel as Senior Manager for engineering and product on RealSense and enterprise server-computing platforms.
Article edited by Jerry Chen
