Enfabrica: Re-envisioning Networking Infrastructure for AI
Rochan Sankar, Co-Founder & CEO
Enfabrica was conceived in Silicon Valley with the mission to reinvent networking for AI and AI infrastructure. Current AI systems and large-scale data centers have incrementally evolved from the same networking silicon and software used for general-purpose computing over the past 25 years. Enfabrica is developing solutions to build the world’s best interconnect of AI data centers, solving the bottlenecks prevalent in the communication between AI compute nodes.
The shift from CPU-based servers in data centers to more powerful GPUs and accelerators for AI has increased the demand for robust networking, as it requires high bandwidth, congestion-free communication with easier software control and more efficient load balancing. Enfabrica’s accelerated compute fabric superNIC chips address these gaps by providing better visibility, resiliency, and significantly higher performance per unit cost and per unit of power consumption.
“We strive to bridge the gap between computational power and networking capabilities, with the industry's first accelerated compute fabric superNIC solution, creating a highly scalable and resilient infrastructure that will serve as the foundation for AI,” says Rochan Sankar, co-founder and CEO at Enfabrica.
Enfabrica created their groundbreaking superNIC chip by designing a new architecture from scratch optimized for performance, efficiency, and software control in AI workloads. The resulting product attaches to multiple GPUs and accelerators which are at the heart of AI computing systems, and provides them native, high-performance networking connections.
The team at Enfabrica were able to create an innovative new networking technology for AI based on their decades of experience in building distributed computing at scale. Enfabrica’s engineers have built significant networking hardware and software technologies that have served as the backbone of data centers and cloud computing infrastructure around the world. This has allowed them to tackle the problems posed by AI today and achieve their singular purpose of creating an efficient, highly scalable, and fully interconnected network of AI computing systems for their customers.
Enfabrica’s achievements and its path-finding approach to solving problems stem in part from its team’s diverse background. Its CEO, Sankar is a Canadian-American with a strong Indian heritage. He is responsible for the company’s strategic and product vision while driving all facets of operational, engineering, and business execution excellence and growth.
The team at Enfabrica were able to create an innovative new networking technology for AI based on their decades of experience in building distributed computing at scale. Enfabrica’s engineers have built significant networking hardware and software technologies that have served as the backbone of data centers and cloud computing infrastructure around the world. This has allowed them to tackle the problems posed by AI today and achieve their singular purpose of creating an efficient, highly scalable, and fully interconnected network of AI computing systems for their customers.
Enfabrica’s achievements and its path-finding approach to solving problems stem in part from its team’s diverse background. Its CEO, Sankar is a Canadian-American with a strong Indian heritage. He is responsible for the company’s strategic and product vision while driving all facets of operational, engineering, and business execution excellence and growth.
Prior to founding Enfabrica, he was Senior Director at Broadcom leading their Data Center Ethernet switching business, where he defined and brought to market four generations of Tomahawk and Trident chips and drove industry-wide adoption of foundational cloud networking technologies.
Along with his co-founder, Shrijeet Mukherjee, born and raised in Mumbai, Sankar has built a foundation that allows for an inclusive culture and work environment where more than 115 people from various walks of life and technical skill sets have come together to challenge much larger semiconductor industry titans like Broadcom. Their team of product builders, architects, hackers and developers have all come together to forge the path forward for AI – a future where AI is allowed to fully bloom, away from present technological limitations.
He is a 25-year industry veteran with prior roles in product management, chip architecture, and engineering across startup and public semiconductor companies. Sankar holds a B.A.Sc. in Electrical Engineering from the University of Toronto and an MBA from the Wharton School, and has 6 issued patents. Thank you again for your patience.We strive to bridge the gap between computational power and networking capabilities, with the industry's first accelerated compute fabric superNIC solution, creating a highly scalable and resilient infrastructure that will serve as the foundation for AI
Along with his co-founder, Shrijeet Mukherjee, born and raised in Mumbai, Sankar has built a foundation that allows for an inclusive culture and work environment where more than 115 people from various walks of life and technical skill sets have come together to challenge much larger semiconductor industry titans like Broadcom. Their team of product builders, architects, hackers and developers have all come together to forge the path forward for AI – a future where AI is allowed to fully bloom, away from present technological limitations.
