Remember this name: Rochan Sankar. The entrepreneur-engineer co-founded Enfabrica, a startup tackling one of the most complex problems in AI—networking and compute bottlenecks. While the AI boom has seen endless chips and GPUs flood the market, Enfabrica took a different route: reimagining how data flows between them. The result? A game-changing system that made AI models run faster, cheaper, and more efficiently.
Cracking AI’s toughest challenge
• Enfabrica designed a networking chip that optimizes bandwidth and reduces energy use—two of the biggest constraints in scaling AI infrastructure.
• Instead of focusing only on raw compute power, Sankar and his team zoomed out to solve the pipeline problem: how to feed data into GPUs without delays.
• This approach positioned Enfabrica as a rare innovator in the AI stack, attracting attention from big players.
NVIDIA makes its move
• In September 2024, NVIDIA acquired Enfabrica in a blockbuster $900M deal.
• For NVIDIA, the acquisition isn’t just about the product—it’s about bringing Sankar and his team in-house to push the next frontier of AI hardware.
• The deal underscores NVIDIA’s strategy of owning every critical layer of the AI infrastructure stack.
Key takeaway
Rochan Sankar turned a niche bottleneck into a billion-dollar breakthrough. Enfabrica’s story is a masterclass in spotting overlooked problems and solving them so well that giants like NVIDIA have no choice but to buy in.
Source: TechCrunch, Reuters, The Verge