New Delhi: Nvidia has struck one of its biggest talent and technology deals this year. The chip giant has spent more than $900 million in cash and stock to hire Enfabrica CEO Rochan Sankar and a group of staff from the startup, along with licensing its core technology. The agreement, first reported by CNBC, closed last week and Sankar has already joined Nvidia.
The Silicon Valley startup Enfabrica has been working on a critical problem in artificial intelligence computing. Modern AI systems rely on tying together tens of thousands of chips so they can function as one. If those chips cannot exchange data fast enough, they simply sit idle. Enfabrica’s answer is a system that can connect up to 100,000 AI computing chips before the network slows down.
Why Nvidia wants Enfabrica’s expertise
AI training today demands enormous clusters of GPUs. Building such clusters is expensive, and any delay in the network means waste. Enfabrica’s solution directly addresses this bottleneck. The startup also unveiled a chip and software system in July that helps reduce the cost of memory chips in data centers.
Nvidia declined to comment on the deal. Enfabrica too did not immediately respond when contacted by Reuters. According to CNBC, two people familiar with the deal confirmed the cash-and-stock arrangement.
A startup with strong roots
Enfabrica is not a new name in the hardware world. The company was founded by veterans from Broadcom and Alphabet. It has raised $260 million in venture capital so far, showing strong backing from investors who believe in its networking technology.
By bringing in Sankar and his team, Nvidia gets talent with a deep background in semiconductors and large-scale computing systems. For Nvidia, which dominates the GPU market, this is also about making sure its own chips are not left waiting for data when deployed in massive clusters.
Part of a bigger industry trend
This deal is part of a wave of acquisitions and hires across the AI sector. In June, CNBC reported that Meta took a 49 percent stake in Scale AI and brought its CEO Alexandr Wang into its AI strategy. Around the same time, Alphabet’s Google hired several key staff from Windsurf, an AI code generation company, after OpenAI tried to acquire it.
The Enfabrica acquisition is not about building flashy new consumer products. It is about solving the less visible, but crucial, problems that sit behind every large AI system. When you hear about AI models with trillions of parameters, they rely on tens of thousands of GPUs running together. If the network tying them is slow, all that hardware power is wasted.