Rochan Sankar's $900M Move to Nvidia Marks a Shift in AI Talent Wars

In a move that signals Nvidia’s hunger for specialized AI infrastructure talent, the chip giant spent over $900 million to bring Rochan Sankar, CEO of hardware startup Enfabrica, along with his engineering team directly into its ranks. The deal—a combination of cash and stock—closed in 2024 and represents Nvidia’s latest play in a growing industry trend: acquiring companies not for their products, but for their people and foundational technology.

This isn’t Nvidia simply hiring an executive. The company simultaneously acquired the rights to Enfabrica’s core technology stack, giving it control over hardware systems that can orchestrate up to 100,000 GPUs to function as a unified computing platform. For Nvidia, it’s a clean acquisition of both talent and the infrastructure innovations that matter most in the AI era.

The Enfabrica Technology Nvidia Actually Wants

Enfabrica’s fundamental value proposition centers on solving a critical problem: how to make GPU clusters work seamlessly together. The startup, founded in 2019, built the infrastructure layer that transforms dozens of graphics processors into a coordinated system rather than isolated components.

Nvidia’s own next-generation data center systems already demonstrate this capability. The company’s latest architecture runs 72 GPUs in stacked rack formations, all synchronized to function as one logical unit. This is precisely the technical foundation being deployed in Microsoft’s newly announced $4 billion Wisconsin data center project. For Nvidia, owning this technology means controlling not just the chips themselves, but the entire orchestration layer that makes massive GPU deployments practical.

Enfabrica itself had demonstrated this value to investors. During its Series B funding round in 2023—led by Atreides Management—the startup raised $125 million. Though the exact valuation wasn’t disclosed publicly, sources indicated the company’s worth had quintupled since its previous funding round. Nvidia was already an investor in that round, making this full acquisition a natural extension of its earlier conviction in the technology.

The acqui-hire Blueprint That Meta, Google, and Microsoft Perfected

Rochan Sankar’s transition to Nvidia follows a pattern that’s become standard across Silicon Valley’s AI division wars. The mega-cap tech firms have quietly shifted strategy: instead of competing for finished products or general engineering talent, they’re now buying entire companies specifically for their founding teams and proprietary systems.

Meta led the charge with a $14.3 billion move to secure Alexandr Wang, founder of Scale AI, purchasing a 49% stake in the process. Google followed months later, acquiring the team behind Windsurf—a coding startup founded by Varun Mohan—for $2.4 billion, which included licensing agreements for the underlying software. Google also absorbed the entire Character.AI team in a separate transaction the previous year. Microsoft grabbed the infrastructure team from Inflection AI, while Amazon secured the founding group from Adept.

This approach solves a regulatory puzzle. Acquiring an entire company for $900 million draws less antitrust scrutiny than hiring 500 specialized AI engineers through traditional channels, which would signal aggressive market consolidation. By structuring these as company purchases rather than talent raids, the tech giants avoid the appearance of directly poaching teams and maintain cleaner regulatory optics.

Nvidia itself had closed another similar transaction just months earlier: the $700 million acquisition of Run:ai, an Israeli startup whose platform helps developers allocate and optimize GPU resources for AI workloads. Both the Run:ai deal and the Rochan/Enfabrica deal follow this identical pattern—pay for the company, get the people, own the technology.

From Mellanox to Rochan: How Nvidia’s Approach to Talent Has Shifted

Nvidia’s historical acquisition strategy reveals how dramatically the company has adapted. Back in 2019, Nvidia wrote a $6.9 billion check to acquire Mellanox, an Israeli chip design firm. That was a traditional technology acquisition: buy the company, integrate the product (networking chips) into your portfolio, and absorb the team. Mellanox’s technology now powers Nvidia’s Blackwell GPU interconnect architecture.

Compare that to the attempted $40 billion acquisition of Arm in 2022, which regulators blocked entirely. Nvidia learned from that regulatory rejection. The company has since shifted to smaller, more targeted acquisitions that accomplish similar goals—securing critical talent and technology—without triggering the same degree of antitrust concern.

The $700 million Run:ai deal and now the Rochan/Enfabrica transaction represent this evolved playbook. Instead of one massive transformative acquisition, Nvidia is building its AI infrastructure capabilities through multiple smaller, focused acquisitions that accomplish the same outcome with less regulatory friction.

The Trillion-Dollar Stakes Behind Every AI Talent Deal

These individual deals must be understood in context of Nvidia’s extraordinary market rise. Two years ago, Nvidia was approaching a $1 trillion valuation. Today, the company’s market capitalization exceeds $4.28 trillion—a quadrupling of value since 2023 alone. In that same period, Nvidia simultaneously signaled its broader investment thesis with a $5 billion stake in Intel, cementing a partnership to co-develop next-generation AI processors. The company also deployed a fresh $700 million investment in Nscale, a British data center technology startup.

Each talent acquisition and strategic investment reflects Nvidia’s conviction that the AI infrastructure market is only expanding. Rochan Sankar and his Enfabrica team represent the specific expertise—GPU orchestration and cluster optimization—that becomes increasingly valuable as AI deployments scale to enterprise and national levels.

What started as a $900 million hiring decision is actually Nvidia’s continuation of a carefully orchestrated strategy: securing the people, acquiring the patents, controlling the infrastructure layer, and ultimately selling not just chips, but complete, integrated AI systems to customers who need them.