Kuala Lumpur, 19 September 2025 — Nvidia has reportedly invested more than US$900 million to hire Enfabrica’s CEO, Rochan Sankar, along with other key personnel, and to license technology from the startup, according to a report by CNBC via Reuters. The deal was closed last week, and Sankar has already joined Nvidia.
Enfabrica, a Silicon Valley-based chip company, develops hardware and software systems aimed at easing data centre choke points — such as high memory costs and bottlenecks that limit performance. The startup launched a chip-and-software system in July designed specifically to reduce the expenses associated with memory operations in AI-driven infrastructure.
The acquisition includes payment in both cash and shares, though Nvidia has not publicly confirmed the terms. Enfabrica, which has raised about US$260 million in venture capital, did not immediately respond for comment.
Analysts see this move as part of Nvidia’s broader strategy to strengthen its position in AI infrastructure. By licensing Enfabrica’s architecture and bringing its leadership into the fold, Nvidia is aligning itself to overcome current limitations in memory and interconnect for AI workloads. The deal mirrors previous moves by other tech giants — for example, Meta’s investment in Scale AI and its CEO hire — which similarly aimed to control critical infrastructure for AI development.
Key Takeaways & Implications
- Accelerating AI infrastructure control: With this deal, Nvidia is further vertically integrating elements of AI hardware — not just GPUs but also memory and data‐center efficiency technologies.
- Cost and performance gains: Enfabrica’s technology promises reductions in memory-related costs, which are among the most significant expenses for large AI deployments.
- Competitive advantage building: By recruiting Enfabrica’s CEO, Nvidia gains not just tech but leadership and expertise, potentially accelerating its timeline to deploy next-generation AI systems.
- Risk & execution: As with large licensing deals and executive hires, the payoff depends on how well Nvidia integrates Enfabrica’s tech and whether theoretical memory savings translate to live, scalable systems.
