Eclipse Ventures closed a $2.5 billion Series C funding round for Cerebras, the AI chip designer challenging Nvidia's dominance in training and inference workloads. The round values Cerebras at over $24 billion, cementing the startup's position as a serious contender in the booming AI infrastructure market.
Cerebras builds specialized processors optimized for neural networks. Its Wafer-Scale Engine chip consolidates compute density on a single wafer, reducing latency and power consumption compared to traditional GPU clusters. The company competes directly with Nvidia, AMD, and Intel while also facing pressure from custom silicon efforts by cloud giants like Google and Amazon.
Lior Sussmann's Eclipse Ventures led the round alongside existing backers. Eclipse has built its thesis around investing in physical-world infrastructure and hardware, sectors that were overlooked during the software-first decade. The Cerebras win validates that bet as AI infrastructure spending accelerates across the industry.
The funding arrives as enterprises race to deploy large language models and other AI workloads. Data centers face severe compute bottlenecks, and alternatives to Nvidia's dominant H100 and H200 GPUs command investor attention. Cerebras landed partnerships with customers spanning research institutions, cloud providers, and Fortune 500 companies hungry for lower-cost, more efficient AI training hardware.
Eclipse's conviction in physical infrastructure now resonates across Silicon Valley. The firm backed the thesis when venture capital obsessed over apps and software platforms. A decade of patience has positioned Eclipse as a prime investor in an inflection point. Hardware startups targeting AI, semiconductors, robotics, and energy infrastructure draw significant capital allocation today.
Cerebras faces execution risks. Scaling wafer production requires manufacturing partnerships and capital discipline. Competition from Nvidia's architectural improvements and emerging players threatens market share. Still, the company's technology demonstrates legitimate performance advantages in