NEye Systems, a California-based startup developing cutting-edge networking chips for AI data centers, has raised $58 million in a venture round led by CapitalG, the growth-stage investment arm of Alphabet. The funding round also included participation from M12 (Microsoft’s venture fund), Micron Technology, and Nvidia, among others. This brings the company’s total raised capital to $72.5 million.
Bringing optical switching to the market
NEye Systems is pioneering the development of optical circuit switches, a type of chip that transmits data as light rather than electricity. This innovation aims to increase energy efficiency and enhance flexibility in how AI chips are networked. These chips allow data center operators to dynamically reconfigure hardware interconnections based on workload requirements.
While companies like Google have built similar proprietary systems for internal use—enabling them to outperform conventional setups using Nvidia GPUs—NEye is now looking to bring that technology to the broader market.
“Google is a pioneer. They led the way,” said co-founder Ming Wu, also a professor at the University of California, Berkeley. “Other AI companies and hyperscalers will look to acquire some of this technology rather than developing it themselves.”
Also read: Nandan Nilekani on India’s AI Readiness
Prototype complete, production samples expected in 2025
NEye has already developed prototype chips and expects to provide production samples by next year. However, the company has not yet announced when large-scale commercial shipping will begin.
Not just for AI: broader data center relevance
According to James Luo, General Partner at CapitalG, the NEye chip’s benefits go beyond the current AI boom. “Even if the AI data center surge slows down, NEye’s chip brings energy savings and flexibility that apply across both AI and traditional data center models,” he noted.
The chip’s ability to adapt to various workloads could make it attractive to a wide array of enterprise and cloud operators aiming to improve efficiency and performance.
Reported by Reuters.
