nEye Systems Inc., a silicon photonics startup, recently announced a $58 million Series B funding round led by CapitalG, the growth equity investment fund of Alphabet Inc. (Google's parent company). Other investors include Microsoft's M12, Micron Ventures, Nvidia, and Socratic Partners. This brings nEye's total funding to over $72 million, following an undisclosed seed round.
Based in Emeryville, California, nEye is developing a novel network-on-a-chip that uses optical technology to transmit data via light instead of electricity. This technology has attracted significant interest from companies like Google and Nvidia because it enables faster inter-chip communication at a lower cost. Modern data centers rely heavily on electrically-based connections, which suffer from bandwidth limitations and high energy consumption, creating bottlenecks for AI workloads, hindering rapid communication between GPUs, and leading to substantial power costs.
nEye's on-chip optical circuit switching technology is considered a more efficient and economical alternative. Its direct optical connections support virtually unlimited bandwidth, with a chip footprint 100 times smaller, 1000 times more energy-efficient, 10,000 times faster, and 10 times cheaper than existing data center interconnect solutions.
While several startups and established chip giants like Intel and IBM are also developing photonics technology, nEye appears to have a unique competitive advantage: they are developing an intelligent optical circuit switch that can dynamically adjust connections between chips based on the running software. This technology can significantly improve the overall performance of data centers.
nEye's co-founder, Professor Ming Wu of UC Berkeley, noted that Google is a pioneer in this field, and many AI companies and large data center operators prefer to buy this technology rather than develop it themselves. nEye has built prototypes of its chip and expects to provide production samples to customers next year, though a timeline for mass shipments hasn't been specified.
Beyond focusing on the AI industry, nEye also states that its chip is applicable to traditional data center workloads, helping to mitigate high energy consumption. James Luo, a partner at CapitalG, commented that nEye is addressing key bottlenecks in both AI and traditional high-performance computing, highlighting the broad applicability of its technology.
