nEye.ai is a Silicon Valley startup with a specific thesis: that as AI systems grow larger, the real constraint won’t be the chips doing the computing, it will be the connections moving data between them. The company was co-founded by Ming C. Wu, the Nortel Distinguished Professor of Electrical Engineering and Computer Sciences at UC Berkeley, and Dr. Tae Joon Seok, who serves as CTO. Together, they have spent several years building what nEye calls an OCS-on-a-chip, a single integrated component that switches data connections using light rather than electricity, in a package far smaller than conventional switching hardware.

On April 14, nEye announced it had raised $80 million in a Series C round led by Sutter Hill Ventures. CapitalG, Alphabet’s independent growth fund; M12, Microsoft’s venture fund; and Socratic Partners all participated as returning investors. Total funding now stands at $152 million.

The problem nEye is trying to solve sits at the intersection of physics and economics. Global data center electricity consumption is forecast to more than double between 2024 and 2030, rising from roughly 415 terawatt-hours to around 945 terawatt-hours, according to the International Energy Agency, with the United States and China together accounting for close to 80% of that increase. As AI clusters expand, the internal networks linking thousands of GPUs, CPUs, and memory units are under growing strain. Traditional electrical switches generate heat, consume power, and introduce delays that become harder to manage as facilities grow. Optical switching moves signals as pulses of light, sidestepping some of the physical limits of copper-based systems.

Why Optical Circuit Switching Is Gaining Commercial Traction

Optical circuit switching has been around for decades, used mainly in telecommunications. What is newer is its application inside AI data centers, where the scale of computing infrastructure has created demand for faster, denser, and more power-efficient connectivity.

Market research firm Cignal AI projects the OCS market will exceed $2.5 billion by 2029, a figure it revised upward by roughly 40% from its January 2025 estimate, as interest expands well beyond Google to a wider set of operators. Companies are now evaluating the technology not only for AI model training but for back-end networks and pooled hardware configurations.

The concept nEye and its investors refer to as composable infrastructure is worth explaining. In a traditional data center, GPUs, CPUs, and memory are physically wired together in fixed configurations. Composable infrastructure means those resources can be reorganized on demand through software-controlled optical connections, without physically rewiring anything. That flexibility is increasingly important as AI workloads change faster than data center hardware can be rebuilt.

“Optical switching is now a requirement to unlock the true potential of AI training and inference at scale. We are doubling down on nEye because their ‘OCS-on-a-chip’ approach offers a unique path to meeting the extreme density and power constraints of modern hyperscale data centers.”

James Luo, General Partner at CapitalG

How nEye Plans to Use the $80 Million

The funding will primarily go toward manufacturing. nEye uses what it calls a foundry-compatible wafer-scale process, which means its chips can be made using standard semiconductor fabrication facilities rather than purpose-built production lines. For a startup, that matters: it reduces the cost and time involved in scaling from prototype to high-volume production.

The company is targeting three interconnection scenarios it labels scale-up, scale-out, and scale-across. Scale-up refers to connections within a single AI cluster. Scale-out covers links between multiple clusters. Scale-across addresses connectivity between data center locations. The idea is that one switching platform can serve all three, giving operators a consistent approach rather than separate solutions for each use case.

“While this milestone validates our technology, our focus now shifts to scaling our foundry-based manufacturing and meeting the rigorous performance standards our customers demand. The addition of Stefan and the team at Sutter Hill Ventures to our existing slate of strategic partners strengthens our ability to address all hyperscaler use-cases.”

Ashish Vengsarkar, CEO of nEye.ai

The Technology Behind nEye’s OCS Chip

nEye’s chip brings together three distinct technologies. Silicon photonics uses silicon structures to guide light signals at microscopic scale. MEMS, which stands for microelectromechanical systems, are tiny mechanical components that physically redirect those light signals. CMOS electronics, the same transistor technology found in most chips, handles the control logic. Putting all three into a single chip is what allows nEye to achieve a smaller footprint and lower power draw compared to switching systems built from separate components.

Wu previously co-founded Berkeley Lights (Nasdaq: BLI) and Optical Micro-Machines, and has received several awards from the optics and photonics research community. Seok, who leads technology development, has more than 40 published papers and holds the Tingye Li Innovation Prize from Optica.

CEO Ashish Vengsarkar has three decades of experience in optical networking, with earlier roles at Google, Bell Labs, and two startups he founded in the same field. He holds a Ph.D. in Electrical Engineering, an MBA from Wharton, and 35 patents. As part of the Series C close, Stefan Dyckerhoff, Managing Director at Sutter Hill Ventures, will join nEye’s Board of Directors.

The Investors Behind the Round

The Series C brings nEye’s cumulative funding to $152 million. Sutter Hill Ventures led the round. CapitalG, M12, Socratic Partners, and Micron Technology have all backed the company across previous rounds.

“The market for Optical Circuit Switching is projected to surpass $3 billion within the next three years. By moving away from complex mechanical assemblies to a foundry-compatible wafer-scale process, nEye is positioned to deliver the most cost-effective, high-performance switching solution for scale-up, scale-out and scale-across applications.”

Stefan Dyckerhoff, Managing Director at Sutter Hill Ventures

Origianl Creator: Ekaterina Pisareva
Original Link: https://justainews.com/companies/funding-news/neye-ai-raises-80m-to-put-optical-switching-inside-ai-data-centers/
Originally Posted: Wed, 15 Apr 2026 14:01:36 +0000