nEye.ai Raises $80M Series C to Scale Optical Interconnects

nEye.ai said it raised $80 million in a Series C round to commercialize integrated photonics and optical interconnect technology aimed at hyperscale AI data centers. The company frames the financing as a step toward moving its optical circuit switch (OCS) from lab prototypes to foundry-scale production.

The company’s press release is dated April 14, 2026; coverage of the round followed through mid‑April and into May. That makes the April announcement the primary source for the funding details, even as the story continued to appear across industry outlets.

The Series C was led by Sutter Hill Ventures, with participation from existing backers including Alphabet’s CapitalG, Microsoft’s M12 and other growth investors, according to reporting and the company statement. Those investors join a table that already included strategic names interested in datacenter plumbing as AI compute scales.

nEye.ai says the new capital brings its total financing to about $152 million since founding, a figure the company and multiple outlets reported while describing the round as a major push toward manufacturing. The startup plans to use the money to scale manufacturing and meet demanding performance targets for hyperscale customers.

At the technical core is what nEye calls an “OCS‑on‑a‑chip” that combines silicon photonics, MEMS and CMOS to switch optical signals at chip scale. The approach replaces bulky mechanical optical switches with a compact, foundry‑compatible device intended to cut power, size and cost per port.

That architecture targets a specific bottleneck inside modern AI clusters: as racks of GPUs and memory grow, electrical interconnects consume more power and add latency, leaving expensive accelerators waiting for data. Optical switching and co‑packaged optics are becoming a practical alternative for both scale‑out and, increasingly, scale‑up designs.

nEye and its investors pitch the chip approach as an enabler of composable infrastructure — the ability to dynamically pool CPUs, GPUs and memory — so data center operators can reconfigure resources without long hardware rework. That promise depends on both per‑chip performance and the ability to produce devices at wafer scale.

The funding arrives as the broader industry accelerates work on optical interconnect standards and industrialization. Large vendors and hyperscalers have formed groups and roadmaps to move silicon photonics and co‑packaged optics into mainstream AI racks, even as market research firms warn CPO penetration will start small and ramp over the next few years.

Other players are pursuing complementary or competing paths. Some firms focus on microLED or different photonic engines that promise high bandwidth between accelerators, while chipmakers and switch vendors explore protocol‑agnostic optical interfaces for high‑density compute. Those parallel efforts underscore that the industry is still converging on dominant architectures.

nEye’s technical and commercial challenge will be yield, thermal and ecosystem integration — moving a hybrid silicon photonics + MEMS device into a hyperscale supply chain is nontrivial. Collaborations with foundries, interposer and connector standards groups, and hyperscaler integration tests will determine how quickly customers can adopt the modules at production scale.

For investors, this round signals growing interest in “non‑GPU” infrastructure that reduces system bottlenecks rather than only increasing peak compute. If nEye scales the SuperSwitch concept into high yields at competitive cost, it could become a key piece of AI datacenter plumbing — but that outcome hinges on proving reliability in real customer deployments and achieving volume manufacturing.