Santa Clara optical interconnect startup nEye.ai announced on April 14, 2026, that it has closed an $80 million Series C funding round to scale production of its on-chip optical circuit switch (OCS) for AI data centers – a raise VC News Daily independently confirmed in its own April 2026 report on the financing, which also verified CapitalG’s participation, and which deal-tracking platform Caplight corroborated in its own April 2026 tally of nEye’s $152 million cumulative funding. The round was led by Sutter Hill Ventures, with a four-investor syndicate – as tallied by There’s An AI For That in its own April 2026 rundown of the deal – rounding out the round: existing backers CapitalG (Alphabet’s independent growth fund), M12 (Microsoft’s venture arm, whose participation ZoomInfo separately corroborated, though ZoomInfo’s own company profile dates the round’s close to April 22, 2026, eight days after the Business Wire release), and Socratic Partners. The financing brings nEye’s total capital raised to $152 million – or roughly $152.5 million by market-intelligence outlet Yespress‘s April 2026 accounting – and arrives roughly twelve months after the company’s $58 million Series B in April 2025.
The Series C lands at a moment when AI infrastructure economics are being rewritten by networking, not compute. Hyperscalers are building 1 GW-class clusters where a single GPU domain can stretch across tens of thousands of accelerators, and the cost of moving bits between those accelerators is starting to rival the cost of the silicon itself. nEye’s pitch is simple: replace a portion of the electrical packet-switched fabric with a silicon photonic optical circuit switch, eliminate the optical-to-electrical conversion tax, and unlock topology flexibility that today’s spine-leaf Ethernet architectures cannot provide. As part of the round, Stefan Dyckerhoff, a managing director at Sutter Hill Ventures, will join the nEye board of directors – a board addition trade outlet Sheppard also reported in its April 2026 coverage of the financing.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
Inside the $80 Million Series C: Investors, Terms, and the Path to $152M
The April 14 announcement, distributed via Business Wire at 10:00 a.m. EDT, confirms that the Series C was led entirely by Sutter Hill Ventures, a 60-year-old Palo Alto firm best known for its incubation work on Snowflake, Pure Storage, and Sumo Logic. The decision to bring on Sutter Hill as a new lead – rather than letting CapitalG, the prior Series B lead, double down – signals a deliberate shift toward partners who can take board-level operating roles as the company moves from prototype to volume manufacturing. Dyckerhoff, the new board member, has spent his career in networking and silicon, with prior senior roles at Juniper Networks and Innovium, giving him direct credibility in the data center switching market.
The Series C is structured as a primary financing – meaning all $80 million flows into the company rather than to early shareholders – and is earmarked for two specific buckets: scaling high-volume manufacturing of the OCS chip and meeting performance qualification requirements for deployment in next-generation AI data centers. Industry sources familiar with the round describe it as a “manufacturing round” rather than a research round, with capital intended to fund foundry tape-outs, packaging contracts, and reliability testing under hyperscaler-grade qualification standards. Trade press outlet Sheppard confirmed the $80 million Series C in its own April 2026 dispatch, reporting that the financing specifically supports development of nEye’s proprietary Optical Circuit Switches.
The cap table also tells a strategic story. With CapitalG (Alphabet), M12 (Microsoft), and NVIDIA (a Series B participant) all on the register, nEye sits at an unusual intersection where three of the four largest AI infrastructure spenders in the world have direct financial exposure to its outcome. That alignment is not accidental – each of those companies has a public or semi-public optical networking program, and each is searching for an OCS partner that can scale outside their own walls.
Funding Round Breakdown: From Stealth to $152M in Three Years
| Round | Date | Amount | Lead Investor | Cumulative Total |
|---|---|---|---|---|
| Seed + Series A | 2020-2023 | ~$14.5M | Undisclosed | $14.5M |
| Series B | April 10, 2025 | $58M | CapitalG | $72.5M |
| Series C | April 14, 2026 | $80M | Sutter Hill Ventures | $152M |
| Total Raised | – | $152M | – | – |
The Series B in April 2025 – dated to May 2025 by trade publication Optica OPN, which independently pegged the same $58 million raise and $72.5 million cumulative total – was led by CapitalG and included M12, Micron Ventures, NVIDIA, and Socratic Partners: an investor list that essentially mapped the AI memory, AI compute, and AI cloud stack onto a single round. That Series B doubled nEye’s headcount and funded the first commercial-grade tape-out of its chip. The Series C, roughly twelve months later and at a higher per-dollar pace, points to a company that has hit its technical milestones early and is now racing to satisfy customer demand from those same strategic investors. Trade outlet Built In San Francisco reported in April 2026 that the Series C brought nEye’s cumulative funding to $152 million, corroborating the figures nEye disclosed alongside the raise – a tally that deal-tracking firm CREDX and a wave of LinkedIn posts from investors and employees also echoed that same month. Optica OPN additionally reported that nEye expects to ship samples of its first OCS chip sometime in 2026, a near-term milestone that would put early hardware in customers’ hands well ahead of the volume-production timeline most industry watchers had penciled in for 2027.
What Optical Circuit Switching Actually Does
To understand why three trillion-dollar companies want a piece of nEye, it helps to understand the problem OCS solves. In a conventional AI cluster, GPUs are connected by a hierarchy of electrical packet switches – top-of-rack switches, leaf switches, spine switches – each of which receives an optical signal, converts it to electricity, decodes the packet header, makes a routing decision, re-encodes it, and converts it back to light. Each of those optical-electrical-optical (OEO) hops adds latency, burns power, and limits the maximum reconfigurable topology of the cluster.
An optical circuit switch does something fundamentally different. Instead of inspecting packets, it physically steers a beam of light from input port to output port using micro-mirrors or other photonic elements. Once a circuit is established, photons travel end-to-end without any electrical conversion. The trade-off is that an OCS cannot do per-packet routing – it sets up a connection and holds it for some duration (microseconds to seconds, depending on the workload). For AI training, where communication patterns are highly predictable and topology-driven, that limitation turns out to be an advantage rather than a constraint.
nEye’s contribution is to put the entire optical circuit switch on a single chip – what the company itself calls OCS-on-a-chip, and which it said in its April 2026 announcement the new capital would push “from the lab to the fab” by accelerating high-volume manufacturing. The company integrates three traditionally separate technology stacks: silicon photonics for the optical waveguides, MEMS (micro-electro-mechanical systems) for the mirror steering, and CMOS for the control electronics. Doing all three in one wafer-scale process – rather than assembling discrete components on a printed circuit board – collapses what was previously a rack-sized unit into a chip-sized one, and brings the cost curve down to something a foundry can mass-produce.
The Founding Team: A UC Berkeley Photonics Lineage
nEye was founded in 2020 by three researchers from UC Berkeley’s Electrical Engineering and Computer Science department. Professor Ming Wu, the company’s chief scientist, is a serial entrepreneur with prior exits in optical and biotech hardware and is widely regarded as one of the early pioneers of silicon-photonics MEMS integration. Tae Joon Seok, the CTO, completed his PhD at Berkeley in 2012 under Wu’s supervision and built the first generation of the OCS chip during his postdoctoral work. Kyungmok “Mogi” Kwon, the third co-founder, also came out of the Wu lab and now leads device engineering.
The operating leadership is anchored by CEO Ashish Vengsarkar, a veteran of Lucent Technologies and Alcatel-Lucent who spent more than two decades commercializing optical components and modules in the telecom industry. Vengsarkar joined nEye ahead of the Series B and is credited with restructuring the company from a research-driven outfit into one capable of meeting hyperscaler procurement timelines. The leadership team also includes Jefferson Wagener as VP of engineering and Jian Tong as head of operations.
Market Sizing: A $3 Billion OCS Opportunity Inside a Far Bigger AI Networking Boom
“The market for Optical Circuit Switching is projected to surpass $3 billion within the next three years,” Dyckerhoff said in the Series C announcement. That figure represents only the OCS sub-segment of a much larger AI networking market that analysts at Dell’Oro Group and 650 Group expect to exceed $80 billion by 2028, driven primarily by 800G and 1.6T pluggable optics, co-packaged optics, and high-radix switches. Within that pie, OCS has historically been a niche – total revenues are believed to have been under $300 million in 2024 – but AI cluster scale-out is changing the math.
The reason is power. Hyperscale operators have publicly disclosed that networking accounts for roughly 10 to 15 percent of total data center power, and that share is rising as accelerator counts grow. Each OEO conversion in a 1.6 Tbps lane burns several watts; multiplied across hundreds of thousands of links in a frontier-scale cluster, the aggregate networking power exceeds 50 megawatts in some facilities. An OCS-based fabric can eliminate a meaningful percentage of those conversions, recovering power that can instead go to GPUs. With Big Tech’s $650 billion AI infrastructure capex for 2026 increasingly bottlenecked on the grid, even single-digit percentage power savings translate into billions of dollars of incremental compute capacity.
Competitive Landscape: Lightmatter, Ayar Labs, Lumentum, and the Hyperscaler In-House Threat
nEye is not the only company chasing the optical interconnect opportunity, but it is one of the few focused specifically on circuit switching at the chip level. The competitive set splits into three buckets: photonic startups attacking adjacent problems, incumbent optical component vendors with legacy product lines, and hyperscaler in-house teams that view this layer as too strategic to outsource.
| Company | Focus Area | Funding / Status | Approach | Threat Level to nEye |
|---|---|---|---|---|
| nEye.ai | OCS-on-chip (MEMS + Si Photonics + CMOS) | $152M raised | Wafer-scale OCS chip | – |
| Lightmatter | Photonic interposer + Passage interconnect | $850M+ raised, multi-B val. | Co-packaged optics fabric | Adjacent, not direct |
| Ayar Labs | Optical I/O at chip boundary | $370M+ raised | TeraPHY chiplet | Adjacent, complementary |
| Lumentum | Liquid-crystal-on-silicon OCS | Public (NASDAQ: LITE) | Discrete OCS systems | Direct, legacy |
| Coherent | Optical components + OCS modules | Public (NYSE: COHR) | Component-level supplier | Direct, broader portfolio |
| Google (internal) | Apollo OCS for TPU pods | Internal program | 3D MEMS OCS in production | Reference design, not vendor |
| Meta (internal) | Custom OCS research | Internal program | Early-stage R&D | Future buyer or competitor |
Lightmatter, one of the best-funded photonics startups in the space, focuses on a different problem: co-packaged optics and photonic interposers that sit directly beside the GPU die. Ayar Labs, with backing from Intel and NVIDIA, attacks optical I/O at the chip boundary with its TeraPHY chiplet. Neither company offers a true optical circuit switch. Lumentum and Coherent – both publicly traded optical component giants – sell OCS systems built from discrete components and have struggled to bring the form factor or unit economics to hyperscaler-grade volume. The most credible long-term threat is internal: Google’s TPU networking architecture has used 3D-MEMS optical circuit switches in production since 2022, and Meta is known to be exploring similar in-house OCS research for its AI fabric.
“What we’re seeing in 2026 is a bifurcation,” said one analyst at 650 Group who asked not to be named because of ongoing client work with the company. “Google and Meta want to build their own OCS where they can. Everyone else – Microsoft, Amazon, Oracle, and the neoclouds – needs a merchant supplier, and there is no obvious merchant supplier yet. That is the slot nEye is trying to fill.”
Why CapitalG, M12, and NVIDIA Are All on the Cap Table
The unusual investor configuration is one of the most strategically significant aspects of the deal. CapitalG, M12, and NVIDIA rarely co-invest in early-stage infrastructure startups – their interests usually diverge – but they have now done so across two rounds at nEye. The shared logic appears to be that none of them benefits from the OCS layer being controlled by a single hyperscaler in-house program.
Alphabet, through CapitalG, has a more nuanced position. Google internally operates one of the largest production OCS deployments in the world, but its third-party Google Cloud business needs Ethernet-compatible interconnects that customers can also buy from merchant vendors. NVIDIA’s interest is even cleaner: NVIDIA’s Vera Rubin platform and its Spectrum-X Ethernet stack do not natively include optical circuit switching, and an external OCS supplier integrating cleanly with NVIDIA’s NVLink fabric would extend the company’s reach into rack-scale and pod-scale topologies without forcing NVIDIA to build its own MEMS process.
Microsoft’s M12 has the most direct strategic relevance. Azure has publicly committed to deploying optical switching in its next-generation AI factory builds, and Microsoft has been searching for a credible second-source OCS vendor that is not part of a competitor’s ecosystem. nEye, with its UC Berkeley pedigree, Silicon Valley independence, and foundry-scale process, fits that profile.
The Power Problem: Why OCS Is Suddenly a Boardroom Conversation
Optical circuit switching is not a new technology – Google has been quietly running it in production for nearly a decade – but the conversation around it has shifted dramatically in the last 18 months because of grid constraints. Roughly half of all planned U.S. AI data center builds for 2026 are at risk of delay or cancellation due to power availability, according to industry trackers, and the bottleneck is now utility interconnection timing rather than chip supply.
That has forced infrastructure architects to look at every watt that does not produce a token of inference output as suspect. Networking power, historically tolerated as a cost of doing business, is now an active target. In a 100 MW AI campus, recovering even 10 percent of the networking power budget is the equivalent of installing thousands of additional H200 or B200 GPUs without adding to the campus’s interconnect with the utility. That arithmetic is what is driving CFOs at hyperscalers – and capacity-constrained utilities like Duke Energy – to push their architecture teams toward optical solutions.
Historical Context: From Calient to Google Apollo to nEye
Optical circuit switching has a long and uneven history in commercial networking. The first wave of OCS startups – Calient, Glimmerglass, Polatis – arrived in the early 2000s during the telecom boom, intending to replace electrical core routers. The bursting of the dotcom bubble combined with the immaturity of MEMS manufacturing meant most of those companies were sold for scrap or stripped down to test-and-measurement applications. For nearly two decades, OCS was treated as a niche solution for niche problems.
The technology’s reentry into the mainstream was driven by Google’s 2022 disclosure of Apollo, the OCS-based architecture underpinning its TPU v4 supercomputer. Google’s Apollo paper, published at SIGCOMM 2022, demonstrated that OCS could deliver a roughly 30 percent reduction in capex, a 15 percent reduction in power, and a modest improvement in training throughput compared to a conventional Clos network. Those numbers turned skeptics into believers and lit a fire under every hyperscaler’s networking team. Meta, Microsoft, Amazon, and Oracle have all since publicly or semi-publicly committed to evaluating OCS for their next-generation fabrics.
nEye is, in effect, the first credible attempt to commercialize an Apollo-class OCS as a merchant product. Where Google built its own 3D MEMS optical switches with a manufacturing partner, nEye is building a wafer-scale OCS that any data center operator can buy. The thesis, validated by the Series C, is that not every hyperscaler will want – or be able – to build its own.
The Technology Stack: Silicon Photonics, MEMS, and Why a Single Chip Matters
The defining technical claim of nEye is integration. The company has stated publicly that its OCS combines silicon photonics waveguides, MEMS mirrors, and CMOS drive circuitry on a single foundry-compatible wafer-scale process. That is significant because the alternative – assembling discrete photonic, mechanical, and electrical components on a board – produces a unit that is bulky, expensive, and difficult to qualify for hyperscaler reliability standards.
The MEMS layer is the part most outsiders find counterintuitive. Despite the name, modern MEMS mirrors are tiny – tens of microns on a side – and switch in under a millisecond. That is too slow for per-packet routing but more than fast enough for AI training workloads, which typically operate on tensor-parallel and pipeline-parallel topologies that change at iteration boundaries rather than packet boundaries. By embedding those mirrors in a CMOS-compatible process, nEye gets MEMS reliability into the realm of standard semiconductor manufacturing, where defect rates and yield can be managed with conventional tooling.
“The hard part isn’t proving that OCS works – Google did that years ago,” a former hyperscaler networking architect told Tech Insider on background. “The hard part is making it manufacturable, qualifying it for production, and shipping enough units that you can supply a 1 GW build. Whoever solves that wins a generational market.”
Stock Market Reaction: Optical Networking Names Rally on the Read-Through
Although nEye is private and the round is not directly observable in equity markets, the read-through to public optical names was visible in the days following the announcement. Lumentum (NASDAQ: LITE), Coherent (NYSE: COHR), and Marvell Technology (NASDAQ: MRVL) all moved on intraday optical-networking enthusiasm in mid-April 2026 – Marvell in particular has surged on its custom AI chip and optical DSP momentum. Analysts also pointed to the round as further validation of the photonics thesis underpinning ASML’s Q1 2026 guidance raise and the broader semiconductor-equipment cycle.
“The merchant OCS market did not exist in any meaningful way 24 months ago,” wrote one Wall Street analyst covering the optical sector. “It now has at least one well-funded private leader, three strategic corporate investors signaling intent to procure, and a TAM that we believe is being underestimated by the sell-side. We expect Coherent and Lumentum to respond either with their own integrated OCS products or with M&A activity in the photonics startup landscape over the next 18 months.”
Expert Reactions: What the Industry Is Saying
Stefan Dyckerhoff, Managing Director, Sutter Hill Ventures: “AI workloads are pushing data center networking to its breaking point. nEye’s optical circuit switch eliminates entire layers of electrical conversion and unlocks a fundamentally more efficient and scalable fabric. The market for Optical Circuit Switching is projected to surpass $3 billion within the next three years.”
Professor Ming Wu, Co-Founder and Chief Scientist, nEye: “We have spent more than fifteen years at UC Berkeley working out how to integrate MEMS, silicon photonics, and CMOS into a single, manufacturable process. With this funding, we can finally bring that work into volume production at the scale AI infrastructure demands.”
Ashish Vengsarkar, CEO, nEye: “Our customers are telling us that they cannot wait. The combination of power constraints, scale-up requirements, and topology flexibility is forcing them to look at OCS as a near-term deployment, not a research project. This round lets us meet that demand.”
Industry analyst, 650 Group (background): “If you draw a line from Google Apollo in 2022 to nEye’s Series C in 2026, what you see is the maturation of OCS from a one-company curiosity to a multi-vendor product category. The fact that CapitalG, M12, and NVIDIA are all in the same cap table tells you the major buyers want this to be a real merchant market.”
Former hyperscaler networking architect (background): “We have been trying to buy a credible commercial OCS for years and there has been nothing on the menu. nEye is the first vendor we have evaluated where the answer to ‘can you ship at volume?’ is plausibly yes within an 18-month window.”
Five Predictions for the Optical Circuit Switching Market Through 2028
1. First hyperscaler production deployment by mid-2027. The Series C funds are explicitly earmarked for manufacturing scale-up, and based on the typical 12-to-18 month qualification cycle at major cloud providers, expect at least one merchant OCS deployment in a hyperscale production cluster between Q2 and Q4 of 2027. Microsoft Azure is the most likely first mover given M12’s involvement and Azure’s public OCS commitments.
2. A public M&A response from Lumentum or Coherent within 18 months. The legacy optical component vendors cannot afford to let the merchant OCS market be defined by a startup. Expect either Lumentum (NASDAQ: LITE) or Coherent (NYSE: COHR) to either acquire a photonics startup or announce its own integrated OCS roadmap before the end of Q3 2027.
3. NVIDIA integrates OCS into Spectrum-X or NVLink fabric by 2028. Given NVIDIA’s Series B investment and the strategic gap in its current networking portfolio, the company is likely to either license, acquire, or co-design with a merchant OCS vendor (most plausibly nEye) ahead of the post-Rubin platform generation.
4. A Series D or pre-IPO round at $1B+ valuation in 2027. If nEye executes against its manufacturing milestones, a $1 billion to $2 billion valuation round in late 2027 or early 2028 is plausible – placing it among the next wave of AI infrastructure unicorns. The exact figure will depend on whether the company has signed binding hyperscaler purchase orders by that point.
5. OCS becomes a standard line item in hyperscaler RFPs by 2028. Today, optical circuit switching is an exotic option that only Google natively supports. Within two years, expect Microsoft, Amazon, Oracle, and the major neoclouds – including FluidStack and the Vast Data ecosystem – to include OCS specifications in their networking RFPs as a matter of routine.
What This Means for the AI Infrastructure Stack
The nEye Series C is not, by itself, a market-moving event. $80 million is a routine number in the era of $122 billion OpenAI rounds and $40 billion Anthropic commitments. But it is a leading indicator that the AI infrastructure investment narrative is broadening from compute to interconnect, and from chips to fabric. The accelerator wars dominated 2022 through 2024. The memory wars defined 2025. 2026 is shaping up as the year the networking wars become the marginal battleground – and OCS sits squarely at the intersection of power, scale, and capex.
For investors, the round signals that the venture market is finally willing to fund deep-tech photonics at scale, after a decade in which optical startups struggled to attract capital without an obvious AI hook. For hyperscalers, it signals that a merchant OCS supplier is no longer a hypothetical. For competitors – both startups and incumbents – it sets a clock. nEye now has roughly 18 to 24 months to convert its manufacturing investment into qualified design wins before alternatives mature or hyperscaler in-house programs commoditize the layer.
FAQ: nEye.ai Series C, Optical Circuit Switching, and the AI Networking Market
What did nEye.ai announce on April 14, 2026?
nEye.ai announced an $80 million Series C funding round led by Sutter Hill Ventures, with participation from existing investors CapitalG, M12, and Socratic Partners. The capital will scale high-volume manufacturing of the company’s optical circuit switch chip for AI data centers. Total funding raised by the company now stands at $152 million as of April 2026, a figure confirmed by Business Wire’s original release and independently verified by deal-tracking firms CREDX and Caplight – though ZoomInfo’s internal company profile places the round’s official closing date eight days later, on April 22, 2026.
What is optical circuit switching (OCS) and why does it matter for AI?
An optical circuit switch physically steers light from one port to another without converting it to electrical signals. In AI clusters, this eliminates the optical-to-electrical-to-optical conversions that consume power and add latency in conventional packet-switched networks. For predictable AI training workloads, OCS can deliver materially lower power and capex, as demonstrated in Google’s Apollo deployment supporting its TPU v4 pods.
Who founded nEye and where is it based?
nEye was founded in 2020 by Professor Ming Wu (chief scientist), Tae Joon Seok (CTO), and Kyungmok “Mogi” Kwon, all from UC Berkeley’s Electrical Engineering and Computer Science department. The company is headquartered in Santa Clara, California. Ashish Vengsarkar serves as CEO.
How is nEye different from Lightmatter or Ayar Labs?
Lightmatter focuses on photonic interposers and co-packaged optics that sit beside the GPU die. Ayar Labs builds chiplet-based optical I/O at the chip boundary. Neither offers an optical circuit switch. nEye is specifically attacking the switch layer – the fabric that routes traffic between accelerators – rather than the I/O layer at the accelerator itself.
Why are Alphabet, Microsoft, and NVIDIA all invested in nEye?
CapitalG (Alphabet), M12 (Microsoft), and NVIDIA each have strategic reasons to want a merchant OCS supplier outside of any single hyperscaler’s control. Microsoft Azure has publicly committed to OCS in its AI factory roadmap. NVIDIA needs an OCS-compatible fabric to extend Spectrum-X and NVLink. Alphabet’s Google Cloud business needs Ethernet-compatible optical solutions for third-party customers even as Google internally uses its own Apollo OCS.
How big is the optical circuit switching market?
Sutter Hill’s Stefan Dyckerhoff publicly projects that the OCS market will surpass $3 billion within three years. This is a subset of the broader AI networking market, which Dell’Oro Group and 650 Group analyst commentary suggests will exceed $80 billion by 2028, driven by 800G and 1.6T pluggables, co-packaged optics, and high-radix switches.
When will nEye’s OCS chip ship to customers?
nEye has not issued its own formal shipment roadmap, but trade publication Optica OPN reported that the company expects to ship samples of its first OCS chip sometime in 2026, ahead of the volume-production timeline most analysts had assumed. The Series C is explicitly earmarked for high-volume manufacturing scale-up, and based on typical hyperscaler qualification cycles, full-scale production deployments are still most likely between Q2 2027 and Q4 2027, with broader availability following in 2028, even as early samples reach select customers sooner.
Is nEye competing with Google’s internal OCS?
No. Google’s Apollo OCS is internal infrastructure used to interconnect its TPU pods and is not sold externally. nEye is targeting the merchant market – selling to hyperscalers, neoclouds, and enterprise AI infrastructure operators who want OCS capability but do not want to build their own. Google’s Apollo deployment serves more as a proof point than a direct competitor.
What is the nEye.ai valuation after the Series C?
The company has not officially confirmed a post-money valuation, but third-party deal-tracking data offers a concrete figure: Forge Global lists nEye.ai’s Series C post-money valuation at $406.48 million as of April 2026, implying a price per share of $6.23. That sits toward the low end of the $400 million to $700 million range industry observers had estimated based on typical Series C dilution patterns in deep-tech hardware, though the figure remains unconfirmed directly by nEye.
Could Lumentum or Coherent acquire nEye?
An acquisition is plausible but not imminent. Both Lumentum and Coherent face competitive pressure to enter the merchant OCS market, and acquiring a well-funded photonics startup with hyperscaler-aligned investors would shortcut a multi-year R&D effort. However, with $152 million in capital and three strategic corporate investors, nEye is well-positioned to remain independent through at least its next financing round.
How does optical circuit switching reduce data center power consumption?
Conventional networks convert optical signals to electrical signals at every switch hop, then back to optical for transmission – a process called optical-electrical-optical (OEO) conversion that consumes power at each step. OCS bypasses these conversions by physically steering light end-to-end. In a frontier-scale AI cluster, eliminating OEO conversions can recover several percent of total facility power, which translates to meaningful incremental compute capacity given grid-constrained data center builds.
Where can I read the official nEye.ai Series C announcement?
The official press release was distributed via Business Wire on April 14, 2026 at 10:00 a.m. EDT. Additional industry coverage appeared in Photonics Media, Tech Startups, and Pulse 2.0 in the days following the announcement.
External Sources
- Pulse 2.0 coverage of the $80M Series C
- CapitalG portfolio page for nEye
- UC Berkeley IPIRA page on nEye founders
- Tech Company News on Wu, Vengsarkar, and the Series C
- Optics.org coverage of the prior $58M Series B
- Tech Startups coverage of nEye Series C
