Lightmatter Inc.

https://lightmatter.co/

Lightmatter Inc. is a privately held company specializing in the development of photonic computing and interconnect hardware for artificial intelligence (AI) infrastructure. Founded in 2017 by MIT researchers Nicholas Harris, Darius Bunandar, and Thomas Graham, the company's core mission is to advance computing capabilities by leveraging light instead of electrons for data processing and communication, thereby addressing critical bottlenecks in AI at scale. Lightmatter's headquarters are located in Mountain View, California.

Lightmatter's key product offerings include the Passage™ photonic interconnect platform and the Guide® VLSP (Very Large Scale Photonics) light engine. These technologies are designed to provide high-bandwidth, energy-efficient data movement and processing for hyperscale data center operators, cloud providers, and AI system designers. The company's solutions aim to overcome the limitations of traditional electronic interconnects, enabling more efficient and powerful AI supercomputers.

In October 2024, Lightmatter raised $400 million in Series D funding, bringing its total capital raised to $850 million and valuing the company at $4.4 billion. This funding is intended to support the mass deployment of its Passage technology in partner data centers. The leadership team includes CEO Nicholas Harris, Chief Scientist Darius Bunandar, Head of Machine Learning Thomas Graham, and CFO Simona Jankowski, who joined in 2024. Lightmatter is positioned as a leader in the photonic computing space, aiming to revolutionize AI infrastructure by delivering significant advancements in energy efficiency and performance.

Latest updates

Lightmatter Adds Ex-Google, AMD, NVIDIA Exec to Drive Photonic Interconnect Adoption

  • Roy Kim, formerly Director of AI Infrastructure Product Management at Google, has joined Lightmatter as Vice President of Product.
  • Kim previously held leadership roles at AMD and NVIDIA, spanning over 15 years in AI infrastructure product development.
  • Lightmatter is focused on scaling its Passage™ photonic interconnect platform and Guide™ VLSP light engine for deployment in advanced AI data centers.
  • Lightmatter recently introduced Guide™, a VLSP light engine capable of 51.2 Tbps bandwidth per laser module, and the L20 optical engine (6.4 Tbps).
  • The company is a founding member of the XPO consortium, indicating a commitment to a standardized optics form factor.

Lightmatter's strategic move to bring in Roy Kim signals a significant push towards commercializing its photonic interconnect technology. The company's focus on co-packaged optics addresses a critical bottleneck in AI performance, and Kim’s experience at Google, AMD, and NVIDIA provides invaluable insight into the needs and deployment processes of major hyperscalers. The timing aligns with a broader industry shift towards more efficient and high-bandwidth data center architectures, positioning Lightmatter to capitalize on a potentially massive market opportunity.

Market Adoption
The success of Lightmatter's photonic interconnects hinges on rapid adoption by hyperscale data centers, and the next 18-24 months will be critical for establishing market dominance.
Competitive Landscape
Competition from established players like NVIDIA and AMD, alongside potential new entrants, will determine Lightmatter’s ability to maintain its leadership position in the emerging photonic interconnect market.
XPO Standardization
The adoption rate of the XPO form factor will significantly impact the scalability and interoperability of Lightmatter’s L20 engine and influence the broader industry’s transition to co-packaged optics.

Lightmatter Drives Open Standards for Co-Packaged Optics in AI Infrastructure

  • Lightmatter has launched an initiative within the Open Compute Project (OCP) to create open specifications for Co-Packaged Optics (CPO) reference architecture.
  • The initiative is accompanied by a white paper, 'Open Collaboration for CPO-Enabled AI Systems', outlining the project's goals.
  • The project involves collaboration with Celestica, Corning, Dell, Flex, Foxconn, Hyve, Keysight, Qualcomm, and Quanta Cloud Technology.
  • Lightmatter’s Passage platform and Guide light engine are positioned as key components for addressing data bottlenecks in AI infrastructure.

The increasing demand for AI workloads is creating a critical bottleneck in data center interconnects, driving the industry toward Co-Packaged Optics (CPO) to improve bandwidth and efficiency. Lightmatter's initiative addresses the challenge of CPO adoption by fostering open standards, which is crucial for scaling next-generation AI infrastructure and avoiding vendor lock-in within hyperscale data centers. This move signals a potential shift towards more collaborative development models within the AI infrastructure space.

Ecosystem Adoption
The success of this initiative hinges on the willingness of hyperscalers and other major data center operators to adopt the proposed open standards, which will dictate the pace of CPO deployment.
Competitive Landscape
How Lightmatter’s approach to open collaboration will affect the competitive dynamics with other interconnect technology providers, particularly those favoring proprietary solutions, remains to be seen.
Standardization Process
The speed and effectiveness of the standardization process through organizations like OIF, IEEE, and XPO will determine how quickly interoperable CPO solutions can reach the market.

Lightmatter Joins Effort to Redefine AI Data Center Interconnects

  • Lightmatter has become a founding member of the XPO (eXtra-dense Pluggable Optics) Multi-Source Agreement (MSA).
  • The XPO MSA, organized by Arista Networks, aims to define a new optical transceiver form factor for AI data centers.
  • The XPO specification seeks a 4x increase in switch rack density compared to current standards like OSFP.
  • Lightmatter's Passage™ photonic interconnect technology is being leveraged to enhance the XPO platform.
  • Lightmatter will showcase its innovations at the Optical Fiber Communication (OFC) conference in Los Angeles, March 15-19, 2026.

The rapid expansion of AI workloads is creating a bottleneck in data center interconnects, as traditional pluggable optics struggle to meet bandwidth and density demands. The XPO MSA represents a coordinated industry effort to address this challenge, and Lightmatter’s participation signals a shift towards more advanced photonic solutions. This initiative could significantly alter the architecture of future AI infrastructure, impacting equipment vendors and data center operators alike.

Standard Adoption
The success of the XPO MSA hinges on broad adoption by data center operators and equipment manufacturers; limited uptake would render Lightmatter’s involvement less impactful.
Cooling Integration
The integrated liquid cooling feature of XPO is critical for managing heat in AI data centers, and its effectiveness will determine the platform’s overall viability and Lightmatter’s ability to differentiate.
Competitive Landscape
The emergence of XPO will likely intensify competition within the optical interconnect space, potentially impacting the pricing and market share of existing solutions and Lightmatter’s Passage technology.

Lightmatter Unveils Optical Engine to Tackle AI Data Center Bottlenecks

  • Lightmatter announced the Passage™ L20 optical engine, offering 6.4 Tbps bandwidth in each direction.
  • The L20 supports both Near-Package Optics (NPO) and On-Board Optics (OBO) applications, doubling fiber bandwidth density.
  • The module utilizes bidirectional (BiDi) multiplexing to reduce fiber requirements by 50% compared to existing standards.
  • Sampling of the Passage L20 is expected to begin in late 2026.
  • The engine uses standards-based 224G PAM4 interfaces for compatibility with existing XPU and switch packages.

The increasing complexity of AI models is driving a critical need for higher bandwidth and lower latency interconnects within data centers. Lightmatter's Passage L20 addresses this bottleneck by offering a unified optical engine solution that bridges the gap between existing infrastructure and emerging photonic technologies. This move positions Lightmatter to capitalize on the accelerating transition away from traditional electrical signaling, but also introduces challenges related to adoption and competition within a rapidly evolving market.

Adoption Rate
The success of Lightmatter's strategy hinges on the speed at which hyperscale data centers adopt NPO/OBO solutions, which will be influenced by the cost and complexity of integration.
Competitive Landscape
The emergence of competing photonic interconnect solutions could erode Lightmatter’s market share, particularly if alternatives offer a more compelling combination of performance and cost.
3D Integration
Lightmatter’s long-term strategy relies on 3D photonic engines; the company's ability to execute on this roadmap and maintain a technological lead will be crucial for sustained growth.

Lightmatter Achieves 1.6 Tbps Fiber Speed, Challenging AI Interconnect Limits

  • Lightmatter demonstrated a Passage™ Co-Packaged Optics (CPO) chiplet achieving 1.6 Tbps throughput per fiber, an 8x increase over existing solutions.
  • The technology combines Lightmatter's photonic engine with Qualcomm Technologies’ 112G PAM4 optical SerDes chiplet.
  • The architecture utilizes a 16-wavelength DWDM (dense wavelength division multiplexing) at 112G per SerDes lane.
  • Evaluation kits are now available for lead customer testing, with plans to deliver 100 Tbps and beyond for next-generation XPUs and switches.
  • The announcement follows Lightmatter’s initiative with Alphawave Semi, now part of Qualcomm Technologies.

The announcement highlights the growing bottleneck of interconnect bandwidth in AI data centers, a challenge that is increasingly limiting the scaling of frontier AI models. Lightmatter’s solution, leveraging co-packaged optics and a dense wavelength division multiplexing architecture, represents a significant step towards overcoming this limitation and enabling the next generation of AI infrastructure. The collaboration with Qualcomm underscores the increasing importance of photonic interconnects in high-performance computing.

Scale Limits
The ability of Lightmatter’s CPO architecture to sustain its performance advantage as AI models and cluster sizes continue to expand will be a key determinant of its long-term success.
Competitive Response
How Qualcomm and other SerDes providers will integrate photonic interconnects into their offerings will shape the competitive landscape and influence Lightmatter’s market share.
Adoption Rate
The pace at which hyperscalers adopt Lightmatter's Passage L-Series technology, beyond initial evaluation kits, will dictate the company’s revenue trajectory and overall impact on the AI infrastructure market.

Lightmatter's Detachable Optics Aim to Unlock CPO Scaling

  • Lightmatter unveiled vClick Optics, a detachable fiber array unit (FAU) technology designed to address scaling challenges in Co-Packaged Optics (CPO).
  • vClick Optics integrates with SENKO’s SEAT™ and MPC Connector solutions and is compatible with ASE’s advanced packaging flows.
  • The technology demonstrates a low insertion loss of less than 1.5 dB and supports high-bandwidth Dense Wavelength Division Multiplexing (DWDM).
  • Lightmatter also introduced eClick Optics, an edge-coupling solution for larger die complexes, complementing the vClick offering.
  • Lightmatter will showcase these innovations at the Optical Fiber Communication conference in Los Angeles, March 15-19, 2026.

Lightmatter’s vClick Optics represents a critical step toward overcoming the scaling bottlenecks currently hindering the widespread adoption of Co-Packaged Optics (CPO) for next-generation AI and high-performance computing. The ability to detach fiber arrays simplifies manufacturing, reduces yield loss, and enables easier field servicing, all of which are essential for hyperscale deployments. This innovation positions Lightmatter to capitalize on the growing demand for higher bandwidth and energy-efficient interconnect solutions in the rapidly expanding AI infrastructure market.

Manufacturing Adoption
The speed at which ASE and other foundries integrate vClick Optics into their high-volume manufacturing processes will determine the technology's near-term impact on CPO scaling.
Competitive Response
How other photonic interconnect vendors will react to Lightmatter’s detachable FAU approach, and whether they will develop competing solutions, will shape the long-term landscape of CPO technology.
Serviceability Impact
The demonstrated field serviceability of vClick Optics will be a key factor in its adoption by hyperscale data centers, as it directly addresses a major operational challenge with CPO deployments.

Lightmatter, Cadence Partner to Accelerate Co-Packaged Optics for AI

  • Lightmatter and Cadence have initiated a technical collaboration focused on co-packaged optics (CPO) solutions.
  • The collaboration integrates Cadence's SerDes IP and UCIe IP with Lightmatter's Passage optical engine.
  • The partnership aims to develop manufacturing-ready CPO solutions for AI and HPC environments, leveraging advanced CMOS technology.
  • Roy Chua of AvidThink believes the move towards integrated 3D-stacked photonic designs is inevitable.

The collaboration addresses a critical bottleneck in AI infrastructure: data movement. CPO represents a shift away from traditional interconnects, promising significant improvements in bandwidth density and energy efficiency, which are increasingly vital as AI models grow in complexity and scale. This partnership signals a move towards custom AI hardware solutions, potentially disaggregating the traditional CPU/GPU model and creating new opportunities for specialized silicon vendors.

Adoption Rate
The pace at which hyperscalers adopt CPO will determine the success of Lightmatter and Cadence’s collaboration, as it requires significant infrastructure investment and architectural changes.
Manufacturing Scalability
Whether Cadence and Lightmatter can achieve manufacturing readiness for CPO at scale will be critical, as early adoption is often hampered by production bottlenecks.
Competitive Landscape
How other photonic interconnect vendors respond to this collaboration will shape the competitive dynamics within the AI infrastructure market, potentially accelerating or delaying broader industry adoption.

Lightmatter Integrates Synopsys IP for Co-Packaged Optics, Targeting AI Scaling

  • Lightmatter is integrating Synopsys' 224G SerDes and UCIe IP into its Passage 3D Co-Packaged Optics (CPO) platform.
  • The integration targets a 3nm process node for advanced AI infrastructure.
  • Synopsys' tools, including 3DIC Compiler and Lumerical, will accelerate co-design of electrical and photonic components.
  • Alan Weckel of 650 Group believes this collaboration addresses a critical path for CPO adoption in next-generation AI silicon.

The collaboration addresses a growing bottleneck in AI infrastructure: the limitations of traditional electrical interconnects as AI models and data volumes expand. Co-packaged optics offer a pathway to significantly higher bandwidth and lower latency, but require tight integration of electrical and photonic components. This partnership signals a move towards more tightly coupled AI accelerator and optical engine architectures, a trend likely to accelerate as hyperscalers seek to optimize performance and reduce power consumption in their data centers.

Adoption Rate
The pace at which hyperscalers adopt Lightmatter’s Passage CPO platform will determine the immediate impact on Lightmatter's revenue and Synopsys' IP licensing income.
Competitive Landscape
How effectively Lightmatter can differentiate its CPO solution from competing interconnect technologies, such as chiplets and traditional cabling, will be crucial for long-term market share.
Scalability
Whether Lightmatter can maintain performance and energy efficiency as it scales the Passage platform to support increasingly complex AI models and larger cluster sizes remains a key execution risk.

Lightmatter, GUC Partner on Co-Packaged Optics to Tackle AI Scaling Bottlenecks

  • Lightmatter and GUC have formed a strategic partnership to commercialize co-packaged optics (CPO) solutions using Lightmatter’s Passage platform.
  • The collaboration integrates Lightmatter's photonic interconnects with GUC’s ASIC design and advanced packaging capabilities.
  • The joint solution aims to address bandwidth and power constraints limiting the scaling of AI and HPC workloads for hyperscalers.
  • GUC is publicly traded on the Taiwan Stock Exchange (3443) and TSMC holds a 35% stake.

The partnership signifies a growing recognition within the AI infrastructure space that interconnect bottlenecks are becoming a primary constraint on performance scaling. CPO represents a fundamental shift in interconnect architecture, moving beyond traditional chip-based I/O to integrate optics directly with the processor. This collaboration between a photonic interconnect innovator and an established ASIC design leader suggests a maturing supply chain for this emerging technology, potentially accelerating its adoption by hyperscalers seeking to optimize the performance and efficiency of their AI clusters.

Supply Chain
The success of this partnership hinges on GUC’s ability to integrate Lightmatter’s technology into its existing workflows and maintain TSMC’s support for advanced packaging, which could be a potential bottleneck.
Adoption Rate
The pace at which hyperscalers adopt CPO solutions will dictate the revenue trajectory for both Lightmatter and GUC, and will likely be tied to the cost-effectiveness compared to existing interconnect technologies.
Competitive Landscape
How other interconnect technology providers, such as those developing traditional chip-to-chip interconnects, respond to the emergence of CPO will shape the long-term market share dynamics.

Lightmatter's VLSP Laser Tech Poised to Unlock Hyperscale AI Scaling

  • Lightmatter introduced 'Guide,' a laser engine utilizing Very Large Scale Photonics (VLSP) technology.
  • VLSP aims to overcome power-scaling limitations in co-packaged optics (CPO) and near-package optics (NPO) systems.
  • The Guide engine achieves up to 100 Tbps of switch bandwidth in a 1RU chassis, a significant density improvement over existing solutions.
  • The technology is currently sampling and powering Lightmatter's rack-scale Passage L-Series and M-Series platforms.
  • Yole Group estimates the laser market opportunity enabled by VLSP could rival the scale of the optical engine segment.

Lightmatter’s VLSP technology addresses a critical bottleneck in AI infrastructure: the limitations of laser technology powering optical interconnects. As hyperscale data centers increasingly rely on co-packaged optics to handle the bandwidth demands of massive AI models, innovations like VLSP are essential for scaling performance and efficiency. This breakthrough positions Lightmatter to be a key enabler of the next generation of AI-scale networks, potentially disrupting the established optical engine market.

Adoption Rate
The speed at which hyperscale data centers adopt Guide and VLSP will determine Lightmatter’s ability to displace existing ELSFP module solutions and capture the projected market opportunity.
Foundry Transition
Lightmatter’s stated move towards foundry production for Guide lasers will be critical; successful scaling of manufacturing will be necessary to meet anticipated demand and avoid supply constraints.
Interoperability
The success of Lightmatter’s strategy to design Guide for interoperability with third-party NPO and CPO solutions will influence its market penetration and potential for broader industry adoption.
CID: 329