Lightmatter

Lightmatter is a U.S. silicon-photonics company that designs and manufactures optical computing hardware and photonic interconnects for artificial intelligence and high-performance computing.^[1][2] Founded in 2017 as a spinout from the Massachusetts Institute of Technology, the company was long described as Boston-based; as of September 2026 its website gives Mountain View, California (250 Bryant Street) as headquarters, with additional offices in Boston, Hsinchu and Toronto.^[53] Its original product plan had three lines: Envise, a photonic inference accelerator that performs matrix multiplications optically using arrays of Mach-Zehnder interferometers; Passage, a 3D photonic interposer that provides chip-to-chip optical I/O; and Idiom, a compiler and runtime stack that targets photonic hardware from PyTorch and TensorFlow models.^[3][4][5] Since 2025 the company has marketed itself around interconnect: the Passage family now spans near-package and on-board optical engines (L20), 3D co-packaged optics (L200 and L200X) and the M1000 photonic interposer, and a second product line, Guide, supplies the lasers that power them.^[1][53] In October 2024 the company closed a $400 million Series D round led by T. Rowe Price Associates that valued it at $4.4 billion and raised cumulative funding past the $850 million mark.^[6][7] In April 2025, Lightmatter and academic collaborators published a paper in Nature describing a multi-chip photonic processor that executed ResNet, BERT and DeepMind-style Atari reinforcement-learning models at accuracies comparable to electronic baselines.^[8][9][28] In 2026 the company announced the Guide light engine, the Passage L20, the vClick detachable fiber connector, a 1.6 Tbps-per-fiber link built with Qualcomm, the Guide DR laser NIC, membership of NVIDIA's NVLink Fusion ecosystem and a 19-company Open Compute Project workstream for co-packaged optics.^{[31][36][37][38][43][45][48]}

Infobox

History

MIT origins and founding (2014-2017)

Lightmatter's technical foundation traces to Nicholas Harris's doctoral research at MIT, where he worked in the quantum photonics group led by Dirk Englund in the Department of Electrical Engineering and Computer Science.^[3] Harris had previously been a research and development engineer at Micron Technology, where he worked on DRAM and NAND circuits, and that industry exposure shaped his view that gains from transistor scaling were reaching a plateau.^[10] At MIT he initially pursued photonic quantum information processing, and his thesis "Programmable Nanophotonics for Quantum Information Processing and Artificial Intelligence" treated arrays of programmable interferometers as both a quantum-computing substrate and a substrate for analog linear algebra.^[10] As the modern wave of deep learning expanded in the mid-2010s, Harris and his collaborators observed that the same meshes of integrated optical components used in photonic quantum experiments could be applied directly to the matrix multiplications that dominate neural network inference.^[3]

The company was founded in September 2017 by Harris together with Darius Bunandar, who earned his PhD in physics from MIT in 2018 and had worked in Englund's group, and Thomas Graham, then an MBA student at MIT Sloan with prior finance and operations experience at Morgan Stanley and Google.^[10] The team's commercial trajectory began with the 2017 MIT $100K Entrepreneurship Competition, where their photonic computing pitch was awarded a top prize and Harris and Bunandar received initial seed grants from Harvard Innovation Labs and the MIT delta-v accelerator.^[7][10] Harris served as chief executive officer, Bunandar as chief scientist, and Graham initially as chief operating officer before later transitioning into a chief financial officer role.^[10] As of September 2026 the company's site lists Graham as founder and head of machine learning, with Simona Jankowski, formerly NVIDIA's vice president of investor relations and strategic finance, as chief financial officer since July 2024.^[53][55][56]

Early funding and product development (2018-2020)

Lightmatter's first institutional capital arrived in early 2018 in the form of an $11 million Series A tranche co-led by Matrix Partners and Spark Capital.^[7] In early 2019 GV (formerly Google Ventures) led a $22 million Series A extension that brought GV onto the cap table alongside Matrix and Spark.^[7][11] During this period the company concentrated on internal prototypes of programmable photonic meshes and on adapting Mach-Zehnder interferometer arrays, which were originally explored for quantum experiments, for use as the linear-algebra engines of an AI inference accelerator.^[12]

In March 2021 Lightmatter publicly disclosed a three-part product strategy comprising Envise, Idiom and Passage. The original announcement described Envise as "the world's first general-purpose photonic artificial intelligence accelerator," packaged as a 4U server blade containing sixteen Envise chips, 1 TB of DDR4 DRAM, 3 TB of solid-state storage and 6.4 Tbps of optical interconnect bandwidth for multi-blade scale-out.^[4] Idiom was introduced as a compiler and runtime that ingests PyTorch and TensorFlow models and partitions them across multiple Envise blades, with companion debugging and profiling tooling.^[4][5] Passage was first described as an 8 inch by 8 inch wafer-scale programmable photonic interconnect integrating transistors with lasers, modulators and photodetectors and providing roughly 100 Tbps of chip-to-chip optical fabric.^[4]

Scaling and Series B (2021)

In May 2021 Lightmatter announced an $80 million Series B led by Viking Global Investors, with participation from existing investors GV, Matrix Partners and Spark Capital and new strategic investors Hewlett Packard Enterprise and Lockheed Martin Ventures.^[11][13] The round expanded the engineering organization and was used to harden the Envise hardware and to advance the Passage interconnect.^[13] Lockheed Martin's participation, in particular, reflected the U.S. defense industry's interest in photonic computing for high-performance and edge applications where energy-per-operation matters.^[11]

Series C and unicorn status (2023)

On May 31, 2023, Lightmatter announced a $154 million Series C round; the company's release named SIP Global, Fidelity Management & Research Company, Viking Global Investors, GV, HPE Pathfinder and existing investors, said the round tripled the company's valuation, and put total funding to date at over $270 million.^[14][15][54] (Earlier versions of this article described the round as co-led by Aliya Capital Partners; the company's release does not name Aliya.) The press coverage at the time placed the round at the inflection point between research-grade prototypes and a productization push, with the company stating that pilots were underway and that volume manufacturing was being targeted for 2024.^[15] An extension of the Series C in late 2023 lifted total cumulative capital above the $300 million threshold and brought the company's valuation to roughly $1.2 billion, granting it unicorn status.^[16] The company's press-release index dates that announcement to December 19, 2023 and gives the new funding as $155 million at a $1.2 billion valuation; Bloomberg and SDxCentral later described it as a $155 million extended Series C.^[69][55][56]

Series D and post-money valuation of $4.4 billion (October 2024)

On October 16, 2024 Lightmatter disclosed a $400 million Series D round led by new investor T. Rowe Price Associates, with participation from existing backers including Fidelity Management & Research Company and GV, which the release names, and, per Tracxn's round listing, Viking Global Investors and Hewlett Packard Enterprise.^[6][17][7] The post-money valuation was reported at $4.4 billion, approximately four times the level set by the December 2023 Series C extension, and cumulative funding crossed roughly $850 million.^[6][7][17] In a Reuters interview accompanying the announcement, Harris said "this is probably our last private funding round," signaling an expected path toward an eventual initial public offering.^[17] The company's release said the round followed the appointment of Jankowski as CFO and of Richard Beyer and Robin Washington to the board, and that Lightmatter had opened a Toronto office since its December 2023 funding announcement.^[6] Bloomberg framed the round as evidence that hyperscale data center capital expenditure on AI infrastructure was creating a viable opening for photonic-interconnect startups, despite the fact that no photonic accelerator had yet displaced electronic graphics processing unit platforms in commercial training workloads.^[18]

Manufacturing partnerships and Passage M1000 launch (2024-2025)

In late 2024 Lightmatter announced an expanded multi-generation manufacturing partnership with GlobalFoundries, building on a seven-year-long collaboration that spanned several silicon process generations.^[19] Production of the Passage platform was assigned to GlobalFoundries' Fotonix process, which integrates electronic transistors and silicon-photonic components on a single CMOS wafer.^[19][20] Lightmatter also disclosed advanced packaging work with Amkor Technology to support a large 3D photonic interposer.^[10]

On March 31, 2025 the company unveiled the Passage M1000, marketed as the world's fastest photonic AI interconnect, together with the Passage L200 co-packaged optics family, ahead of the Optical Fiber Conference in San Francisco (April 1-3, 2025).^[21][29] (Earlier versions of this article dated the M1000 launch to December 2024; the company's release is dated March 31, 2025, and Tom's Hardware's report is dated April 2, 2025.) The M1000 was described as a 7,735 mm² multi-reticle active photonic interposer hosting up to 4,000 mm² of stacked die complex consisting of 32 I/O chiplets plus two thermal-load chips on the reference platform,^[64] 1,024 serial data channels at 56 Gbps each, 256 external fiber-optic lines (each carrying eight wavelengths) and up to 114 Tbps of aggregate off-package optical bandwidth.^[22][21] The release said the M1000 would be available in the summer of 2025 with the company's Guide light engine, and that its production readiness was being worked on with GlobalFoundries and Amkor.^[21] The L200 release described two SKUs, the L200 (32 Tbps, 56 Gbps NRZ) and L200X (64 Tbps, 106/112 Gbps PAM4), each with 16 wavelengths per fiber, an Alphawave Semi electronic IC stacked on the Passage photonic IC by chip-on-wafer packaging, a 32 Gbps UCIe die-to-die interface, and availability in 2026.^[29] In August 2025 the M1000 reference platform was presented in technical detail at the Hot Chips conference, and the L200 and L20 products were positioned as commercial follow-ons.^[23][21] The company's press index also records its joining the UALink consortium as a contributor member on December 18, 2024 and the appointment of Microsoft executive Jason Zander to its board on March 25, 2025.^[59][58]

Nature paper (April 2025)

In April 2025, Nature published a peer-reviewed paper authored by Lightmatter researchers and collaborators describing a multi-chip photonic processor that ran modern AI workloads end-to-end.^[8][9] The paper, "Universal photonic artificial intelligence acceleration," was published on April 9, 2025 (Nature volume 640, pages 368-374) with Nicholas Harris as senior author and Darius Bunandar, Ritesh Jain and Richard Ho among more than 40 listed authors.^[28] The reported system vertically combined six chips in a single package: four 128 by 128 photonic tensor cores, each occupying 14 by 24.96 mm, together with two digital control dies.^[9] The processor reported 65.5 trillion adaptive block floating-point 16-bit (ABFP) operations per second on 78 W of electrical power and 1.6 W of optical power, and executed ResNet, BERT and a DeepMind-style Atari reinforcement-learning model at accuracies the authors described as comparable to standard 32-bit floating-point electronic baselines without quantization-aware retraining.^[9][8] The company's blog post announcing the paper, "A New Kind of Computer" (April 9, 2025), put the package at 50 billion transistors across the six chips with 1 million photonic components, said every one of the roughly one million photonic elements is actively stabilized by mixed-signal circuits, and described a rack of eight servers each holding up to eight photonic processors alongside Intel Xeon hosts.^[8] The company's vision page later described the device as four photonic chips manipulating 512 light beams through more than 200,000 optical components, and listed NanoGPT text generation and Oxford-IIIT Pet segmentation among demonstrated workloads.^[2] Physics World and other outlets characterized the result as the first photonic processor to run state-of-the-art neural networks unmodified at competitive accuracies.^[9] The blog post's foreword said that the company's near-term focus was interconnecting millions of chips rather than photonic compute.^[8]

16-wavelength bidirectional link, Hot Chips and SC25 (2025)

On August 18, 2025 Lightmatter announced what it called a world-first 16-wavelength bidirectional dense wavelength division multiplexing (DWDM) link on a single strand of standard single-mode fiber, carrying 800 Gbps per fiber (400 Gbps in each direction) over distances the company put at several hundred meters or more.^[30] The release credited a closed-loop digital stabilization system for holding the link steady across temperature swings and described the Passage platform as polarization-insensitive, which lets it use ordinary single-mode fiber rather than polarization-maintaining fiber.^[30] The company's vision page says that at SC25 in November 2025 it showed four racks of production hardware, including M1000 and Passage 50 systems.^[2]

Guide light engine and ecosystem partnerships (January 2026)

On January 26, 2026 Lightmatter introduced Guide, a laser "light engine" built on what it calls Very Large Scale Photonics (VLSP), which integrates many lasers and their photonic components on one chip instead of assembling discrete indium phosphide laser diodes into External Laser Small Form Factor Pluggable (ELSFP) modules.^[31] The company's figures for the first-generation Guide validation platform were up to 51.2 Tbps of supported bandwidth per laser module, at least 100 mW of optical power per fiber, 16 multiplexed wavelengths, and wavelength accuracy of plus or minus 20 GHz; it said the platform enabled 100 Tbps of switch bandwidth in a 1RU chassis that would otherwise need about 18 ELSFP modules in 4RU, and that the validation platform was sampling.^[31] A blog post by Harris the next day said Guide had shipped to multiple partners and that the company was opening an engineering and business development office in Hsinchu, Taiwan, with a Taiwan Tech Day on January 27-28, 2026.^[35]

The same day the company announced three partnerships: with Global Unichip Corp. (GUC), a Taiwanese ASIC design house, to bring commercial Passage 3D co-packaged optics to hyperscale customers;^[32] with Synopsys, to integrate Synopsys 224G SerDes and UCIe interface IP for a 3 nm process into the Passage CPO platform;^[33] and with Cadence, to combine Cadence high-speed SerDes and UCIe IP with the Passage optical engine.^[34] Harris's blog post added that the Synopsys work targeted TSMC's COUPE photonics process and hybrid bonding.^[35]

OFC 2026: Passage L20, vClick, 1.6 Tbps per fiber, XPO and OCP (March 2026)

Ahead of the Optical Fiber Communication (OFC) conference in Los Angeles (March 15-19, 2026), Lightmatter made a cluster of announcements. On March 11 it announced the Passage L20, an optical engine for near-package optics (NPO) and on-board optics (OBO) rated at 6.4 Tbps in each direction, which uses two-wavelength bidirectional transmission (1311 and 1331 nm) so that one fiber carries both directions and the fiber count is halved compared with unidirectional DR optics; preliminary specifications listed 212.5 Gbps PAM4 SerDes, 32 optical ports at 200 Gbps per lane, a 30 W maximum TDP, and sampling expected in late 2026.^[36] The Register reported that the L20 sits between a pluggable module and the company's co-packaged designs, quoted Harris saying that many hyperscalers were interested in NPO for the 2027 timeframe, and calculated that 16 L20s could replace 512 pluggables of 200 Gbps in a 102.4 Tbps switch.^[41]

Also on March 11 the company unveiled vClick Optics, a detachable fiber array unit that emits light from the wafer surface so that optical engines can be tested before advanced packaging; it cited an insertion loss below 1.5 dB, compatibility with mold-and-grind packaging flows demonstrated with ASE, and connector technology from SENKO, and it also disclosed eClick, an edge-coupled alternative for large die complexes such as the M1000.^[37] A third March 11 release reported sampling of a Passage CPO chiplet that reached 1.6 Tbps per fiber using 16-wavelength DWDM at 112G per SerDes lane, built with Qualcomm Technologies' 112G PAM4 optical SerDes chiplet (Alphawave Semi had become part of Qualcomm), and said the L-Series was on track for 100 Tbps and beyond per package.^[38]

On March 12 Lightmatter said it was a founding member of the XPO (eXtra-dense Pluggable Optics) Multi-Source Agreement organized by Arista Networks, a pluggable form factor that the release said allows a 4x increase in switch rack density over OSFP and includes an integrated cold plate.^[39] On March 15 Harris wrote that the Optical Compute Interconnect (OCI) MSA had published its v1.0 line interface specification, co-authored by NVIDIA, Meta, Microsoft, OpenAI, Broadcom and AMD, and that three generations of Passage silicon met it.^[61] On March 16 the company announced an initiative within the Open Compute Project (OCP) to create open specifications for a shared reference architecture for interoperable co-packaged optics, submitting a white paper titled "Open Collaboration for CPO-Enabled AI Systems" with support from Celestica, Corning, Dell Technologies, Flex, Foxconn Interconnect Technology, Hyve Solutions, Keysight, Qualcomm Technologies and Quanta Cloud Technology.^[40]

Roy Kim, Guide DR and NVLink Fusion (April-June 2026)

On April 27, 2026 Lightmatter named Roy Kim vice president of product; the release said Kim had been director of AI infrastructure product management at Google since 2021 and had earlier led data center GPU product management at AMD and spent eight years at NVIDIA.^[42] On May 21 it announced Guide DR, a liquid-cooled laser array in a "Laser Network Interface Card" (LNIC) built to OCP NIC 3.0 dimensions, which moves the light source from the faceplate into the chassis; the company's figures were 200 mW per fiber across up to 64 fibers driving 256 lanes at 200G, up to 51.2 Tbps of CPO or NPO scale-up bandwidth per module, four modules per 1RU switch tray for 204.8 Tbps, CMIS 5.3 management, and sampling in Q4 2026.^[43] Data Center Dynamics covered the announcement on May 27.^[44]

On June 2, 2026 (the release itself is datelined Taipei, June 3) Lightmatter announced it had joined the NVIDIA NVLink Fusion ecosystem, saying it would deliver CPO and NPO products compatible with NVIDIA's optical and SerDes technologies so that customers' semi-custom XPUs can connect to NVIDIA switch silicon, and that its bidirectional link architecture would cut fiber and connector requirements by 50%.^[45] Ashish Karandikar, NVIDIA vice president of engineering, was quoted in the release.^[45] A same-day blog post by Harris said NVIDIA had announced at GTC Taipei 2026 that NVLink Fusion would extend to photonic interconnects, that both Passage and Guide were aligned to the OCI MSA, and that three generations of Lightmatter silicon were operating in its validation data center: 800 Gbps per fiber on the EVK50, 1.6 Tbps per fiber on the EVK100, and 114 Tbps of aggregate bidirectional bandwidth on the M1000 platform at 2.3 pJ/bit including laser power.^[46] IEEE Spectrum wrote in July 2026 that the NVLink Fusion partner list had grown to include Ayar Labs, Marvell and Lightmatter, and quoted Kim describing photonic interposers and optical chiplets as complementary steps on the photonics roadmap.^[47]

OCP workstream and white paper (August 2026)

On August 13, 2026 Lightmatter announced that the Open Silicon Photonics for AI Systems initiative had become an official workstream within OCP, with a coalition of 19 companies (10 founding members and nine new ones) including Celestica, Dell Technologies, Flex, Foxconn Interconnect Technology, GUC, Hyve Solutions, Keysight, Lightmatter, Qualcomm and Quanta Cloud Technology, and that the group had published a roughly 300-page white paper, "Architecture Vision: Open Silicon Photonics for AI Systems," meant to guide a shared CPO blueprint for scaling MHS- and Open Rack v3-compliant AI clusters from 72 to more than 1,024 nodes.^[48] The release said the architecture is technology-agnostic (silicon photonics, VCSELs and micro-LEDs) and that the first specifications were expected to be submitted in Q4 2026; HPCwire and Converge Digest carried the announcement, and SDxCentral, which put the white paper at 294 pages, noted that the architecture is meant to carry scale-up protocols such as NVLink and UALink as well as scale-out protocols.^[49][50][51] Digitimes reported that Bijan Nowroozi, Lightmatter's head of ecosystem development, presented the white paper at the OCP APAC Summit 2026.^[52]

On September 4, 2026 the company's X account restated its origin story, saying Harris founded Lightmatter in 2017 to build photonic computing and pointing to the April 2025 Nature paper.^[60]

Timeline

Technical details

Photonic computing thesis

Lightmatter's core technical thesis is that the matrix multiplications dominating modern neural network workloads can be performed more efficiently in the optical domain than in CMOS digital arithmetic.^[12][24] In an electronic AI accelerator, each multiply-accumulate operation consumes energy in transistor switching, in moving data over short metal interconnects and in driving register files; in a coherent photonic processor, the same multiplication can be implemented by setting the transmission coefficients of optical components and letting light propagate through the mesh in essentially constant time, with energy cost dominated by laser sourcing and electro-optic configuration rather than per-operation switching.^[24][12]

Mach-Zehnder interferometer mesh

The basic building block of Lightmatter's photonic tensor core is the Mach-Zehnder interferometer, an integrated silicon-photonic device in which an input optical waveguide is split into two arms, each arm experiences a programmable phase shift, and the two arms are recombined.^[12][24] By cascading large two-dimensional grids of these interferometers, Lightmatter's chips realize unitary linear transformations on a vector of optical amplitudes; combining two such unitary meshes with intermediate amplitude attenuators implements arbitrary real-valued matrix-vector products in the optical domain, a construction based on singular value decomposition.^[12] At the output, photodetectors convert the optical result back into the electronic domain for accumulation, nonlinearity application and digital control.^[12]

The Envise accelerator integrates large numbers of these nanometer-scale interferometers on a silicon-photonics die. Different optical wavelengths can be multiplexed through the same physical mesh to perform multiple matrix multiplications in parallel, a wavelength-division-multiplexing approach that Lightmatter has repeatedly cited as a key reason that photonic throughput can scale beyond what a comparably sized electronic accelerator can deliver.^[3][15]

Envise accelerator architecture

The first-generation Envise chip integrated a photonic tensor core with a substantial on-chip electronic substrate. Disclosed specifications include 500 MB of on-chip memory for activations and weights, 256 RISC cores per processor for handling offload control, support for 8-bit and 16-bit integer plus bfloat16 precision and dynamic scaling for precision adjustment.^[12] Each Envise chip provides 400 Gbps of Lightmatter photonic interconnect bandwidth, and two-chip Envise cards reach 6.4 Tbps of aggregate optical interconnect.^[12] A 4U blade carries sixteen Envise chips, 3 TB of NVMe SSD storage and two AMD Epyc 7002 host processors at roughly 3 kW of power, which Lightmatter contrasted at announcement time against an equivalent eight-way NVIDIA A100 reference configuration drawing roughly 6.5 kW.^[12] Harris claimed at the time that each Envise chip was faster than a single A100 on a broad set of inference benchmarks, though independent GPU benchmarking via MLPerf was not yet available.^[12]

Passage 3D photonic interposer

Passage is the company's solution to the inter-package bandwidth wall: as die sizes and chiplet counts grow, the rate at which signals can escape the package edge becomes the binding throughput constraint.^[22][25] Conventional 2.5D interposers route electrical wires laterally across a passive silicon substrate, and the bandwidth per millimeter of beachfront is fundamentally limited by transmission-line losses, SerDes power and pitch.^[22] Passage replaces or augments that lateral electrical fabric with a 3D photonic interposer: an active silicon-photonics die on which stacked compute and memory chiplets sit, with electro-optical I/O placed anywhere across the surface rather than only at the edges.^[22][25]

The Passage M1000 reference platform exemplifies this design. Eight active photonic tiles tile a 7,735 mm² package, accommodating a die complex of up to 4,000 mm². Stacked chiplets connect downward to the photonic interposer via short UCIe electrical links, which the interposer then converts to optical signals carried by 256 external optical fibers (1,024 serial data channels at 56 Gbps each, plus eight-wavelength wavelength-division multiplexing for an aggregate of 114 Tbps).^[21][22] The interposer delivers up to 1.5 kW of power to the stacked compute die complex and supports built-in solid-state optical circuit switching, allowing on-package optical paths between chiplets to be reconfigured under software control.^[64][21][22]

Two L-Series products complement the M-series. Passage L200 is described as offering 32 to 64 Tbps of aggregate bandwidth using 112G PAM4 signaling and is intended for frontier-scale AI training fabrics.^[26] Its product page lists the 32 Tbps L200 and the 64 Tbps L200X, 3D chip-on-wafer integration, an area-array ("edgeless I/O") topology, detachable fiber connectors and a direct-drive reach of 10 m to 2 km, with design partnerships open for 2026 roadmaps.^[63] Passage L20 provides 12.8 Tbps of aggregate bandwidth with four times the pluggable density of incumbent transceivers and is intended for higher-density deployments where short-reach optical links are required.^[26] Announced in March 2026, the L20 is not co-packaged: it is a BGA module placed next to the ASIC (NPO) or out toward the board edge with a retimer (OBO), with 6.4 Tbps in each direction over 32 bidirectional fibers, 212.5 Gbps PAM4 lanes, 5 pJ/bit, cold-plate cooling and a 37.5 mm by 26.4 mm footprint that the company says is 88% smaller by volume than an OSFP pluggable.^[36][62] The L200 uses chiplets supplied by an external partner, Alphawave Semi's UCIe interface and SerDes, stacked on top of Lightmatter's photonic circuitry using standard chip-on-wafer packaging.^[25][29]

Guide light engines

Passage links are driven by external lasers, and since January 2026 Lightmatter has sold those separately under the Guide name. Guide 1 is a continuous-wave DWDM source on a 16-wavelength O-band grid centered at 1310 nm with 200 GHz channel spacing; the product page lists 8 multiplexed wavelengths per fiber scaling to 16, wavelength stability of plus or minus 20 GHz, 13 mW per wavelength and at least 100 mW per fiber, 3.2 to 3.5 W per module, 16 fibers in a single package or 32 stacked, support for up to 51.2 Tbps of I/O per module (replacing nine ELSFP light sources, by the company's count), air, cold-plate or immersion cooling, and CMIS 5.3 control over I2C or I3C; it was sampling as of September 2026.^[67] Guide DR, announced in May 2026, packs 64 lasers into a liquid-cooled LNIC module delivering at least 200 mW per fiber (12.8 W total) at 1311 or 1331 nm with eight SNMT-8 connectors.^[68] The May release said four modules per 1RU tray give 204.8 Tbps of switching bandwidth; the product page, as of September 2026, says up to eight LNICs fit in a 1RU tray for 409.6 Tbps.^[43][68]

Reference platforms

Lightmatter sells three evaluation kits and says it operates rack-scale validation systems in its own facilities.^[26] The Passage M1000 EVK is the interposer platform: 114.6 Tbps of total bidirectional bandwidth over 1,024 SerDes lanes at 56 Gbps NRZ, 2.3 pJ/bit at system level including laser power (about 2.0 pJ/bit for the SerDes), an eight-tile (2 by 4) active interposer of 4,000 mm² with 256 fibers and more than 1 kW of power delivery, 32 I/O chiplets and two thermal-load chips stacked by chip-on-wafer, eClick detachable fibers, 8-wavelength DWDM from Guide, an integrated optical circuit-switching layer, and a liquid-cooled 19-inch 3U server with an SDK for bit-error-rate monitoring and telemetry.^[64] The Passage EVK100 is a 2D multi-chip package pairing the photonic IC with optical SerDes chiplets: 16-wavelength 112G PAM4 DWDM with 200 GHz spacing at 1310 nm, 1.6 Tbps unidirectional per fiber (3.2 Tbps aggregate over two fibers) at a target of 3.2 pJ/bit including laser, validated from 25 to 105 degrees C.^[65] The Passage EVK50 is the bidirectional platform: two Passage links, 16 wavelengths per fiber (eight in each direction), 800 Gbps per fiber at 56 Gbps NRZ with off-the-shelf XSR SerDes, single-mode fiber and a measured 2.6 pJ/bit including laser.^[66]

Idiom software stack

Idiom is the compiler and runtime that links existing deep-learning frameworks to Envise hardware. The stack ingests PyTorch and TensorFlow model graphs, applies optimization passes specific to photonic tensor cores (including precision lowering and operator fusion), and emits an execution plan that can target single Envise chips or large clusters of blades.^[5][15] Lightmatter identifies three subcomponents publicly: idCompile, the graph compiler; idProfiler, a runtime profiler; and idBug, a debugger.^[10][15] Idiom can automatically detect Envise topology and partition large models across blades, an important property given the multi-chip, multi-blade scale at which photonic systems are intended to operate.^[5][15]

Funding history

The following table summarizes Lightmatter's disclosed primary funding rounds. Cumulative capital is reported by multiple sources as approximately $850 million following the October 2024 Series D.^[7][17][6]

Variants and products

Partnerships

• GlobalFoundries: A roughly seven-year manufacturing collaboration was expanded in November 2024 to mass-produce Passage on the foundry's Fotonix silicon-photonics platform, which integrates photonic and electronic components on the same CMOS wafer.^[19][20]
• Amkor Technology: Advanced packaging partner for what the company described as the world's largest 3D photonic package, announced in late 2024.^[10]
• Hewlett Packard Enterprise: Strategic investor through both the Series B and the Hewlett Packard Pathfinder venture program in subsequent rounds.^[11][15] HPE's interest is consistent with the data-center server market for which Passage is being positioned.
• Lockheed Martin Ventures: Strategic investor since the 2021 Series B; the relationship has been described by Lightmatter and by press coverage as reflecting U.S. defense interest in photonic computing for high-performance and energy-efficient applications.^[11][13]
• Alphawave Semi / Qualcomm Technologies: Chiplet supplier for the UCIe electrical interfaces stacked above the photonic interposer in Passage L200.^[25] After Alphawave became part of Qualcomm, the collaboration produced the 1.6 Tbps-per-fiber Passage CPO chiplet announced in March 2026, and Qualcomm is a supporter of the OCP co-packaged optics initiative.^[38][40]
• ASE: Named alongside GlobalFoundries and Amkor as a manufacturing partner for the L200; demonstrated vClick compatibility in its advanced packaging flows.^[29][37] The company's site also lists TSMC and Tower Semiconductor as manufacturing partners.^[1]
• Global Unichip Corp. (GUC): January 2026 partnership to bring commercial Passage 3D co-packaged optics to hyperscale customers using GUC's ASIC design and packaging services; GUC is also a member of the OCP workstream.^[32][48]
• Synopsys and Cadence: January 2026 collaborations to integrate each company's 224G SerDes and UCIe IP with the Passage platform.^[33][34]
• SENKO: Supplies the SEAT and MPC connector technology used in vClick Optics.^[37]
• NVIDIA: Lightmatter joined the NVLink Fusion ecosystem in June 2026, committing to CPO and NPO products compatible with NVIDIA's optical and SerDes technologies.^[45]
• Standards bodies and consortia: The company's site lists OIF, IEEE, the Advanced Photonics Coalition, UALink, Ultra Ethernet, OCP, JEDEC and the UCIe Consortium; it joined UALink in December 2024, the XPO MSA in March 2026, and leads the OCP Open Silicon Photonics for AI Systems workstream.^{[1][59][39][48]}

Leadership and board

As of September 5, 2026 the company's website lists the following executives.^[53]

The board of directors listed on the same page is Nicholas Harris, Darius Bunandar, Olivia Nottebohm, Erik Nordlander, Jason Zander, Jeff Smith, Kushagra Vaid, Robin Washington and Rich Beyer; Dirk Englund and David Miller are technical advisors and Natasha Jen is creative advisor.^[53] Company releases record Beyer and Washington joining in July 2024 (optics.org described Beyer as a former Freescale Semiconductor CEO and Washington as an Alphabet executive) and Zander, who leads Microsoft's Strategic Missions and Technologies division, joining on March 25, 2025.^[6][57][58] Jankowski joined as CFO in July 2024 after nearly seven years at NVIDIA and, before that, a long stint as a chip analyst at Goldman Sachs.^[55][56] The site gives Mountain View as headquarters and lists Boston, Hsinchu and Toronto as other locations; the Toronto office was opened between December 2023 and October 2024 and the Hsinchu office in January 2026.^[53][6][35]

Significance and applications

Lightmatter's products are positioned at two intersecting bottlenecks in modern AI infrastructure: the per-operation energy cost of matrix multiplication in GPU inference and the bandwidth wall imposed by electrical chip-to-chip I/O.^[18][22] The Envise accelerator is targeted at neural-network inference workloads, including transformer and convolutional models, where the company has demonstrated end-to-end execution of BERT, ResNet and reinforcement-learning benchmarks in its Nature paper.^[9] The Passage product line is aimed at the much larger problem of optically interconnecting thousands to millions of accelerator dies within hyperscale data center AI training clusters.^[6][22]

The broader significance of Lightmatter, as repeatedly framed by Harris in press interviews and on the company's own materials, is the contention that continued scaling of large neural network training and inference will require co-design of compute and interconnect substrates rather than improvements to electronics alone, and that silicon photonics is the most production-ready substrate to fill that role.^[17][18][3] Passage in particular is described by Lightmatter as the world's first 3D-stacked photonics engine capable of connecting many processors at the speed of light at extreme scale.^[6]

Limitations and challenges

Lightmatter and external analysts have publicly acknowledged several persistent challenges in photonic AI hardware:

• Thermal sensitivity: Mach-Zehnder interferometers rely on precise phase relationships between the two arms of each device. Local temperature variations on a silicon-photonics chip shift the refractive index of the waveguides and therefore the relative phases, which in turn changes the computed matrix coefficients.^[24] Industrial photonic processors must therefore include on-chip thermal management and ongoing phase calibration to maintain numerical accuracy.^[24][9]
• Calibration overhead: Programming a large mesh of phase shifters to implement a target matrix requires either calibrating each element individually or running adaptive control loops. The IEEE Spectrum survey of optical-interconnect startups noted that Lightmatter's most aggressive interposer designs had been disclosed but, at the time of publication, not yet fully demonstrated in production hardware.^[25]
• Process integration: Silicon photonics requires fabrication steps (waveguide etching, modulator implants, photodetector integration) that are not standard parts of leading-edge logic CMOS processes. Lightmatter's reliance on GlobalFoundries' Fotonix platform is a strategic bet that monolithic electro-photonic integration on mature nodes is a more productive path than co-packaging discrete photonics with bleeding-edge logic dies.^[19][20]
• Software ecosystem: Photonic accelerators must compete against the deeply entrenched CUDA software stack and the dominant PyTorch runtime path that targets NVIDIA hardware. Idiom is designed to abstract that gap by ingesting PyTorch and TensorFlow graphs natively, but the absence of a published independent benchmark of Envise on standardized inference suites such as MLPerf has been a recurring observation in industry coverage.^[12][5]
• Commercialization risk: A wider set of photonic AI startups, including some with substantial venture funding, has faced reorganization or strategic pivots. Luminous Computing, which raised significant capital in 2022, reportedly hit technical hurdles and pivoted away from pure photonic compute in 2023.^[27] Lightmatter's strategy of building a separate, near-term-shippable interconnect product (Passage) in parallel with its longer-horizon compute product (Envise) is partly a response to that commercialization risk.^[17]

Competitive landscape

Lightmatter operates in two adjacent markets: photonic AI compute and photonic chip-to-chip interconnect. The following table summarizes principal competitors as identified in industry surveys.^[27][25]

Electronic AI-accelerator vendors are also implicit competitors at the system level. NVIDIA's position in AI training and inference is well established, and incumbents Cerebras Systems (wafer-scale electronic engines), Groq (deterministic LPU), SambaNova Systems, Tenstorrent and Etched Sohu all pursue alternative-architecture strategies, though none use silicon photonics as their primary compute substrate.^[27] At the interconnect layer Lightmatter must also compete against the trajectory of NVIDIA's own optical-interconnect roadmap and against established switch silicon vendors such as Broadcom and Marvell, both of which have been integrating co-packaged optics offerings of their own.^[27] By mid-2026 the picture was partly cooperative: IEEE Spectrum reported that Ayar Labs, Marvell and Lightmatter had all joined NVIDIA's NVLink Fusion ecosystem, and Lightmatter's own blog described Passage and Guide as aligned to the OCI MSA optical specification co-authored by NVIDIA, Broadcom and AMD.^[47][46]

Comparison with electronic AI accelerators

The qualitative case for photonic AI hardware versus mainstream electronic accelerators can be summarized along several axes. The table below collects publicly reported claims rather than independently audited benchmarks.

Lightmatter's Nature paper has been read by external commentators as the first peer-reviewed evidence that a photonic processor can match electronic accuracy on production-scale neural networks without bespoke retraining, a claim the company itself has subsequently echoed in marketing materials.^[9][8]

See also

• Massachusetts Institute of Technology
• AI accelerator
• Graphics processing unit
• NVIDIA
• NVIDIA A100
• Cerebras Systems
• Groq
• SambaNova Systems
• Tenstorrent
• Etched Sohu
• Data Center
• PyTorch
• TensorFlow
• BERT
• ResNet
• Transformer
• DeepMind
• Deep Learning
• Neural Network
• Reinforcement learning
• Convolutional Neural Network
• CUDA
• Inference

References

1. ^1 ^2 ^3 ^4 ^5 ^6Lightmatter, "Photonic chips for AI", lightmatter.co, 2026. lightmatter.co. Accessed 2026-09-05.
2. ^1 ^2 ^3 ^4 ^5Lightmatter, "Vision: Building What Comes Next", lightmatter.co, 2026. lightmatter.co/vision. Accessed 2026-09-05.
3. ^1 ^2 ^3 ^4 ^5Zach Winn, "Startup accelerates progress toward light-speed computing", MIT News, 2024-03-01. news.mit.edu/...-toward-light-speed-computing-0301. Accessed 2026-05-20.
4. ^1 ^2 ^3 ^4 ^5 ^6 ^7Nicholas Harris, "Introducing Envise, Idiom and Passage: Next Generation AI Compute, Compile and Interconnect Platforms", Lightmatter (Medium), 2021-03-10. medium.com/...ompile-and-interconnect-331878d6cea5. Accessed 2026-05-20.
5. ^1 ^2 ^3 ^4 ^5 ^6Lightmatter, "Idiom: Machine Learning and Neural Network Software", lightmatter.co, 2024. lightmatter.co/...idiom. Accessed 2026-05-20.
6. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13Lightmatter, "Lightmatter Raises $400M Series D; Quadruples Valuation to $4.4B as Photonics Leader for Next-Gen AI Data Centers", Business Wire, 2024-10-16. businesswire.com/...r-for-Next-Gen-AI-Data-Centers. Accessed 2026-05-20.
7. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11Tracxn, "Lightmatter: Funding Rounds and List of Investors", Tracxn, 2024. tracxn.com/...funding-and-investors. Accessed 2026-05-20.
8. ^1 ^2 ^3 ^4 ^5 ^6 ^7Nick Harris, "A New Kind of Computer", Lightmatter blog, 2025-04-09. lightmatter.co/...a-new-kind-of-computer. Accessed 2026-09-05.
9. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8Isabelle Dumé, "Photonic computer chips perform as well as purely electronic counterparts, say researchers", Physics World, 2025-04-30. physicsworld.com/...c-counterparts-say-researchers. Accessed 2026-05-20.
10. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10Contrary Research, "Lightmatter Business Breakdown and Founding Story", research.contrary.com, 2025. research.contrary.com/...lightmatter. Accessed 2026-05-20.
11. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9Jordan Williams, "Corporates illuminate Lightmatter in $80m Series B", Global Venturing, 2021-05-10. globalventuring.com/...ates-illuminate-lightmatter. Accessed 2026-05-20.
12. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18Nicole Hemsoth Prickett, "Lightmatter Normalizing Silicon Photonics for AI", The Next Platform, 2021-03-17. nextplatform.com/...izing-silicon-photonics-for-ai. Accessed 2026-05-20.
13. ^1 ^2 ^3 ^4 ^5FinSMEs, "Lightmatter Raises $80M in Series B Funding", FinSMEs, 2021-05-10. finsmes.com/...tter-raises-80m-in-series-b-funding. Accessed 2026-05-20.
14. ^1 ^2Lightwave Editorial Staff, "Photonic processor startup Lightmatter lands $154 million in Series C funding", Lightwave, 2023-05-31. lightwaveonline.com/...million-in-series-c-funding. Accessed 2026-05-20.
15. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8Devin Coldewey, "Lightmatter's photonic AI hardware is ready to shine with $154M in new funding", TechCrunch, 2023-05-31. techcrunch.com/...o-shine-with-154m-in-new-funding. Accessed 2026-05-20.
16. ^1 ^2 ^3Sacra, "Lightmatter valuation, funding and news", Sacra, 2024. sacra.com/...lightmatter. Accessed 2026-05-20.
17. ^1 ^2 ^3 ^4 ^5 ^6 ^7Chris Williams (citing Reuters), "Lightmatter: Photonic Supercomputing Company Raises $400 Million (Series D) At $4.4 Billion Valuation", Pulse 2.0, 2024-10-16. pulse2.com/...on-series-d-at-4-4-billion-valuation. Accessed 2026-05-20.
18. ^1 ^2 ^3Bloomberg News, "Photonic Computing Startup Lightmatter Reaches $4.4 Billion Valuation", Bloomberg, 2024-10-16. bloomberg.com/...ter-reaches-4-4-billion-valuation. Accessed 2026-05-20.
19. ^1 ^2 ^3 ^4 ^5 ^6Sebastian Moss, "Lightmatter turns to GlobalFoundries to bring photonics interconnect to market in 2025", Data Center Dynamics, 2024-11-19. datacenterdynamics.com/...onnect-to-market-in-2025. Accessed 2026-05-20.
20. ^1 ^2 ^3 ^4Lightmatter, "Lightmatter and GlobalFoundries Partner to Mass Produce Passage Platform", lightmatter.co, 2024-11-19. lightmatter.co/...to-mass-produce-passage-platform. Accessed 2026-05-20.
21. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9Lightmatter, "Lightmatter Unveils Passage M1000 Photonic Superchip, World's Fastest AI Interconnect", lightmatter.co press release, 2025-03-31. lightmatter.co/...p-worlds-fastest-ai-interconnect. Accessed 2026-05-20.
22. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9Anton Shilov, "Lightmatter unveils high-performance photonic 'superchip', claims world's fastest AI interconnect", Tom's Hardware, 2025-04-02. tomshardware.com/...worlds-fastest-ai-interconnect. Accessed 2026-05-20.
23. ^1 ^2Patrick Kennedy, "Lightmatter Passage M1000 at Hot Chips 2025", ServeTheHome, 2025-08-26. servethehome.com/...assage-m1000-at-hot-chips-2025. Accessed 2026-05-20.
24. ^1 ^2 ^3 ^4 ^5Daniel Brunner et al., "Photonic Computing Takes a Step Toward Fruition", Physics (APS), 2025-05-19. link.aps.org/...Physics.18.84. Accessed 2026-05-20.
25. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9Samuel K. Moore, "Start-ups Replace Copper with Optical Links for GPUs", IEEE Spectrum, 2025. spectrum.ieee.org/optics-gpu. Accessed 2026-05-20.
26. ^1 ^2 ^3 ^4Lightmatter, "Passage: 3D Photonics for AI Applications", lightmatter.co, 2024. lightmatter.co/...passage. Accessed 2026-05-20.
27. ^1 ^2 ^3 ^4 ^5 ^6 ^7ts2.tech, "Photonic AI Accelerators vs. GPUs: The Battle for AI's Future in Efficiency, Cost, and Scale", ts2.tech, 2025. ts2.tech/...is-future-in-efficiency-cost-and-scale. Accessed 2026-05-20.
28. ^1 ^2 ^3Sufi R. Ahmed, Reza Baghdadi, Mikhail Bernadskiy et al. (senior author Nicholas C. Harris), "Universal photonic artificial intelligence acceleration", Nature 640, 368-374, 2025-04-09. nature.com/...s41586-025-08854-x. Accessed 2026-09-05.
29. ^1 ^2 ^3 ^4 ^5 ^6Lightmatter, "Lightmatter Announces Passage L200, the Fastest Co-Packaged Optics for AI", lightmatter.co press release, 2025-03-31. lightmatter.co/...astest-co-packaged-optics-for-ai. Accessed 2026-09-05.
30. ^1 ^2 ^3 ^4Lightmatter, "Lightmatter Achieves World-First 16-Wavelength Bidirectional Link on Single-Mode Optical Fiber", lightmatter.co press release, 2025-08-18. lightmatter.co/...ink-on-single-mode-optical-fiber. Accessed 2026-09-05.
31. ^1 ^2 ^3 ^4 ^5Lightmatter, "Lightmatter Introduces Guide Light Engine for AI, Featuring VLSP Technology", lightmatter.co press release, 2026-01-26. lightmatter.co/...for-ai-featuring-vlsp-technology. Accessed 2026-09-05.
32. ^1 ^2 ^3Lightmatter, "Lightmatter and GUC Partner to Produce Co-Packaged Optics (CPO) Solutions for AI Hyperscalers", lightmatter.co press release, 2026-01-26. lightmatter.co/...po-solutions-for-ai-hyperscalers. Accessed 2026-09-05.
33. ^1 ^2 ^3Lightmatter, "Lightmatter Collaborates with Synopsys to Integrate Advanced Interface IP with Its Passage Co-Packaged Optics Platform", lightmatter.co press release, 2026-01-26. lightmatter.co/...sage-co-packaged-optics-platform. Accessed 2026-09-05.
34. ^1 ^2 ^3Lightmatter, "Lightmatter and Cadence Collaborate to Accelerate Optical Interconnect for AI Infrastructure", lightmatter.co press release, 2026-01-26. lightmatter.co/...terconnect-for-ai-infrastructure. Accessed 2026-09-05.
35. ^1 ^2 ^3 ^4Nick Harris, "Building out the Photonic Stack", Lightmatter blog, 2026-01-27. lightmatter.co/...building-out-the-photonic-stack. Accessed 2026-09-05.
36. ^1 ^2 ^3 ^4 ^5Lightmatter, "Lightmatter Expands Photonic Interconnect Roadmap with Passage L20 Unified Optical Engine for NPO and OBO Applications", lightmatter.co press release, 2026-03-11. lightmatter.co/...ine-for-npo-and-obo-applications. Accessed 2026-09-05.
37. ^1 ^2 ^3 ^4 ^5 ^6Lightmatter, "Lightmatter Unveils vClick Optics, Industry-First Detachable Fiber Array Unit for CPO Advanced Packaging and High-Volume Production", lightmatter.co press release, 2026-03-11. lightmatter.co/...aging-and-high-volume-production. Accessed 2026-09-05.
38. ^1 ^2 ^3 ^4Lightmatter, "Lightmatter Achieves Record 1.6 Tbps Per Fiber to Accelerate AI Optical Interconnect", lightmatter.co press release, 2026-03-11. lightmatter.co/...celerate-ai-optical-interconnect. Accessed 2026-09-05.
39. ^1 ^2 ^3Lightmatter, "Lightmatter Joins XPO MSA as Founding Member to Accelerate High-Density Optical Interconnects for AI Data Centers", lightmatter.co press release (also on Business Wire), 2026-03-12. lightmatter.co/...nterconnects-for-ai-data-centers. Accessed 2026-09-05.
40. ^1 ^2 ^3Lightmatter, "Lightmatter Announces Reference Architecture Initiative with Industry Leaders in the Open Compute Project for Co-Packaged Optics", lightmatter.co press release, 2026-03-16. lightmatter.co/...e-project-for-co-packaged-optics. Accessed 2026-09-05.
41. ^Tobias Mann, "Lightmatter says latest photonics will slash datacenter fiber bills in half", The Register, 2026-03-11. theregister.com/...lightmatter_passge_l20_fiber. Accessed 2026-09-05.
42. ^1 ^2Lightmatter, "Lightmatter Names Roy Kim Vice President of Product to Lead Global Deployment of Photonic Interconnects", lightmatter.co press release (also on Business Wire), 2026-04-27. lightmatter.co/...oyment-of-photonic-interconnects. Accessed 2026-09-05.
43. ^1 ^2 ^3 ^4 ^5Lightmatter, "Lightmatter Unveils Guide DR: Industry First Liquid-Cooled Laser NIC that Quadruples Rack Density", lightmatter.co press release (also on Business Wire), 2026-05-21. lightmatter.co/...nic-that-quadruples-rack-density. Accessed 2026-09-05.
44. ^Charlotte Trueman, "Lightmatter unveils Guide DR laser NIC to support CPO scale-up", Data Center Dynamics, 2026-05-27. datacenterdynamics.com/...-to-support-cpo-scale-up. Accessed 2026-09-05 (via web.archive.org).
45. ^1 ^2 ^3 ^4 ^5Lightmatter, "Lightmatter Joins NVIDIA NVLink Fusion and Powers Next-Generation AI Infrastructure with Photonic Interconnects", lightmatter.co press release (Business Wire, 2026-06-02; datelined Taipei, June 3, 2026). lightmatter.co/...atter-joins-nvidia-nvlink-fusion. Accessed 2026-09-05.
46. ^1 ^2 ^3Nick Harris, "Scale-Up is a Problem Made for Photonics", Lightmatter blog, 2026-06-02. lightmatter.co/...-is-a-problem-made-for-photonics. Accessed 2026-09-05.
47. ^1 ^2Knvul Sheikh, "Is Optical Scale-Up Finally Approaching? NVLink Fusion hints that optics may be moving inside the rack", IEEE Spectrum, 2026-07-09. spectrum.ieee.org/nvlink-fusion-optics. Accessed 2026-09-05.
48. ^1 ^2 ^3 ^4 ^5Lightmatter, "Industry Leaders Formally Launch CPO System Architecture Initiative Within the Open Compute Project", lightmatter.co press release, 2026-08-13. lightmatter.co/...-within-the-open-compute-project. Accessed 2026-09-05.
49. ^1 ^2HPCwire (source: Lightmatter), "Lightmatter, Industry Partners Launch CPO Architecture Initiative Within OCP", HPCwire, 2026-08-14. hpcwire.com/...-architecture-initiative-within-ocp. Accessed 2026-09-05.
50. ^Ben Wodecki, "Lightmatter and OCP work to standardize silicon photonic interconnects for future AI cluster backbones", SDxCentral, 2026-08-14. sdxcentral.com/...s-for-next-gen-ai-infrastructure. Accessed 2026-09-05.
51. ^Jim Carroll, "OCP Launches Open Silicon Photonics for AI Systems Initiative", Converge! Network Digest, 2026-08-13. convergedigest.com/...tonics-ai-systems-initiative. Accessed 2026-09-05.
52. ^DIGITIMES, "Lightmatter white paper targets fragmented supply chain, barriers to photonic interconnect deployment", DIGITIMES Asia, 2026-08-13. digitimes.com/...re-supply-chain-transmission-2026. Accessed 2026-09-05.
53. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10Lightmatter, "About", lightmatter.co, 2026. lightmatter.co/about. Accessed 2026-09-05.
54. ^1 ^2 ^3Lightmatter, "The Future Is Bright -- Lightmatter Raises $154M to Deliver Photonic Products to Customers", Business Wire, 2023-05-31. businesswire.com/...en. Accessed 2026-09-05 (via web.archive.org).
55. ^1 ^2 ^3 ^4 ^5Ian King, "Nvidia's Jankowski Takes CFO Job at Chip Startup Lightmatter", Bloomberg (via Yahoo Finance), 2024-07-01. finance.yahoo.com/...owski-takes-cfo-job-153000073. Accessed 2026-09-05.
56. ^1 ^2 ^3 ^4 ^5 ^6SDxCentral, "Former Nvidia exec Simona Jankowski appointed CFO of computing company Lightmatter", SDxCentral, 2024-07-02. sdxcentral.com/...of-computing-company-lightmatter. Accessed 2026-09-05.
57. ^optics.org, "Google backs Lightmatter again in $400M fundraising", optics.org, 2024-10-22. optics.org/...ightmatter-again-in-400m-fundraising. Accessed 2026-09-05.
58. ^1 ^2 ^3Lightmatter, "Lightmatter Welcomes Industry Leader Jason Zander to Its Board of Directors", lightmatter.co press release (Business Wire), 2025-03-25. lightmatter.co/...zander-to-its-board-of-directors. Accessed 2026-09-05.
59. ^1 ^2 ^3Lightmatter, "Lightmatter Joins UALink Consortium to Propel AI Interconnect into the Photonic Era", lightmatter.co press release, 2024-12-18. lightmatter.co/...terconnect-into-the-photonic-era. Accessed 2026-09-05.
60. ^Lightmatter (@LightmatterCo), post on X, 2026-09-04. x.com/...2095935332411965942. Accessed 2026-09-05.
61. ^Nick Harris, "Wide and Parallel Wins: DWDM is the Future of Scale-Up Networking", Lightmatter blog, 2026-03-15. lightmatter.co/...wide-and-parallel-wins. Accessed 2026-09-05.
62. ^1 ^2Lightmatter, "Passage L20: Scalable BiDi Optical Interconnect", lightmatter.co, 2026. lightmatter.co/...passage-l20. Accessed 2026-09-05.
63. ^Lightmatter, "Passage L200: First Edgeless I/O 3D Co-Packaged Optics", lightmatter.co, 2026. lightmatter.co/...l200. Accessed 2026-09-05.
64. ^1 ^2 ^3 ^4Lightmatter, "Passage M1000 EVK: 3D Photonic Interposer for Exascale AI", lightmatter.co, 2026. lightmatter.co/...passage-m1000-evk. Accessed 2026-09-05.
65. ^1 ^2Lightmatter, "Passage EVK100: Exascale Connectivity for AI", lightmatter.co, 2026. lightmatter.co/...passage-evk100. Accessed 2026-09-05.
66. ^1 ^2Lightmatter, "Passage EVK50: World-First 16-wavelength BiDi CPO Link", lightmatter.co, 2026. lightmatter.co/...passage-evk50. Accessed 2026-09-05.
67. ^1 ^2Lightmatter, "Guide 1: Lasers redefined for the CPO era", lightmatter.co, 2026. lightmatter.co/...guide-1. Accessed 2026-09-05.
68. ^1 ^2 ^3Lightmatter, "Guide DR: Lasers reimagined for the scale-up era", lightmatter.co, 2026. lightmatter.co/...guide-dr. Accessed 2026-09-05.
69. ^1 ^2 ^3 ^4Lightmatter, "Press Releases", lightmatter.co, 2026. lightmatter.co/press-releases. Accessed 2026-09-05.