UNITED STATES DISTRICT COURT EASTERN DISTRICT OF TEXAS MARSHALL DIVISION
OPTIWAVE SEMICONDUCTOR CORP.,
Plaintiff,
v.
LUMICORE TECHNOLOGIES, INC.,
Defendant.
Case No. 2:24-cv-00319-JRG
Judge James R. Gilstrap
JURY TRIAL DEMANDED
EXPERT REPORT OF GREGORY FALK REGARDING THE SILICON-PHOTONICS TRANSCEIVER MARKET, COMPETITIVE LANDSCAPE, TECHNOLOGY EVOLUTION, AND COMMERCIAL SIGNIFICANCE OF THE PATENTED INNOVATIONS
DRAFT — ATTORNEY WORK PRODUCT — CONFIDENTIAL
Final Report Due: June 15, 2025
This report contains information designated "CONFIDENTIAL" and "HIGHLY CONFIDENTIAL — ATTORNEYS' EYES ONLY" under the Stipulated Protective Order entered in this action. This report also incorporates information derived from source-code materials designated "HIGHLY CONFIDENTIAL — SOURCE CODE." Distribution is restricted to persons qualified to receive such materials under the Protective Order.
Prepared for:
Counsel for Plaintiff Optiwave Semiconductor Corp. Calderwood & Harkness LLP 350 South Grand Avenue, 48th Floor Los Angeles, CA 90071
I. INTRODUCTION
I, Gregory Falk, submit this report in the above-captioned action pursuant to Federal Rule of Civil Procedure 26(a)(2)(B). I have been retained by Calderwood & Harkness LLP on behalf of Plaintiff Optiwave Semiconductor Corp. ("OWS" or "Optiwave") to provide expert opinions regarding (i) the silicon-photonics transceiver market, (ii) the competitive landscape among OWS and its competitors, including Defendant Lumicore Technologies, Inc. ("Lumicore"), (iii) the evolution of the relevant technology, and (iv) the commercial significance of the innovations embodied in U.S. Patent Nos. 10,234,847 (the "'847 Patent"), 10,891,193 (the "'193 Patent"), 9,678,562 (the "'562 Patent"), and 11,402,015 (the "'015 Patent") (collectively, the "Asserted Patents").
This report reflects the opinions I have formed to date based on the materials I have reviewed and my professional experience. Because discovery is ongoing and further materials — including but not limited to Lumicore's FY 2024 financial productions, additional depositions, and the Court's forthcoming claim-construction ruling — remain outstanding, I reserve the right to supplement or revise this report as appropriate. See Part XIV, infra.
II. QUALIFICATIONS AND BACKGROUND OF EXPERT
A. Professional Summary
I am an independent consultant specializing in silicon-photonics and optical interconnect technologies. I have more than thirty (30) years of continuous experience in the optical communications and semiconductor industries, spanning discrete optical component design, photonic integrated circuit ("PIC") development, waveguide engineering, transceiver module architecture, and the commercial and manufacturing aspects of silicon-photonics products.
From the company's founding in 2003 through its dissolution in 2020, I served as Vice President of Engineering at Solara Optics, Inc., a California corporation that specialized in photonic integrated circuit design and optical component R&D. In that role, I directly oversaw the research and engineering teams responsible for, among other things, the technology that became the subject matter of the '562 Patent. I am a named inventor on several U.S. patents in waveguide design and photonic-component fabrication, and I have published in peer-reviewed technical literature and presented at leading industry conferences, including the Optical Fiber Communication Conference ("OFC") and the IEEE Photonics Conference.
B. Retention in This Matter
I was retained by Calderwood & Harkness LLP pursuant to an Expert Retention Agreement dated August 8, 2024. My billing rate for this engagement is $650 per hour, and the estimated budget for my engagement is approximately $375,000. My compensation is not contingent, in whole or in part, on the substance of the opinions I render or on the outcome of this litigation.
The scope of my engagement, as reflected in the Retention Agreement, is to provide expert opinions on:
(i) the silicon-photonics transceiver market, including its size, growth, structure, and demand drivers;
(ii) the competitive landscape between OWS and its competitors, including Lumicore;
(iii) the evolution of the relevant technology and the technical significance of the Asserted Patents; and
(iv) the commercial significance of the innovations embodied in the Asserted Patents, including issues of long-felt need, failure of others, industry recognition, copying, and nexus between commercial success and the patented features.
C. Education and Prior Positions
I hold a Ph.D. in Electrical Engineering (specialization: Integrated Optics and Photonics), an M.S. in Electrical Engineering, and a B.S. in Physics. My full academic credentials, employment history, publications, patents, professional affiliations (including the Institute of Electrical and Electronics Engineers ("IEEE") and Optica, formerly the Optical Society of America ("OSA")), and prior expert-testimony engagements are set forth in my curriculum vitae, attached as Appendix B. In brief:
• 2003–2020: Vice President of Engineering, Solara Optics, Inc. — responsible for all optical component R&D, PIC design, process development liaison with foundry partners, and technical due diligence supporting Solara's licensing and strategic transactions. Following Solara's dissolution in 2020, I have served as records custodian for the company's archived technical files.
• 1994–2003: Senior positions in optical component and semiconductor companies, culminating in Director-level engineering responsibility over waveguide device design.
• 1988–1994: Design engineer and senior engineer roles in optical communications component development.
I am a named inventor on multiple issued U.S. patents in waveguide design and photonic-component fabrication, and I have authored or co-authored peer-reviewed publications and conference papers in the field. I have previously served as an expert witness in patent matters involving optical and photonic technologies; those engagements are itemized in Appendix B.
D. Materials Reviewed
In forming the opinions expressed in this report, I have reviewed and considered the materials listed comprehensively in Appendix A. Without limitation, the principal materials I have relied upon include:
• The Asserted Patents — U.S. Patent Nos. 10,234,847; 10,891,193; 9,678,562; and 11,402,015 — together with their complete prosecution histories, as compiled and provided to me by Calderwood & Harkness LLP in September 2024;
• The teardown reports prepared by Benchmark Semiconductor Analysis, LLC (Kevin Ng, Lead Engineer, 2244 Mission College Boulevard, Santa Clara, CA 95054), specifically: (i) the LumiLink 400G-ZR+ teardown dated December 12, 2022; (ii) the LumiLink 800G-DR8 teardown dated June 2023; and (iii) the LumiEdge 1600G-LR teardown dated February 2024;
• Lumicore design documents and engineering specifications produced in discovery, including materials designated "Highly Confidential — Attorneys' Eyes Only" and "Highly Confidential — Source Code";
• OWS product datasheets and internal engineering documentation for the OWS-400ZR, OWS-800XR, and OWS-1600DR transceiver module families;
• The Asset Purchase Agreement dated June 15, 2018 between Solara Optics, Inc. and OWS, pursuant to which OWS acquired the '562 Patent along with 11 other patents for $34.5 million (the "Solara APA");
• OWS's FY 2023 audited financial statements (audited by Gresham & Whitfield LLP, San Jose office), reflecting revenue of $487.3 million and net income of $61.4 million;
• Lumicore's FY 2023 Form 10-K filing, reflecting estimated revenue of $312.6 million;
• Lumicore product-level financial data produced in discovery, reflecting accused-product gross margins ranging from 52.1% (LumiLink 400G-ZR+) to 58.3% (LumiEdge 1600G-LR);
• Industry reports, market-research publications, IEEE journals, OFC conference proceedings, and other technical and market literature identified in Appendix A;
• Deposition transcripts and exhibits from the depositions of Dr. Jian-Wei Huang (OWS CTO, November 5, 2024); Dr. Sandra Michaud (Lumicore CTO, November 19, 2024); Angela Reeves (Managing Director, FairPoint Licensing Group, LLC, December 10, 2024); and Kevin Ng (Benchmark Semiconductor Analysis, LLC, January 7, 2025);
• Pre-suit licensing correspondence from FairPoint Licensing Group, LLC to Lumicore dated January 15, 2023 and May 15, 2023, together with contemporaneous records of the April 20, 2023 in-person licensing meeting held in Austin, Texas; and
• Draft expert reports of Dr. Caroline Ashford (Northfield Institute of Technology) and Dr. Lawrence Keating (Meridian Economic Consulting Group, LLC).
E. Compensation
As noted above, my hourly rate for this engagement is $650. Neither my compensation nor my continued engagement is in any way contingent on the substance of my opinions or on the outcome of this litigation.
III. SUMMARY OF OPINIONS
Based on the materials I have reviewed, the analysis I have performed, and my more than thirty years of experience in the field, I have formed the following opinions, each of which is developed in detail in the sections that follow:
Opinion 1. The silicon-photonics transceiver market for hyperscale data-center interconnects is a large, rapidly growing, and strategically critical segment of the global semiconductor industry, driven by the sustained bandwidth demands of cloud computing, artificial-intelligence training and inference workloads, and machine-learning infrastructure.
Opinion 2. OWS occupies a significant competitive position in this market as the third-largest U.S. supplier of silicon-photonics transceiver modules for hyperscale data-center applications. That position is built upon foundational innovations in (a) low-insertion-loss waveguide multiplexing ('847 Patent), (b) coherent transceiver integration on silicon ('193 Patent), (c) embedded micro-channel thermal management of photonic integrated circuits ('562 Patent), and (d) polarization-diverse grating coupling ('015 Patent).
Opinion 3. The innovations embodied in the Asserted Patents addressed long-felt but previously unsolved needs in the silicon-photonics industry and represent commercially significant technological breakthroughs.
Opinion 4. Lumicore's entry into the market with its LumiLink 400G-ZR+ (Q3 2022), LumiLink 800G-DR8 (Q1 2023), and LumiEdge 1600G-LR (Q4 2023) product lines has materially harmed OWS through price erosion, customer diversion, and loss of design-in opportunities.
Opinion 5. Prior to OWS's innovations (and the Solara Optics innovation that became the '562 Patent), multiple competitors attempted and failed to achieve the low-insertion-loss waveguide multiplexing, integrated coherent transceiver architectures, embedded micro-channel thermal management, and polarization-diverse grating coupling that the Asserted Patents describe and claim.
Opinion 6. The evidence supports the conclusion that Lumicore copied OWS's patented technology in designing its accused product lines, including through documented design-phase review of OWS's published patent applications and continued introduction of infringing products after receipt of pre-suit notice.
Opinion 7. The features claimed in the Asserted Patents are the basis for customer demand for the accused transceiver modules and for OWS's competing modules.
IV. THE SILICON-PHOTONICS TRANSCEIVER MARKET
A. Market Overview and Size
Silicon photonics is a technology platform that integrates optical components — waveguides, modulators, photodetectors, wavelength multiplexers, and grating couplers — onto silicon substrates using fabrication techniques compatible with mainstream CMOS foundry processes. Its principal commercial application today is high-bandwidth, short-, medium-, and long-reach optical interconnects within and between hyperscale data centers.
The market for silicon-photonics transceivers has grown at compound annual rates exceeding 30% over the past several years, driven principally by:
• Exponential growth in traffic between and within hyperscale data centers, itself driven by cloud-computing services and the emergence of large-scale AI training clusters;
• The industry-wide transition from 400 Gb/s to 800 Gb/s per-lane and per-module data rates, with 1600 Gb/s modules now beginning volume ramp; and
• The unsustainability of copper-based interconnects at the reach and bandwidth levels required by modern GPU-based AI clusters.
Transceivers interconnect switches, servers, GPUs, and storage nodes within a data-center campus and across long-haul metro links. They are, in effect, the sensory-and-nervous-system components of the modern data center; without them, the compute and storage infrastructure cannot exchange data at the rates required by contemporary workloads.
B. Key Market Participants
Optiwave Semiconductor Corp. OWS is a fabless semiconductor company incorporated in Delaware and headquartered in Santa Clara, California. Founded in 2011, OWS today employs approximately 1,250 personnel and reported FY 2023 revenue of $487.3 million and net income of $61.4 million (per audited financials issued by Gresham & Whitfield LLP, San Jose office). OWS's principal product families are the OWS-400ZR, OWS-800XR, and OWS-1600DR silicon-photonics transceiver modules. OWS holds 142 issued U.S. patents and has 58 pending U.S. patent applications. Since 2019, Ridgeline Capital Partners, LP (200 Clarendon Street, 40th Floor, Boston, MA 02116) has held a 55% majority equity stake in OWS, acquired for $780 million; Franklin Osei, Managing Partner at Ridgeline, serves as Chair of the OWS Board.
Lumicore Technologies, Inc. Lumicore is a Delaware corporation headquartered at 9200 Spectrum Boulevard, Building C, Austin, Texas 78759. It was founded in 2015, employs approximately 870 personnel, and reported estimated FY 2023 revenue of $312.6 million. Lumicore completed its initial public offering on the Nasdaq Global Select Market in March 2022 at $24.00 per share; as of the April 22, 2024 filing date of the Complaint in this action, Lumicore's shares traded at approximately $37.00, yielding a market capitalization of approximately $3.4 billion based on 91.8 million shares outstanding. Lumicore's CEO is Thomas ("Tom") Kessler; its CTO is Dr. Sandra Michaud. Its accused product lines are the LumiLink 400G-ZR+, LumiLink 800G-DR8, and LumiEdge 1600G-LR. Lumicore's auditor is Brennan Kellogg & Associates LLP (Austin office); its lead investment bank is Silverstone Capital Markets LLC.
Beyond OWS and Lumicore, the market includes a small number of additional competitors — principally larger, vertically integrated semiconductor manufacturers and a handful of smaller specialty suppliers. However, at the highest performance tiers (state-of-the-art 800G-DR8 and 1600G-LR modules qualified into leading hyperscaler production networks), OWS and Lumicore effectively operate as the two leading U.S. suppliers in a functionally two-supplier competitive dynamic. This narrow two-supplier structure means that each design-in loss and each unit of price erosion has an outsized impact on OWS's competitive position.
C. Customer Base and Demand Drivers
Silicon-photonics transceivers for hyperscale data-center applications are purchased almost exclusively by a narrow universe of hyperscale cloud operators, tier-1 telecommunications operators, and large content and platform companies. Based on my review of OWS's customer records, discovery from Lumicore, and my discussions with OWS engineering and sales personnel, OWS and Lumicore compete head-to-head at 12 of the top 20 hyperscale data-center operators; 7 of the top 20 have qualified and purchased from both OWS and Lumicore during the FY 2023 period.
Design-in cycles in this market are lengthy and consequential. A hyperscale customer typically undertakes an 18-to-24-month qualification process before qualifying a transceiver module for volume deployment. Once a module is qualified and deployed into a data-center fabric, switching costs are substantial: qualification testing must be re-run, firmware and management-plane integration must be re-validated, and the customer must accept the risk of introducing a new supplier into a mission-critical infrastructure element. As a result, each design-in loss represents not merely a single-unit sale, but the loss of a multi-year revenue stream — a dynamic that meaningfully amplifies the competitive impact of Lumicore's accused products.
Within the Eastern District of Texas, customers of Lumicore's accused product lines include, among others, Pinnacle Data Systems, LLC (1800 Judson Road, Longview, TX 75601); Frontier Cloud Services, Inc. (400 East Houston Street, Marshall, TX 75670); and CrossPoint Networking Corp. (7500 Dallas Parkway, Suite 200, Plano, TX 75024).
D. Market Share Analysis
Approximately $310 million of OWS's $487.3 million FY 2023 revenue — roughly 63.6% — is attributable to sales of products that practice one or more of the Asserted Patents. That is a substantial revenue base reflecting sustained customer acceptance and repeat purchases of OWS's patented technology.
Beginning in Q3 2022, coincident with the launch of the LumiLink 400G-ZR+, Lumicore's market-share trajectory has climbed steeply, principally at the expense of OWS. Based on my analysis of customer overlap, head-to-head competitive engagements at the 12 top-20 hyperscale operators where both suppliers are qualified, and appropriate adjustments for hyperscale customers where OWS is not an approved vendor, I have determined that approximately 38.2% of Lumicore's accused-product sales represent revenue that would, but for Lumicore's infringement, have been captured by OWS. The remainder of Lumicore's accused-product sales — approximately 61.8% — reflect sales to customers where either OWS is not qualified or capacity or design-cycle timing would have foreclosed OWS from the opportunity in any event. My market-share displacement methodology is described more fully in Section X.C, infra, and in Appendix C.
V. TECHNOLOGY EVOLUTION IN SILICON-PHOTONICS TRANSCEIVERS
A. Historical Development of Silicon Photonics
Silicon photonics emerged, over the past two decades, from the fundamental insight that many of the functions historically performed by discrete optical components — beam splitters, wavelength multiplexers, modulators, detectors — could be integrated onto a single silicon substrate using fabrication techniques closely aligned with those used in mainstream CMOS microelectronics. This convergence permitted photonic-component manufacturing to leverage the enormous installed base of CMOS foundry capacity, achieving cost, yield, and scale advantages that discrete III-V compound-semiconductor platforms could not match.
The transition from discrete-optics assemblies to photonic integrated circuits, however, was neither immediate nor straightforward. Each functional element in a transceiver — the multiplexer, the modulator, the photodetector, the coupling structures between fiber and chip, and the thermal-management subsystem — presented distinct engineering challenges when reduced to silicon-integrable form. The four Asserted Patents address four of those challenges, each of which had persisted as an open problem in the industry until it was solved by the invention that the corresponding patent claims.
B. Waveguide Multiplexing Technology (Context for the '847 Patent)
In wavelength-division-multiplexed ("WDM") transceivers, multiple optical carrier wavelengths are combined into and separated from a single fiber. In silicon photonics, this multiplexing function is typically implemented using one of three families of on-chip devices: arrayed waveguide gratings ("AWGs"), echelle gratings, or ring-resonator-based structures. Each of these prior-art families exhibited significant limitations at the performance levels demanded by 400G-class and higher transceivers:
• AWGs offered moderate channel counts but suffered from relatively high insertion losses (typically 3–5 dB) and were sensitive to temperature and fabrication variation;
• Echelle gratings offered compact footprints but similarly suffered from insertion loss and polarization sensitivity;
• Ring-resonator-based multiplexers were compact but exhibited narrow passbands and pronounced thermal sensitivity, requiring active tuning.
The '847 Patent, entitled "Multi-Channel Silicon Photonic Waveguide Multiplexer with Low Insertion Loss," was filed September 14, 2016 and issued March 19, 2019. It was originally assigned to OWS. The patent claims a specific waveguide multiplexer configuration that, in my technical judgment, achieves materially lower insertion loss than prior-art approaches while preserving compact on-chip footprint and manufacturability in standard silicon-photonics foundry processes. The '847 Patent is part of a continuation family that also includes the '562 Patent's underlying technology stream, and the two patents together reflect a coordinated technology arc from thermal-management foundations to advanced multiplexer design.
C. Coherent Optical Transceiver Integration (Context for the '193 Patent)
The '193 Patent, entitled "High-Bandwidth Coherent Optical Transceiver Integrated on Silicon Substrate," was filed April 7, 2017 and issued January 12, 2021. It was originally assigned to OWS.
Coherent optical detection is an architecture family that historically has been implemented in discrete III-V compound-semiconductor components. Integrating a coherent transceiver on a silicon substrate — while preserving bandwidth, sensitivity, and reach — has long been a central engineering challenge in silicon photonics. Prior-art approaches relied on hybrid integration (mounting discrete III-V die adjacent to silicon PICs) or heterogeneous bonding (die-to-wafer or wafer-to-wafer bonding of III-V material to silicon). Each of these approaches introduced yield, thermal, alignment, and cost penalties.
The '193 Patent claims an integrated coherent transceiver architecture that overcomes these limitations. The '193 Patent shares a common provisional application (No. 62/499,117, filed April 7, 2017) with the '015 Patent, reflecting the fact that the underlying technology stream was conceived and disclosed as an integrated program of transceiver-architecture innovation.
D. Thermal Management of Photonic Integrated Circuits (Context for the '562 Patent)
Silicon exhibits a strong thermo-optic effect: refractive index varies with temperature. In densely integrated PICs, uncontrolled temperature variation causes wavelength drift, insertion-loss increases, and, in extreme cases, functional failure of ring-resonator-based devices. Prior to the '562 Patent, the industry's principal approaches to thermal management — conventional heat sinks and thermoelectric coolers (TECs) at the package level — were increasingly inadequate as PIC integration density and per-die power dissipation grew.
The '562 Patent, entitled "Method for Thermal Management of Photonic Integrated Circuits Using Embedded Micro-Channel Heat Sinks," was filed November 22, 2014 and issued June 13, 2017. It was originally assigned to Solara Optics, Inc., and is currently assigned to OWS by virtue of the Solara APA dated June 15, 2018, under which OWS acquired the '562 Patent together with 11 other Solara patents for aggregate consideration of $34.5 million.
I have direct, personal knowledge of the development of the technology that became the '562 Patent. During my tenure as Vice President of Engineering at Solara Optics, I supervised the engineering team responsible for the design and reduction to practice of the embedded micro-channel thermal-management structures that the patent claims. I also served as records custodian for Solara Optics' archived technical files following the company's dissolution in 2020; in that capacity, I responded to a third-party subpoena issued on September 9, 2024 for Solara corporate records relating to the '562 Patent.
E. Polarization-Diverse Grating Coupling (Context for the '015 Patent)
Fiber-to-chip coupling in silicon photonics is commonly implemented using grating couplers, which diffract light out of the plane of the chip into an optical fiber (and vice versa). Conventional one-dimensional grating couplers, however, are polarization-selective: they efficiently couple only one polarization state of the incoming light. Because the polarization state of light emerging from a standard single-mode fiber is generally unknown and varies over time, approximately half of the incoming optical power is lost or must be recovered through external polarization control — an approach that increases module cost, complexity, and physical footprint.
The '015 Patent, entitled "Polarization-Diverse Grating Coupler for Silicon Photonic Transceivers," was filed August 30, 2019 and issued August 2, 2022. It was originally assigned to OWS. The '015 Patent claims a two-dimensional, polarization-diverse grating coupler configuration that efficiently couples both polarization states of the incoming light onto orthogonal on-chip waveguides, eliminating the need for external polarization controllers. As noted above, the '015 Patent shares a common provisional application (No. 62/499,117, filed April 7, 2017) with the '193 Patent.
F. Interrelationship of the Four Patented Technologies
The four innovations do not exist in isolation. In a complete silicon-photonics transceiver module such as OWS's OWS-400ZR, OWS-800XR, or OWS-1600DR, the four technologies function together as an integrated whole:
(i) Light enters the chip through a polarization-diverse grating coupler ('015 Patent), which captures both polarization states with high efficiency;
(ii) The multi-channel waveguide multiplexer ('847 Patent) combines or separates wavelength channels with low insertion loss;
(iii) The integrated coherent transceiver ('193 Patent) performs high-bandwidth modulation and detection on each channel; and
(iv) Embedded micro-channel thermal-management structures ('562 Patent) maintain the on-chip temperature profile within tolerances required for stable multi-channel operation.
Each innovation independently contributes to the commercial performance of OWS's transceiver modules; collectively, the four innovations define the differentiated performance envelope on which OWS competes.
VI. COMMERCIAL SIGNIFICANCE OF THE PATENTED INNOVATIONS
A. Commercial Success of OWS Products Embodying the Asserted Patents
OWS's FY 2023 total revenue of $487.3 million included approximately $310 million from transceiver modules — principally the OWS-400ZR, OWS-800XR, and OWS-1600DR product families — that practice one or more of the Asserted Patents. This represents both an absolute revenue base of commercial significance and, in relative terms, the majority of OWS's total revenue.
OWS's revenue attributable to patent-practicing products has grown steadily since the introduction of the relevant product families, reflecting sustained customer acceptance. Product-family-level revenue trends are set forth in Appendix C. In addition:
• OWS is the third-largest U.S. supplier of silicon-photonics transceiver modules for hyperscale data-center applications;
• OWS's gross margin on the patent-practicing products is approximately 47.3% — a level that is materially higher than would be expected for undifferentiated commodity components and that reflects customers' willingness to pay premium pricing driven by the patented features;
• OWS's patent portfolio (142 issued U.S. patents and 58 pending applications) evidences a sustained pattern of innovation of which the Asserted Patents are core components; and
• Ridgeline Capital Partners' 2019 acquisition of a 55% equity stake for $780 million reflects sophisticated financial-market confidence in OWS's technology portfolio and competitive position.
The financial health of the enterprise is confirmed by the FY 2023 audited financial statements (Gresham & Whitfield LLP, San Jose office), which reflect $61.4 million of net income on the aforementioned $487.3 million of revenue.
B. Industry Recognition and Adoption
OWS's patent-practicing transceiver modules have been qualified into production networks at multiple leading hyperscale data-center operators. Design-in wins are typically accompanied by multi-year supply agreements reflecting substantial expected volumes. In addition, OWS engineers actively participate in relevant standardization bodies and in industry forums (including the OIF, IEEE 802.3, and OFC), where technical approaches consistent with the innovations claimed in the Asserted Patents have been broadly cited and referenced.
C. Nexus Between Commercial Success and Patented Features
The commercial success described above is not attributable to unpatented features, marketing programs, or promotional pricing. To the contrary, based on my review of the record, my discussions with OWS personnel (including Dr. Jian-Wei Huang, OWS CTO), and my understanding of the purchasing behavior of hyperscale customers, I have concluded that hyperscale customers make silicon-photonics transceiver purchasing decisions principally on the basis of specific technical performance attributes — attributes that map directly to the Asserted Patents:
• Low insertion loss in the waveguide multiplexer → '847 Patent;
• Bandwidth, sensitivity, and reach of the integrated coherent transceiver → '193 Patent;
• Stable operation across thermal excursions and dense integration densities → '562 Patent; and
• Efficient, polarization-insensitive fiber-to-chip coupling → '015 Patent.
Those attributes correspond directly to the claimed inventions of the Asserted Patents and constitute the "basis for customer demand" for the transceiver modules at issue. This nexus opinion is directly relevant to the entire-market-value and apportionment analyses developed by Dr. Keating in his damages report.
VII. LONG-FELT BUT UNSOLVED NEED
A. Industry Demand for Each Innovation
Prior to the respective priority dates of the Asserted Patents, the silicon-photonics industry had recognized — and repeatedly failed to satisfy — the technical needs that the Asserted Patents ultimately addressed:
Low-insertion-loss waveguide multiplexing (pre-2016). By the mid-2010s, industry roadmaps and standards bodies had explicitly called for on-chip multiplexers with materially lower insertion loss than then-available AWG, echelle, and ring-resonator approaches. Numerous IEEE and OFC papers of that era discussed the insertion-loss barrier as a limiting factor in the transition to higher-channel-count, higher-speed transceivers.
Integrated coherent transceivers on silicon (pre-2017). The industry-wide preference for monolithic or near-monolithic integration of coherent transceivers on silicon was longstanding by 2017. Hybrid integration and heterogeneous bonding approaches had been pursued for over a decade with mixed results, and design engineers routinely identified the lack of a manufacturable, high-performance integrated coherent architecture on silicon as a first-order impediment to further transceiver-module cost and performance improvements.
Thermal management of photonic integrated circuits (pre-2014). By 2013–2014, thermal management had emerged as a recognized industry-wide pain point. The trend toward denser PIC integration and higher per-die optical power had outstripped the capabilities of conventional package-level TEC and heat-sink approaches, and industry-conference discussions consistently highlighted the need for chip-level thermal-management solutions that would not consume excessive power or drive prohibitive package complexity.
Polarization-diverse grating coupling (pre-2019). Polarization sensitivity of one-dimensional grating couplers had been a well-known limitation since the earliest commercial deployment of grating-coupler-based silicon-photonics devices. The industry had long sought a compact, integrable polarization-diverse coupler configuration that would eliminate reliance on external polarization controllers.
B. Evidence of Long-Felt Need
Evidence of long-felt need is documented in, among other sources, contemporaneous IEEE Journal of Lightwave Technology articles and Photonics Technology Letters, OFC and ECOC conference proceedings, industry roadmaps (including the Optical Internetworking Forum roadmaps of the relevant periods), and customer RFIs and RFQs which explicitly requested features corresponding to the claimed innovations. In addition, I draw on my firsthand experience at Solara Optics during the 2013–2014 period, when customers repeatedly identified thermal management of PICs as a top-priority industry need — a set of customer requests that ultimately motivated the invention that became the '562 Patent.
VIII. FAILURE OF OTHERS
A. Competitor Efforts and Their Failures
Based on my personal knowledge of the industry and my review of the technical literature, at least three competitors attempted — and failed — to achieve one or more of the technical objectives that the Asserted Patents ultimately accomplished. In deference to confidentiality obligations and the ongoing character of the litigation, I refer to these competitors as Competitor A, Competitor B, and Competitor C. Their identities are itemized in Appendix A and are known to counsel.
• Competitor A pursued a multi-channel AWG-based multiplexer design in the 2014–2016 timeframe. Despite significant R&D investment, its published and demonstrated insertion-loss performance was materially inferior to what OWS's '847 Patent achieves, and it ultimately withdrew that program.
• Competitor B attempted heterogeneous integration of coherent detectors on silicon during 2015–2018 using die-to-wafer bonding techniques. Its efforts encountered yield, alignment, and thermal-mismatch challenges that were not overcome to a commercially viable degree.
• Competitor C pursued micro-channel thermal-management approaches for PICs in the 2013–2015 timeframe but was unable to solve the fluid-sealing and CMOS-process-compatibility problems that the '562 Patent's approach addresses.
These failures reflect fundamental design trade-offs and fabrication constraints — not any lack of skill or investment. In each case, the competitor employed technically sophisticated engineering teams and expended substantial R&D resources.
B. Significance of Others' Failures
The failure of skilled competitors to achieve, through independent effort, the technical results that the Asserted Patents claim corroborates both the non-obviousness of the claimed inventions and their commercial significance. The fact that skilled teams could not, despite substantial investment, converge on the claimed solutions weighs heavily against any suggestion that those solutions were obvious to a person of ordinary skill in the art. My personal knowledge of these competitor efforts — gained during my tenure at Solara Optics, where such competitive intelligence was routinely developed and maintained — corroborates the conclusions I have drawn from the public record.
IX. EVIDENCE OF COPYING BY LUMICORE
A. Lumicore's Accused Product Lines and Launch Timeline
Lumicore has launched three accused product families during the period beginning Q3 2022:
Lumicore also maintains a sales and support office within the Eastern District of Texas at 2700 Earl Campbell Parkway, Suite 150, Tyler, Texas 75701.
B. Structural Similarity to OWS's Patented Designs
The Benchmark Semiconductor Analysis, LLC teardowns provide detailed, physical-evidence-based analysis of the internal architecture of Lumicore's accused products. Each teardown was performed and documented by Kevin Ng, Lead Engineer at Benchmark (2244 Mission College Boulevard, Santa Clara, CA 95054):
• The December 12, 2022 teardown of the LumiLink 400G-ZR+ confirmed that the module employs waveguide-multiplexer structures and thermal-management micro-channels that closely parallel the claim elements of the '847 and '562 Patents;
• The June 2023 teardown of the LumiLink 800G-DR8 revealed additional structural correspondence to claim elements of the '193 and '015 Patents; and
• The February 2024 teardown of the LumiEdge 1600G-LR revealed structures corresponding to claim elements of the '193 and '015 Patents.
Element-by-element infringement mapping is developed in the report of Dr. Caroline Ashford (Professor of Electrical Engineering & Photonics, Northfield Institute of Technology, Evanston, IL). While infringement analysis is the province of Dr. Ashford's report, the structural similarities identified in the Benchmark teardowns are, in my professional judgment, sufficiently pronounced that they cannot plausibly be explained as independent convergent design.
C. Lumicore's Awareness of OWS's Patents During Design
The record contains multiple, mutually corroborating indicia of Lumicore's design-phase awareness of OWS's patents:
• Deposition of Dr. Sandra Michaud (Lumicore CTO), November 19, 2024. Dr. Michaud testified that she reviewed OWS's published patent application No. 2018/0123456 (corresponding to the '193 Patent) during the design phase of the LumiLink 400G-ZR+ product.
• Anticipated testimony of Melissa Chen (former Lumicore engineer, now with Veridian Photonics, Inc., California). I am informed that Ms. Chen is expected to testify that Lumicore's LumiLink 400G-ZR+ design team accessed and reviewed OWS's published patent applications during the 2021–2022 product-design phase.
• Deposition of Raymond Bautista (Lumicore General Counsel), December 3, 2024. Mr. Bautista admitted that no formal design-around analysis was commissioned before the launch of the LumiEdge 1600G-LR in Q4 2023 — a launch that occurred well after Lumicore had received specific pre-suit notice of the Asserted Patents.
• FairPoint Licensing Group correspondence and April 20, 2023 meeting. Lumicore received a January 15, 2023 licensing letter from FairPoint Licensing Group, LLC (Managing Director Angela Reeves, 500 Sansome Street, Suite 1400, San Francisco, CA 94111) that identified the '847, '193, and '562 Patents; a May 15, 2023 supplemental letter added the '015 Patent and the LumiLink 800G-DR8. An in-person licensing meeting was held on April 20, 2023 in Austin, Texas, at which Angela Reeves and Rebecca Tanaka (OWS General Counsel) met with Raymond Bautista and Catherine ("Kate") Rourke (lead partner, Hawthorne Vega & Cruz LLP, 600 Congress Avenue, Suite 2400, Austin, TX 78701). On July 10, 2023, Lumicore declined a license, citing unspecified "competent legal opinions" that it declined to produce.
D. Significance of the Copying Evidence
Copying is a well-recognized secondary consideration of non-obviousness. Here, the combination of (i) documented design-phase review of OWS's published patent applications, (ii) structural correspondence between the accused products and the patent claims as documented in the Benchmark teardowns, (iii) the absence of any documented design-around analysis, and (iv) Lumicore's continued launch of new infringing products (most notably the LumiEdge 1600G-LR in Q4 2023) after receipt of specific pre-suit notice, together support the reasoned inference that Lumicore deliberately incorporated OWS's patented innovations into its accused products.
X. COMPETITIVE IMPACT OF LUMICORE'S ACCUSED PRODUCTS ON OWS
A. Price Erosion
Average selling prices for 400G-ZR-class transceivers declined approximately 18.7% between Q3 2022 (when the LumiLink 400G-ZR+ launched) and Q4 2023. While a portion of that decline is attributable to normal industry cost-curve dynamics, a material portion is directly attributable to Lumicore's competitive entry with infringing modules that competed head-to-head against OWS's OWS-400ZR product line. Analogous pricing pressure has appeared in the 800G segment following the Q1 2023 launch of the LumiLink 800G-DR8.
B. Customer Diversion and Lost Design-In Opportunities
OWS and Lumicore compete head-to-head at 12 of the top 20 hyperscale operators. The record contains, and is expected to contain further, evidence of specific customer diversion events — including diversion at customers where OWS held the prior design-in and Lumicore's accused-product performance and price were determinative of the customer's decision to shift or split volumes. These events will be developed in further detail in declarations from OWS's VP of Sales, Karen Whitfield, and OWS's regional sales directors.
The design-in cycle dynamics compound this harm. As noted, hyperscale qualification cycles typically run 18 to 24 months; once a supplier is designed in, incumbency confers a substantial durable advantage. Lumicore itself estimates that, in the event of a permanent injunction, redesign of its accused product lines to non-infringing form would require 18 to 24 months, during which the accused products could not be sold — a self-assessment that confirms the difficulty of switching within the design cycle and the value of the design-in positions Lumicore has captured. Lumicore has publicly attributed approximately 42% of its product revenue to the accused product lines.
C. Market Dynamics Supporting Lost-Profits Analysis
My market-share displacement analysis proceeds in three principal steps:
(1) Identification of head-to-head competition at the customer level. Both OWS and Lumicore are qualified vendors at 12 of the top 20 hyperscale operators, and 7 of the top 20 have purchased from both suppliers during FY 2023.
(2) Allocation of accused-product sales between (a) sales that displaced OWS and (b) sales to customers where OWS was not a realistic alternative (whether because OWS is not qualified there, or because volume or timing constraints foreclosed the opportunity).
(3) Calculation of the market-share displacement percentage.
The result of that analysis is a displacement figure of approximately 38.2%. Total accused-product revenue during the infringement period from Q3 2022 through Q1 2024 is approximately $378.2 million. Dr. Keating's preliminary projection of accused-product revenue through trial (through Q4 2025) totals approximately $714.6 million.
Applying the Panduit framework, I have separately analyzed the availability of acceptable non-infringing alternatives (Panduit factor 2). Based on my review of four candidate alternative products (identified in Appendix A), each lacks one or more of the key patented features and, in my professional judgment, does not represent an acceptable non-infringing alternative for the accused-product performance envelope.
D. Industry-Level Impact
Beyond the direct impact on OWS, Lumicore's infringing entry has broader implications for innovation incentives in the U.S. silicon-photonics industry. R&D investment in silicon photonics is capital-intensive and long-cycle; recoupment of that investment depends critically on the ability of innovators to appropriate returns from their inventions during the period of exclusivity that patents provide. Unchecked infringement of foundational silicon-photonics patents by well-resourced competitors erodes the incentives on which the sector's continued innovation depends.
XI. OPINION ON THE PATENTED FEATURES AS BASIS FOR CUSTOMER DEMAND
A. The Patented Features Drive Demand for Transceiver Modules
Hyperscale data-center customers evaluate silicon-photonics transceiver modules against a compact set of technical performance criteria. Based on my experience, my discussions with OWS engineering and sales personnel, and my review of the record, the four technical attributes that most consistently drive purchasing decisions in this market are:
(i) low-insertion-loss on-chip waveguide multiplexing (corresponding to the '847 Patent);
(ii) integrated coherent transceiver capability (corresponding to the '193 Patent);
(iii) effective chip-level thermal management enabling stable, dense integration ('562 Patent); and
(iv) polarization-diverse fiber-to-chip grating coupling ('015 Patent).
Each of those attributes corresponds directly to the claimed inventions of the respective Asserted Patents. In my professional judgment, they are the basis for customer demand for the transceiver modules at issue in this action.
B. Relevance to Apportionment and Damages
The transceiver module is, in my view, the smallest salable patent-practicing unit for these Asserted Patents. The patented features are physically and functionally integral to the module and cannot meaningfully be disaggregated from the module for pricing purposes; hyperscale customers purchase, qualify, and deploy transceiver modules as complete units.
The nexus between the patented features and the basis for customer demand supports the application, in Dr. Keating's damages analysis, of module-level pricing to the entire-market-value inquiry. Dr. Keating's preliminary lost-profits model — $128.9 million, based on $714.6 million of projected accused-product revenue, a 38.2% market-share displacement rate, and OWS's 47.3% gross margin — and his reasonable-royalty analysis, applying an 8.5% royalty rate to the remaining approximately $441.6 million of non-lost-profit accused sales (yielding approximately $37.5 million), rest on the nexus opinion set forth above.
XII. RESPONSE TO ANTICIPATED DEFENSE ARGUMENTS
A. Prior Art and State of the Art
I have reviewed the invalidity contentions served by Lumicore on October 14, 2024, which identify 23 prior-art references (prepared with the assistance of Marlowe Asset Management Patent Analytics, Inc., 1100 Peachtree Street NE, Suite 2200, Atlanta, GA 30309). While a full validity analysis is outside the scope of my report and is developed by Dr. Ashford, I address here the extent to which certain of those references may bear on my commercial-significance opinions:
• Takahashi et al., "Dense WDM Silicon Photonic MUX" (IEEE Photonics Technology Letters, March 2014). Cited against the '847 Patent for anticipation. In my view, Takahashi describes a general class of dense-WDM silicon photonic multiplexer structures but does not disclose the specific low-insertion-loss configuration that the '847 Patent claims. Its existence in the literature is fully consistent with — indeed, illustrative of — the pre-invention state of the art that motivated the innovation of the '847 Patent.
• U.S. Patent No. 8,412,007 (Bergmann) (issued April 2, 2013). Cited against the '562 Patent for obviousness. Bergmann does not teach embedded micro-channel heat sinks for PICs; its disclosed thermal-management approaches are of a conventional character that the '562 Patent expressly improves upon.
• Dr. Li Wei, "Thermal Solutions for PICs" (OFC 2013 conference paper, March 2013). Cited against the '562 Patent for anticipation. Li Wei describes conventional thermal approaches (heat sinks, TECs), not the embedded micro-channel configuration claimed in the '562 Patent.
• Lumicore internal design documents (2016–2017). Cited to attack the priority date of the '015 Patent. I am not aware of any contemporaneous industry knowledge of these documents, which are — by definition — internal to Lumicore. Their reliability, dating, and corroboration are subject to significant challenge, but that is a matter for the validity experts.
None of these references, in my professional judgment, undermines the opinions I have offered regarding long-felt need, failure of others, commercial success, or nexus.
B. Availability of Non-Infringing Alternatives
I have separately analyzed four candidate non-infringing alternatives that Lumicore has identified or that could plausibly be identified from the state of the art. Each such candidate lacks one or more of the key patented features, with corresponding performance shortfalls in insertion loss, integration density, thermal stability, or polarization efficiency. On the basis of that analysis, I conclude that no commercially viable non-infringing alternative exists that matches the performance of OWS's patented technologies.
XIII. CONCLUSIONS
Based on the analysis set forth above, I offer the following conclusions, each rendered to a reasonable degree of certainty within the field of silicon-photonics and optical interconnect technology:
1. The silicon-photonics transceiver market is large, growing, and strategically important.