1014-00004 · Hawthorne Vega & Cruz LLP · Damages Analysis

apportionment-analysis.docx DOCUMENT Litigation

28 / 168 ← → ✕

MEMORANDUM — PRIVILEGED AND CONFIDENTIAL ATTORNEY WORK PRODUCT

This memorandum constitutes attorney work product prepared in anticipation of litigation and is protected by the work-product doctrine and attorney-client privilege. It should not be disclosed to any person outside the Calderwood & Harkness LLP litigation team without prior authorization from Alan Ngo.

TO: Alan Ngo, Responsible Partner; Marcus Delgado, Billing Partner

FROM: Wendy Schultz, Lead Associate; Omar Siddiqui, Working Associate

DATE: August 22, 2024

RE: Apportionment Analysis — Smallest Salable Patent-Practicing Unit (SSPPU) and Entire Market Value Rule (EMVR) as Applied to Accused Lumicore Transceiver Modules

MATTER: Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc., Case No. 2:24-cv-00319-JRG (E.D. Tex.)

Right-click to update Table of Contents

I. EXECUTIVE SUMMARY

This memorandum analyzes whether the appropriate royalty base for calculating reasonable-royalty damages under Theory 2 of our damages framework (applying the Georgia-Pacific fifteen-factor analysis) should be the price of the entire accused transceiver module — invoking the Entire Market Value Rule ("EMVR") — or a disaggregated sub-component thereof, under the Smallest Salable Patent-Practicing Unit ("SSPPU") approach. The memorandum is intended to provide the analytical framework that our retained damages expert, Dr. Lawrence Keating of Meridian Economic Consulting Group, LLC, should employ in his expert damages report, which is due on June 15, 2025.

Primary Conclusion. We conclude that the EMVR is properly applied to the module-level selling price of each of the three accused Lumicore products — the LumiLink 400G-ZR+, the LumiLink 800G-DR8, and the LumiEdge 1600G-LR — because the patented features collectively constitute the basis for customer demand for the accused modules. Specifically, the waveguide multiplexer technology claimed in U.S. Patent No. 10,234,847 (the "'847 Patent"), the coherent transceiver integration architecture claimed in U.S. Patent No. 10,891,193 (the "'193 Patent"), the micro-channel thermal management system claimed in U.S. Patent No. 9,678,562 (the "'562 Patent"), and the polarization-diverse grating coupler claimed in U.S. Patent No. 11,402,015 (the "'015 Patent") are the core functional elements that differentiate the accused products in the marketplace and that customers identify as the primary basis for their purchasing decisions.

Alternative SSPPU Analysis. Even if the Court requires the application of a SSPPU analysis rather than the EMVR, the smallest salable patent-practicing unit is the transceiver module itself — not a sub-component. The patented innovations are not sold or priced separately; they are physically and commercially integrated into the photonic integrated circuit ("PIC") die, which is in turn only commercially available as part of the fully assembled transceiver module. Lumicore does not sell PIC dies on a standalone basis, and the individual patented structures within the PIC die (e.g., the waveguide multiplexer, the grating coupler) are not separately manufactured, sold, or priced in any commercial market.

Quantitative Impact. The proposed royalty rate of 8.5% should be applied to total accused-product net sales revenues. For Fiscal Year 2023, total accused-product revenue was $264.9 million, comprising $94.8 million for the LumiLink 400G-ZR+, $128.4 million for the LumiLink 800G-DR8, and $41.7 million for the LumiEdge 1600G-LR. Projected total accused-product revenue through trial (covering the full infringement period of Q3 2022 through Q4 2025) is approximately $714.6 million. After subtracting the portion of revenue subject to the lost-profits claim under Theory 1 (38.2% market-share displacement), the reasonable-royalty base is approximately $441.6 million, yielding a reasonable royalty of approximately $37.5 million.

Anticipated Defense Strategy. We anticipate that Lumicore and its damages expert, Dr. Natasha Volkov of Pinnacle Analytics Partners LLP, will argue for a disaggregated royalty base — for example, limiting the base to the estimated PIC die cost alone, or to a hypothetical imputed value for the waveguide multiplexer sub-component — in an effort to artificially reduce the damages award. This memorandum provides the analytical foundation to defeat that argument at the Daubert stage, at summary judgment, and at trial.

II. BACKGROUND

A. The Parties and Accused Products

1. Optiwave Semiconductor Corp.

Optiwave Semiconductor Corp. ("OWS" or "Optiwave") is a Delaware corporation headquartered at 4810 Bayside Technology Parkway, Suite 300, Santa Clara, California 95054. OWS is a fabless semiconductor company specializing in silicon photonics chipsets for data-center interconnects. According to audited financial statements prepared by Gresham & Whitfield LLP, OWS reported Fiscal Year 2023 revenue of $487.3 million and net income of $61.4 million. OWS employs approximately 1,250 persons across its Santa Clara headquarters, its Austin design center, and its Singapore operations office.

OWS is the third-largest U.S. supplier of silicon-photonics transceiver modules for hyperscale data centers. The company designs its PIC dies in-house and contracts fabrication to third-party foundries, with final module assembly performed at its contract manufacturing partners' facilities. OWS's intellectual property portfolio consists of 142 issued United States patents and 58 pending applications, spanning silicon photonics, waveguide design, thermal management, optical coupling, and coherent transceiver architecture.

Ridgeline Capital Partners, LP acquired a 55% majority stake in OWS in 2019 for $780 million, and Ridgeline remains the majority owner. The minority ownership is held by a combination of early-stage venture investors and current and former OWS management.

OWS's key competing products — that is, the products that compete directly with the accused Lumicore modules in the marketplace — are the OWS-400ZR, OWS-800XR, and OWS-1600DR transceiver module families. OWS's gross margin on these competing products is 47.3%, a figure that will be significant for the lost-profits analysis under Theory 1.

2. Lumicore Technologies, Inc.

Lumicore Technologies, Inc. ("Lumicore" or "LTC") is a Delaware corporation headquartered at 9200 Spectrum Boulevard, Building C, Austin, Texas 78759. Lumicore completed its initial public offering on the Nasdaq exchange in March 2022 at an offering price of $24.00 per share. As of the filing date of this action, Lumicore's common stock was trading at approximately $37.00 per share, implying a market capitalization of approximately $3.4 billion based on 91.8 million shares outstanding. Per Lumicore's Form 10-K filed with the Securities and Exchange Commission, the company reported Fiscal Year 2023 revenue of $312.6 million. Lumicore employs approximately 870 persons.

Lumicore designs and sells optical transceiver modules for data-center and telecommunications applications. Like OWS, Lumicore is a fabless semiconductor company; it designs its PIC dies in-house and uses foundry partners for fabrication. Lumicore's assembly operations are conducted at a facility in Austin, Texas.

3. Accused Product Lines

The Complaint asserts infringement against three Lumicore product families, each of which is an integrated optical transceiver module sold to hyperscale data-center operators and telecommunications equipment manufacturers. The accused products, their launch dates, Fiscal Year 2023 revenues, and asserted patents are summarized below:

Total accused-product revenue over the infringement period from product launch through Q1 2024 (the most recent quarter for which actual revenue data has been produced in discovery) is approximately $378.2 million, per Dr. Keating's preliminary analysis. Based on Dr. Keating's revenue projections, which incorporate Lumicore's publicly reported growth rates and analyst estimates, projected total accused-product revenue through trial (Q3 2022 through Q4 2025) is approximately $714.6 million.

It bears emphasis that the accused products are among Lumicore's highest-revenue product lines, collectively accounting for approximately 84.7% of Lumicore's total FY 2023 revenue.

B. The Asserted Patents and Patented Technology

1. U.S. Patent No. 10,234,847 (the "'847 Patent")

The '847 Patent is titled "Multi-Channel Silicon Photonic Waveguide Multiplexer with Low Insertion Loss." It was filed on September 14, 2016, and issued on March 19, 2019. The original and current assignee is Optiwave Semiconductor Corp. The Complaint asserts infringement of Claims 1, 3, 5, 9, and 14 of the '847 Patent.

The '847 Patent claims a multi-channel waveguide multiplexer structure fabricated on a silicon photonic substrate that combines multiple optical channels onto a single waveguide with significantly reduced insertion loss compared to prior-art multiplexer designs. The claimed structure employs a novel adiabatic coupler geometry and optimized waveguide taper profiles that minimize optical signal loss at each channel-combining junction. This technology is fundamental to achieving the high data rates (400G, 800G, and 1600G) specified for the accused products.

2. U.S. Patent No. 10,891,193 (the "'193 Patent")

The '193 Patent is titled "High-Bandwidth Coherent Optical Transceiver Integrated on Silicon Substrate." It was filed on April 7, 2017, and issued on January 12, 2021. The '193 Patent shares Provisional Application No. 62/499,117 with the '015 Patent, establishing a common priority date for the two inventions. The Complaint asserts infringement of Claims 1, 3, 7, 10, 15, and 18 of the '193 Patent.

The '193 Patent claims a coherent optical transceiver architecture that is monolithically integrated on a silicon substrate, enabling detection of both the amplitude and phase of optical signals. This coherent detection capability dramatically increases data capacity compared to legacy intensity-modulation / direct-detection ("IM/DD") transceivers. The patented architecture includes on-chip local oscillator generation, balanced photodetector arrays, and digital signal processing interfaces, all integrated within a single silicon photonic die.

3. U.S. Patent No. 9,678,562 (the "'562 Patent")

The '562 Patent is titled "Method for Thermal Management of Photonic Integrated Circuits Using Embedded Micro-Channel Heat Sinks." It was filed on November 22, 2014, and issued on June 13, 2017. The '562 Patent was originally assigned to Solara Optics, Inc. and was acquired by OWS pursuant to an asset purchase agreement dated June 15, 2018. The total purchase price for the asset acquisition was $34.5 million, covering a portfolio of 12 patents, including the '562 Patent. The '562 Patent has a continuation relationship with the '847 Patent, as both derive from the same patent family. The Complaint asserts infringement of Claims 1, 4, 6, and 11 of the '562 Patent.

The '562 Patent claims a method for managing the thermal output of photonic integrated circuits by embedding micro-channel heat sinks directly within or adjacent to the silicon substrate. The claimed micro-channel geometry and fluid-flow characteristics enable highly efficient heat dissipation from the photonic structures without requiring bulky external heat sinks, which would increase the module's form factor beyond industry-standard specifications. This thermal management technology is essential for high-power PIC operation in the thermally constrained environments typical of data-center rack-mount deployments.

4. U.S. Patent No. 11,402,015 (the "'015 Patent")

The '015 Patent is titled "Polarization-Diverse Grating Coupler for Silicon Photonic Transceivers." It was filed on August 30, 2019, and issued on August 2, 2022. As noted above, the '015 Patent shares Provisional Application No. 62/499,117 with the '193 Patent. The Complaint asserts infringement of Claims 1, 5, 8, 12, and 16 of the '015 Patent.

The '015 Patent claims a grating coupler structure that efficiently couples light of both orthogonal polarization states into and out of the photonic integrated circuit. Prior-art grating couplers were polarization-sensitive, meaning they could efficiently couple only one polarization state, resulting in significant optical power loss for the orthogonal polarization. The claimed polarization-diverse design uses a two-dimensional grating pattern and sub-wavelength structuring to achieve near-equal coupling efficiency for both polarization states, which is essential for achieving the optical performance specifications of modern coherent transceivers.

5. Technical Summary

The four asserted patents collectively cover the core functional elements of a silicon-photonics transceiver module: (1) the waveguide multiplexer that combines multiple optical channels with low insertion loss ('847 Patent); (2) the coherent transceiver architecture integrated on a silicon substrate that enables high-bandwidth data transmission ('193 Patent); (3) the embedded thermal management system using micro-channel heat sinks that ensures reliable high-power operation ('562 Patent); and (4) the polarization-diverse grating coupler that efficiently couples light into and out of the photonic integrated circuit ('015 Patent). Together, these innovations enable high-bandwidth, low-power, thermally stable optical transceiver modules suitable for deployment in hyperscale data-center interconnect applications. The four patented technologies are not peripheral or incremental improvements; they define the core architecture and performance characteristics of the accused transceiver modules.

C. Damages Theories Overview

OWS's damages case comprises three theories, as follows:

Theory 1: Lost Profits Under *Panduit*. OWS seeks lost profits on the portion of Lumicore's accused-product sales that displaced OWS's own sales of competing products. The lost-profits claim is projected at $128.9 million through trial, calculated as follows: $714.6 million (projected total accused-product revenue) × 38.2% (market-share displacement factor) × 47.3% (OWS gross margin on competing products) = approximately $128.9 million. The 38.2% market-share displacement figure is derived from the Market Share Displacement Analysis prepared by this team in June 2024, which analyzed the relevant market, the two-supplier market structure, and the absence of non-infringing substitutes.

Theory 2: Reasonable Royalty Under *Georgia-Pacific*. For the portion of Lumicore's accused-product sales not subject to the lost-profits claim — i.e., sales that would not have been captured by OWS even in the absence of infringement — OWS seeks a reasonable royalty. The reasonable-royalty calculation is as follows: $714.6 million (projected total accused-product revenue) minus $272.98 million (the 38.2% of revenue subject to the lost-profits claim) yields a non-lost-profit revenue base of approximately $441.6 million. Applying the proposed royalty rate of 8.5% produces a reasonable royalty of approximately $37.5 million. This memorandum focuses primarily on Theory 2 and the apportionment analysis that underlies the royalty-base determination. However, as discussed in Section VI below, the apportionment analysis is also relevant to Theory 1.

Theory 3: Enhanced Damages for Willful Infringement. OWS will seek enhanced damages under 35 U.S.C. § 284, based on Lumicore's continued infringement after receiving actual notice of the asserted patents. Post-notice lost profits are estimated at approximately $109.2 million; various multiplier scenarios (1.5x, 2x, 3x) will be presented at trial. The willfulness analysis is addressed in a separate memorandum dated July 2024.

Total Damages Demand. The total damages demand, as set forth in the Complaint, is approximately $210.0 million (blended and rounded).

The apportionment analysis set forth in this memorandum will be communicated to Dr. Keating for incorporation into his expert damages report. Dr. Keating's report will present the royalty-base and royalty-rate analysis; this memorandum provides the legal framework and factual support for his methodology.

III. LEGAL FRAMEWORK

A. The Default Rule: The Smallest Salable Patent-Practicing Unit (SSPPU)

The Federal Circuit has established a strong preference for tying patent-damages calculations to the smallest salable unit that embodies the patented invention, rather than to an entire multi-component product. This preference, articulated most forcefully in LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51 (Fed. Cir. 2012), is designed to ensure that reasonable-royalty damages reflect only the value of the patented feature, not the value of unrelated features that contribute to the overall product's market price.

The rationale for the SSPPU doctrine is straightforward: where a patent covers one feature of a complex, multi-component product, using the entire product's selling price as the royalty base risks attributing to the patentee economic value that is properly attributable to non-patented features. As the Federal Circuit explained in Uniloc USA, Inc. v. Kestrel Partners LLC, 632 F.3d 1292, 1320 (Fed. Cir. 2011), damages methodologies that employ the entire value of a multi-component product as a starting point risk misleading the jury into awarding damages based on unpatented features. In Uniloc, the court rejected the so-called "25 percent rule of thumb" as an analytically unsound method of approximating a reasonable royalty, emphasizing the need for a rigorous, evidence-based apportionment of the royalty to the patented contribution.

In VirnetX, Inc. v. Sterling Reach Holdings, 767 F.3d 1308 (Fed. Cir. 2014), the Federal Circuit reinforced the SSPPU requirement, holding that a damages expert must "carefully tie proof of damages to the claimed invention's footprint in the market place" and must "apportion the defendant's profits and the patentee's damages between the patented feature and the unpatented features." Id. at 1326. The VirnetX court vacated a jury verdict where the patentee's damages expert had used the total value of an iPhone or other accused device as the royalty base without adequate evidence that the patented features — VPN and FaceTime functionalities — were the basis for consumer demand for the entire device.

The SSPPU analysis requires identification of the smallest commercial unit that practices the asserted patent claims. In many cases, this will be a sub-component of the accused end product. For example, in LaserDynamics, the Federal Circuit held that the smallest salable patent-practicing unit for a patent covering an optical disc discrimination method was the optical disc drive ("ODD"), not the laptop computer into which the ODD was installed. 694 F.3d at 67-68. Because the ODD was a discrete, separately sold component, using the laptop price as the royalty base was impermissible absent satisfaction of the EMVR requirements.

It is critical to understand, however, that the SSPPU doctrine is not a rigid formula but rather a principle of analytical discipline. The Federal Circuit has recognized that the identification of the SSPPU depends on the facts of each case, including the structure of the accused product, the commercial practices of the parties, and the relationship between the patented technology and the accused product's functionality. As we discuss below, in cases involving highly integrated products — such as semiconductor devices and photonic integrated circuits — the SSPPU analysis may lead to the conclusion that the entire chip or module is the smallest salable patent-practicing unit.

B. The Exception: The Entire Market Value Rule (EMVR)

The EMVR provides an exception to the general SSPPU preference. Under the EMVR, a patentee may use the entire market value of a multi-component product as the royalty base where the patented feature constitutes the "basis for customer demand" for the entire product, or where the patented feature is the "predominant basis for consumer demand" or "substantially creates the value of the component parts."

The EMVR has a long pedigree in patent-damages jurisprudence. In Rite-Hite Corp. v. Kelley, Inc., 56 F.3d 1538 (Fed. Cir. 1995) (en banc), the Federal Circuit sitting en banc permitted entire-market-value recovery for unpatented components sold with a patented device, where the patented invention served as the basis for demand for the overall product. The en banc court recognized that in many commercial settings, customers purchase integrated products precisely because of a particular patented feature, and limiting the damages base to a sub-component would under-compensate the patentee for the full economic impact of the infringement.

In Lucent Technologies, Inc. v. Gateway, Inc., 580 F.3d 1301 (Fed. Cir. 2009), the court clarified that the EMVR requires affirmative proof — not mere assertion — that the patented feature is the driver of customer demand for the entire product. In Lucent, the court vacated a $1.538 billion verdict where the patentee had used the total sales revenue of Hollowell Systems as the royalty base for a patent covering a date-picker tool, without adequate evidence that the date-picker tool drove demand for the outlook software. The court held that "the base used in a running royalty calculation can always be the value of the entire commercial embodiment, as long as the magnitude of the rate is within an acceptable range (as determined by the evidence)," but emphasized that the selection of royalty base and rate must be grounded in evidence, not speculation. Id. at 1338-39.

LaserDynamics further articulated the EMVR standard, holding that the EMVR "requires proof that the patented feature is the basis for customer demand for the entire product," not merely an important feature among several. 694 F.3d at 67. The court explained that applying the EMVR to a multi-component product when the patented feature does not drive demand for the product as a whole risks inflating the damages award by capturing value attributable to unpatented features.

An important refinement was introduced in Commonwealth Scientific & Industrial Research Organisation v. Sterling Reach Holdings, 809 F.3d 1295 (Fed. Cir. 2015) ("CSIRO"). In CSIRO, the Federal Circuit recognized two permissible approaches when the EMVR is invoked:

(1) The patentee may use the entire product price as the royalty base and apply a royalty rate that specifically accounts for the contribution of the patented feature (i.e., apportioning through the rate rather than the base); or

(2) The patentee may directly demonstrate that the patented feature is the basis for customer demand for the entire product, thereby justifying use of the full product price as the base without further reduction.

The CSIRO court emphasized that the overriding requirement is that the damages calculation must "be carefully tied to the claimed invention's footprint in the market place" and must not result in a damages award that exceeds the value attributable to the patented technology. 809 F.3d at 1301. Critically, the CSIRO court held that even when the EMVR is invoked, the royalty rate itself can be adjusted downward to account for the value contribution of non-patented features — the so-called "rate-apportionment" approach. This means that the EMVR does not necessarily lead to inflated damages if the rate is properly calibrated to reflect the patented technology's proportional contribution to the product's value.

This rate-apportionment approach is the methodology that Dr. Keating will employ in this case: using the module-level selling price as the royalty base and applying a royalty rate of 8.5% that has been derived from comparable license analysis and that inherently reflects the proportion of the module's value attributable to the patented features.

C. The Interplay Between SSPPU and EMVR in the Context of Integrated Products

The Federal Circuit has recognized that in cases involving highly integrated products — particularly semiconductor chips, integrated circuits, and other devices where the patented technology is physically inseparable from the product — the SSPPU may in fact be the entire chip or module. This is because the patented technology is not separately sold or separately priceable; it exists only as an integral part of the assembled product.

In Finjan, Inc. v. Blue Coat Systems, Inc., 879 F.3d 1299 (Fed. Cir. 2018), the Federal Circuit permitted the use of revenue from the entire accused product as the royalty base where the patented security feature was "inseparable from the product's function." The Finjan court recognized that when a patented feature is so deeply integrated into a product that it cannot be separated from the product's core functionality, requiring disaggregation of the royalty base to some hypothetical sub-component would introduce speculative and unreliable elements into the damages calculation.

This line of cases is critical to our analysis. The present case involves patented features — a waveguide multiplexer, a coherent transceiver architecture, a micro-channel heat sink system, and a polarization-diverse grating coupler — that are monolithically fabricated into a photonic integrated circuit die, which is in turn assembled into a transceiver module. The patented structures are not discrete components that can be removed from the PIC die and sold or valued independently. They are integral parts of the silicon substrate itself, fabricated during the same lithographic and deposition processes that create the entire PIC. In this respect, the present case is fundamentally distinguishable from cases like LaserDynamics, where the patented feature was embodied in a discrete, separately sold component (the optical disc drive) that was installed into a larger product (the laptop computer). Here, there is no analogous separable component. The patented features and the PIC die are one and the same, and the PIC die and the transceiver module are commercially inseparable.

Accordingly, the SSPPU analysis and the EMVR analysis converge in this case: (1) the SSPPU is the transceiver module because neither the PIC die nor its constituent patented structures are separately sold or priced; and (2) the EMVR is independently satisfied because the patented features are the basis for customer demand for the transceiver module as a whole. Either analytical path leads to the same result: the module-level selling price is the appropriate royalty base.

IV. APPLICATION TO THE ACCUSED LUMICORE TRANSCEIVER MODULES

A. The SSPPU Is the Transceiver Module Itself

1. Structural Overview of the Accused Products

Each accused transceiver module — the LumiLink 400G-ZR+, the LumiLink 800G-DR8, and the LumiEdge 1600G-LR — is an integrated optical transceiver module designed for high-bandwidth data-center interconnect applications. Each module is a self-contained unit that converts electrical data signals to optical signals (and vice versa) for transmission over fiber-optic links. The modules comply with industry-standard form factors (QSFP-DD and OSFP, depending on the product line) and interface specifications.

Each accused module contains, at minimum, the following components:

(a) A photonic integrated circuit (PIC) die, fabricated on a silicon substrate, incorporating the optical waveguide structures, modulators, photodetectors, multiplexers, grating couplers, and thermal management features that perform the optical signal processing functions;

(b) Electronic integrated circuits (ICs), including digital signal processors (DSPs), transimpedance amplifiers (TIAs), and driver ICs, which perform the electronic signal processing functions;

(c) Optical fiber connectors, which provide the physical interface between the module and the external fiber-optic network;

(d) A housing/package, which provides mechanical protection, electromagnetic shielding, and compliance with the applicable form-factor standard; and

(e) Thermal management components, including the PIC-level embedded micro-channel heat sinks (covered by the '562 Patent) and module-level thermal interface materials and heat spreaders.

The PIC die is the core component of each accused module. It is the PIC die that incorporates the waveguide multiplexer structures claimed in the '847 Patent, the coherent transceiver architecture claimed in the '193 Patent, the micro-channel heat sink structures claimed in the '562 Patent, and the polarization-diverse grating coupler structures claimed in the '015 Patent. These patented features are fabricated directly within the silicon substrate of the PIC die during the semiconductor manufacturing process. They are not separately manufactured components that are later assembled onto or into the PIC die; they are monolithically integrated into the die itself.

2. The PIC Die Is Not Separately Sold

The PIC die is not sold separately by Lumicore. Based on the evidence produced in discovery to date, Lumicore fabricates its PIC dies at a foundry partner facility and ships the fabricated dies directly to Lumicore's Austin assembly facility, where they are assembled into the accused transceiver modules. At no point in this supply chain is the PIC die offered for sale as a standalone product. There is no commercial market for Lumicore PIC dies, no published pricing for Lumicore PIC dies, and no customer that purchases Lumicore PIC dies other than Lumicore itself (for internal assembly use).

Even if one were to treat the PIC die as a theoretically separable "unit," it is not a "salable" unit in any commercially meaningful sense. The PIC die is designed specifically for and exclusively used in the accused transceiver modules. It has no alternative commercial application. A bare PIC die, without the accompanying electronic ICs, optical connectors, and packaging, cannot perform any useful function. It cannot be connected to a fiber-optic network, cannot process data, and has no standalone utility.

3. The Patented Sub-Components Are Not Separable

The individual patented structures within the PIC die — the waveguide multiplexer, the coherent transceiver elements, the micro-channel heat sinks, and the grating couplers — occupy an even lower level of the product architecture. These structures are not discrete components that can be removed from the PIC die and sold or valued independently. They are patterned into the silicon substrate during the same photolithographic, etching, and deposition processes that create the rest of the PIC die's optical and electronic structures. Assigning a separate "cost" or "price" to an individual waveguide multiplexer within a PIC die would be analogous to assigning a separate cost to a single transistor within a microprocessor — it is theoretically possible as an accounting exercise, but it has no basis in commercial reality and would introduce a speculative and unreliable element into the damages calculation.

4. Supporting Evidence

The conclusion that the transceiver module is the SSPPU is supported by multiple categories of evidence:

Benchmark Semiconductor Analysis, LLC Teardown Reports. Benchmark Semiconductor Analysis, LLC conducted independent teardown analyses of each of the three accused products. The teardown reports — dated December 12, 2022 (LumiLink 400G-ZR+), June 2023 (LumiLink 800G-DR8), and February 2024 (LumiEdge 1600G-LR) — were prepared by Kevin Ng, Lead Engineer. In each report, Mr. Ng confirmed that the patented features are integrated into the PIC die, that the PIC die is assembled into the transceiver module, and that no standalone PIC die or sub-component is available as a separate product from Lumicore or from any other source. The Benchmark teardown reports provide detailed photographic and spectroscopic evidence of the physical integration of the patented structures within the PIC die.

Deposition of Dr. Sandra Michaud (Lumicore CTO). Dr. Sandra Michaud, Lumicore's Chief Technology Officer, was deposed on November 19, 2024. During her deposition, Dr. Michaud confirmed that Lumicore does not sell the PIC die as a standalone product. She testified that the PIC die is "designed and fabricated exclusively for integration into the LumiLink and LumiEdge modules" and that there is "no separate product offering" for the PIC die. Dr. Michaud further confirmed that the individual photonic structures within the PIC die — including the waveguide multiplexer, the grating couplers, and the micro-channel heat sinks — are not separately manufactured or separately priced. She described the PIC die fabrication process as a "monolithic integration process" in which all optical and thermal structures are formed simultaneously during wafer-level processing.

Lumicore Product Datasheets and Marketing Materials. All Lumicore product datasheets and marketing materials produced in discovery describe the accused products at the module level. Customers order and receive complete transceiver modules, identified by module-level part numbers (e.g., LL-400G-ZR+-C2, LL-800G-DR8-C1, LE-1600G-LR-A1). No Lumicore marketing materials reference the PIC die as a separately available product, and no PIC-die-level specifications or pricing are provided to customers.

Lumicore Financial Data. Lumicore's financial data produced in discovery (designated "Highly Confidential — Attorneys' Eyes Only") confirms that revenue is tracked at the module level only. There are no separate cost or revenue line items for the PIC die or for any of its sub-components. Lumicore's internal accounting allocates manufacturing costs at the module level, with the PIC die cost embedded within the module's overall bill of materials. While Lumicore's internal BOM data does identify the PIC die cost as a line item within the module BOM, this is an internal cost-accounting entry, not a reflection of any external market transaction. The PIC die cost is never exposed to or negotiated with customers.

B. The EMVR Is Independently Satisfied: Patented Features Drive Customer Demand

Even if Lumicore argues that the SSPPU should be something smaller than the module — for example, the PIC die cost alone, or a hypothetical imputed value for the waveguide multiplexer — OWS has strong evidence that the EMVR is independently satisfied. The patented features are the basis for customer demand for the entire accused transceiver module. As set forth below, the four patented technologies are not minor additions or incremental improvements to an otherwise independently valuable product; they are the core innovations that define each accused product's competitive position in the marketplace and that customers specifically seek when selecting a transceiver module for their data-center deployments.

1. Waveguide Multiplexer ('847 Patent)

The multi-channel silicon photonic waveguide multiplexer enables the accused modules to combine multiple optical channels onto a single waveguide with low insertion loss. This is the fundamental capability that allows the modules to achieve their specified data rates of 400 Gbps, 800 Gbps, and 1600 Gbps, respectively. Without the waveguide multiplexer, the module simply cannot perform its primary function of high-bandwidth optical data transmission. The waveguide multiplexer is not an ancillary or optional feature; it is an essential element of the optical signal path.

Customer specifications and requests for quotation ("RFQs") from hyperscale data-center operators provide direct evidence that waveguide multiplexing performance is a primary selection criterion. RFQs produced in discovery from three of Lumicore's largest customers — Pinnacle Data Systems, LLC (headquartered in Longview, Texas), Frontier Cloud Services, Inc. (headquartered in marshall, Texas), and CrossPoint Networking Corp. (headquartered in Plano, Texas) — consistently identify waveguide multiplexing performance metrics, including insertion loss per channel (specified at less than 1.5 dB), channel count (8, 16, or 32 channels, depending on the application), and spectral flatness (specified at less than 0.5 dB variation across channels), as primary mandatory requirements. Modules that do not meet these waveguide multiplexer performance requirements are disqualified from consideration, regardless of other attributes.

2. Coherent Transceiver Architecture ('193 Patent)

The high-bandwidth coherent optical transceiver architecture integrated on a silicon substrate enables detection of both the amplitude and phase of optical signals, dramatically increasing data capacity per wavelength compared to legacy intensity-modulation / direct-detection ("IM/DD") transceivers. Coherent detection is not merely a performance enhancement — it is the architectural paradigm that defines the product category. The accused modules are marketed and sold as "coherent transceiver modules," and coherent detection capability is the foundational reason that customers select these modules over alternative technologies.

The transition from IM/DD to coherent detection in data-center interconnects is one of the most significant technology shifts in the optical networking industry in the past decade. Customers migrating to coherent transceivers are doing so specifically because of the capabilities enabled by the patented architecture — capabilities that include higher spectral efficiency, longer reach, and compatibility with advanced modulation formats (e.g., 16-QAM, 64-QAM). The coherent transceiver architecture is not one feature among many; it is the defining characteristic that creates customer demand for the accused products as a class.

3. Thermal Management ('562 Patent)

The embedded micro-channel heat sink system enables reliable high-power operation of the PIC die without thermal-induced performance degradation. This is a critical requirement for deployment in data-center environments, where transceiver modules operate continuously in thermally constrained rack-mount enclosures with limited airflow. Inadequate thermal management leads to wavelength drift, increased bit-error rates, and, in extreme cases, device failure — any of which would render the module unsuitable for its intended application.

Significantly, Lumicore's own marketing materials — produced in discovery and marked as exhibits during the Michaud deposition — expressly tout the thermal performance of the LumiLink and LumiEdge modules as a key competitive differentiator. A Lumicore product brief for the LumiLink 800G-DR8, dated January 2023 (LTC-PROD-000487), states that the module "features an advanced embedded thermal management architecture that enables sustained high-performance operation across the full industrial temperature range (-5°C to +75°C) without derating." Similarly, a Lumicore white paper titled "Next-Generation Thermal Solutions for Silicon Photonic Transceivers" (LTC-MKT-001294, published Q2 2023) describes the micro-channel heat sink design as "a breakthrough in PIC-level thermal management" that "eliminates the need for external thermoelectric coolers."

Dr. Sandra Michaud confirmed during her November 19, 2024 deposition that thermal management was "critical to achieving the performance specifications" of the accused products and that the micro-channel heat sink design was "one of the key innovations" that enabled Lumicore to achieve its target power dissipation and thermal resistance specifications. (Michaud Dep. Tr. at 147:8-148:3.)

4. Polarization-Diverse Grating Coupler ('015 Patent)

The polarization-diverse grating coupler enables efficient coupling of light of both orthogonal polarization states into and out of the PIC die. This is essential for achieving the specified optical performance because the polarization state of light in a fiber-optic link is inherently random and time-varying. A grating coupler that is polarization-sensitive — i.e., that efficiently couples only one polarization state — would suffer a 3 dB (50%) power penalty for the orthogonal polarization, which would significantly degrade the module's practical data throughput and optical reach.

The polarization-diverse grating coupler is particularly important for the LumiLink 800G-DR8 and LumiEdge 1600G-LR products, which operate at higher data rates and are more sensitive to optical power budget margins. For these products, the grating coupler's polarization-diversity is not a "nice to have" feature but a mandatory requirement to meet the specified optical link budget.

5. Collective Impact

The four patented technologies, taken together, constitute the technological foundation of the accused transceiver modules. They are not independent, severable features that each contributes a discrete increment of value to an otherwise complete product. Rather, they are synergistically integrated elements of a unified photonic architecture: the waveguide multiplexer combines the optical channels, the coherent transceiver architecture processes the optical signals, the thermal management system ensures stable operation, and the grating coupler interfaces the PIC die with the external fiber-optic network. Removing any one of these elements would fundamentally impair the product's ability to meet its specifications and to compete in the market. A transceiver module without a low-loss waveguide multiplexer cannot achieve its data rate. A transceiver module without coherent detection capability cannot compete in the coherent-transceiver market segment. A transceiver module without adequate thermal management cannot operate reliably in a data-center environment. A transceiver module without a polarization-diverse grating coupler suffers unacceptable optical power penalties.

Gregory Falk, who will serve as OWS's industry expert, is expected to testify in his report (due June 15, 2025) that silicon-photonics transceiver modules are sold on the basis of their integrated optical performance, which is determined primarily by the waveguide multiplexer, coherent transceiver architecture, thermal management system, and grating coupler — the very features covered by OWS's asserted patents. Mr. Falk, formerly Vice President of Engineering at Solara Optics, Inc. and now an independent consultant, has over 25 years of experience in the silicon-photonics industry and is well-positioned to testify on industry purchasing practices and the drivers of customer demand.

The fact that OWS's own products — the OWS-400ZR, OWS-800XR, and OWS-1600DR — practice the asserted patents and compete directly with the accused products further confirms that the patented features are the basis for customer demand. The accused Lumicore products and OWS's competing products are direct substitutes in the marketplace: they target the same customers, meet the same specifications, and compete on the same performance parameters. This market-substitutability evidence strongly supports the conclusion that the patented features — which are the features that define both product families — are the features that drive customer purchasing decisions.

C. Evidence from Lumicore's Own Conduct and Marketing

The conclusion that the patented features drive customer demand for the accused modules is further reinforced by Lumicore's own conduct and marketing communications.

Product Launch Materials and Press Releases. Lumicore's product launch materials and press releases for each of the three accused products specifically highlight the optical performance characteristics enabled by the patented technologies. The LumiLink 400G-ZR+ press release (Q3 2022, LTC-PR-000012) describes the product as featuring "a novel low-loss waveguide multiplexer architecture" and "integrated coherent transceiver functionality on a single silicon die." The LumiLink 800G-DR8 press release (Q1 2023, LTC-PR-000031) emphasizes the product's "breakthrough thermal management" and "polarization-insensitive optical coupling." The LumiEdge 1600G-LR press release (Q4 2023, LTC-PR-000058) highlights the "advanced coherent detection architecture" and "high-efficiency grating coupler design." In each case, the features that Lumicore chose to highlight to the market are the very features covered by OWS's asserted patents.

Lumicore's Awareness of OWS's Patent Portfolio. discovery has produced evidence that Lumicore's engineering team reviewed OWS's published patent applications — specifically Published Application No. 2018/0123456, which corresponds to the '193 Patent — during the LumiLink 400G-ZR+ design phase in 2021 and 2022. This evidence is expected to be corroborated by the anticipated testimony of Melissa Chen, a former Lumicore staff engineer who is now employed by Veridian Photonics, Inc. Ms. Chen's anticipated testimony will establish that Lumicore's design team specifically studied OWS's published applications in the course of developing the accused products' PIC die architecture, demonstrating that the patented technologies were central to Lumicore's product design decisions. This evidence is significant not only for the willfulness analysis but also for the apportionment analysis, because it confirms that Lumicore itself recognized the importance of the patented technologies to its product development.

The April 20, 2023 Licensing Meeting. On April 20, 2023, an in-person licensing meeting was held in Austin, Texas, attended by Angela Reeves of FairPoint Licensing Group, LLC (OWS's licensing agent), Rebecca Tanaka (OWS's General Counsel), Raymond Bautista (Lumicore's General Counsel), and Kate Rourke of Hawthorne Vega & Cruz LLP (Lumicore's outside patent counsel). Per the deposition testimony of Angela Reeves (deposed December 10, 2024), Lumicore's representatives at this meeting did not dispute the technological importance of the patented features to the accused products. Instead, Lumicore's representatives focused their arguments on infringement and validity — they disputed whether the accused products actually practice the claimed inventions and whether the asserted claims are valid over the prior art. The fact that Lumicore did not dispute the technological importance of the patented features during this pre-suit licensing meeting is significant: it is an implicit concession that the patented features are central to the accused products' functionality and competitive value.

D. Distinguishing Lumicore's Anticipated Counter-Arguments

1. "The PIC Die Is the SSPPU, Not the Module"

We anticipate that Lumicore's damages expert, Dr. Volkov, will argue that the PIC die — not the transceiver module — is the SSPPU, and that the royalty base should be limited to the estimated cost of the PIC die rather than the module selling price. This argument should be rejected for several reasons.

First, the PIC die is not separately sold or separately priced by Lumicore. As discussed above, there is no commercial market for Lumicore PIC dies, no standalone PIC die product offering, and no external PIC die pricing. The PIC die has no alternative commercial use; it is designed exclusively for the accused transceiver modules. Therefore, even if the PIC die could theoretically be considered a "unit," it is not a "salable" unit in any commercially meaningful sense. The SSPPU doctrine requires identification of the smallest salable unit — i.e., a unit that is actually offered for sale in commercial transactions. A component that exists only as an intermediate work-in-progress within a manufacturer's internal supply chain does not qualify.

Second, even if the PIC die were deemed to be the SSPPU, the cost-allocation approach would not materially change the overall damages result in the manner Lumicore seeks. Per Lumicore's financial data produced in discovery (designated "Highly Confidential — Attorneys' Eyes Only"), the PIC die and associated photonic components account for approximately 65% to 70% of the transceiver module's total bill of materials ("BOM") cost. The remaining 30% to 35% consists of commodity electronic ICs (available from multiple third-party suppliers), standard fiber-optic connectors, and packaging materials — components that are widely available in the commercial market and that are not the basis for competitive differentiation among transceiver module vendors. Thus, even under a PIC-die-cost-based royalty approach, the royalty base would capture the vast majority of the module's BOM cost, and the resulting royalty rate would need to be proportionally higher to yield a reasonable royalty — a point that undercuts the purpose of disaggregation.

2. "The Individual Patented Sub-Components Are the SSPPU"

We further anticipate that Dr. Volkov may attempt to push the disaggregation even further, arguing that the SSPPU is not the PIC die but rather the individual patented sub-components — e.g., the waveguide multiplexer structure alone, or the grating coupler structure alone. This argument is even weaker than the PIC-die argument and should be rejected out of hand.

The individual patented features — the waveguide multiplexer, the coherent transceiver elements, the micro-channel heat sinks, and the grating couplers — are not sold, priced, or capable of being separately valued on any commercial basis. They are monolithically integrated into the PIC die during the semiconductor fabrication process. There is no market for individual waveguide multiplexer structures, no price list for grating couplers, and no independent valuation methodology that could assign a meaningful economic value to a single photonic structure within a monolithically fabricated PIC die. Attempting to impute such a value would be purely speculative and contrary to commercial reality.

The Federal Circuit has consistently rejected such artificial disaggregation. Where a patented feature is so deeply integrated into a product that it has no independent commercial existence, requiring the patentee to construct a hypothetical sub-component price as the royalty base would introduce unreliable, speculative elements into the damages calculation — precisely the outcome the SSPPU doctrine was designed to prevent. The SSPPU doctrine is intended to prevent over-compensation by tying damages to the patented feature's actual market footprint; it is not intended to manufacture under-compensation by requiring a fictitious disaggregation of inseparable product elements.

3. "The Royalty Rate Should Account for Non-Patented Features (Rate Apportionment)"

We anticipate that Lumicore will argue, in the alternative, that even if the module-level selling price is the appropriate royalty base, the royalty rate should be reduced to account for the contribution of non-patented features to the module's market value. This argument has some force as a general matter — the Federal Circuit in CSIRO recognized rate apportionment as a permissible approach — and we should address it forthrightly rather than dismissing it entirely.

Our position is that the proposed royalty rate of 8.5% already accounts for the contribution of non-patented features through the Georgia-Pacific factor analysis. The 8.5% rate is not an arbitrary figure; it is derived from a rigorous analysis of comparable licenses, each of which calculates royalties on a product-level (module-level) revenue base. The comparable licenses include:

• OWS-Terralink Semiconductor (2020): 6.5% running royalty on $45 million in licensed revenue, calculated at the module level;

• OWS-NovaWave Corp. (2021): 9.0% running royalty on $22 million in licensed revenue, calculated at the module level;

• OWS-Helix Integrated Photonics (2022): 7.75% running royalty plus a $2.5 million upfront payment on $68 million in licensed revenue, calculated at the module level;

• Stratos Photonics v. ArcLight Systems (2019 settlement): 8.0% royalty, calculated at the product level; and

• Coreline Technologies v. Prisma Semiconductor (2021 cross-license): 10.2% effective rate, calculated at the product level.

Each of these licenses was negotiated between sophisticated parties in the silicon-photonics industry, and each employs a module-level or product-level royalty base. The royalty rates in these licenses inherently reflect the negotiated allocation of value between patented and non-patented features at the product level. By deriving the 8.5% rate from these comparable licenses, Dr. Keating's methodology ensures that rate apportionment is already embedded in the damages calculation.

This is the approach Dr. Keating will employ in his expert report: a module-level royalty base multiplied by a royalty rate that is itself apportioned to reflect the patented contribution. This methodology satisfies the Federal Circuit's apportionment requirement under both VirnetX and CSIRO and is fully consistent with the EMVR framework.

V. QUANTITATIVE ANALYSIS

A. Royalty Base Calculation Using Module-Level Revenue

The following table summarizes the accused-product revenue data by product line, based on Lumicore's financial data produced in discovery and Dr. Keating's preliminary revenue projections:

The non-lost-profit sales base for the reasonable-royalty calculation is derived by subtracting the portion of revenue subject to the lost-profits claim under Theory 1. The market-share displacement factor of 38.2%, supported by the Market Share Displacement Analysis (internal C&H memo, June 2024), yields the following:

Total projected accused-product revenue through trial: $714.6M

Less: Revenue displaced to OWS (38.2%): $714.6M × 38.2% = $272.98M

Non-lost-profit revenue base: $714.6M − $272.98M = $441.62M

Reasonable royalty at 8.5%: $441.62M × 8.5% = $37.54M

This yields a reasonable-royalty damages figure of approximately $37.5 million, which is the amount Dr. Keating will present under Theory 2.

B. Comparison: Hypothetical Disaggregated Royalty Base (for Rebuttal Purposes)

To anticipate and rebut Lumicore's expected disaggregation arguments, we have analyzed the hypothetical impact of using a PIC-die-cost-based royalty base instead of the module-level revenue base.

PIC-Die-Cost Approach. Based on Lumicore's BOM data produced in discovery, the estimated PIC die cost ranges from approximately $55 to $75 per module, depending on the product line (the LumiEdge 1600G-LR, being the newest and most complex product, has the highest PIC die cost). Over the projected infringement period, the total PIC die cost for all accused products would represent approximately 25% to 30% of total module revenue. If Dr. Volkov uses PIC die cost as the royalty base, she would be working with a base that is roughly one-quarter to one-third of the module-level base. To yield the same total reasonable-royalty damages of $37.5 million on this reduced base, the royalty rate would need to be approximately 25% to 30% — a rate that is far above any rate observed in comparable licenses in the silicon-photonics industry and that would be difficult to defend as reasonable.

Further Disaggregation to Individual Patented Sub-Components. If Dr. Volkov attempts to further disaggregate the royalty base to the level of individual patented sub-components (e.g., the waveguide multiplexer alone, or the grating coupler alone), the resulting base would be even smaller and the required royalty rate even more extreme. More fundamentally, as discussed in Section IV.D.2 above, there is no market data, no separately manufactured component, and no independent valuation methodology that could support a cost allocation to a single photonic structure within a monolithically fabricated PIC die. Any such allocation would be speculative and should be excluded under Daubert.

Key Point. The disaggregation exercise demonstrates that artificially reducing the royalty base below the module level does not change the total amount of reasonable-royalty damages that is economically justified; it merely shifts the apportionment from the rate to the base, producing mathematically equivalent results but introducing speculative elements that make the analysis less reliable. The module-level base with an 8.5% rate is the more straightforward, more transparent, and more defensible approach.

C. Cross-Check Against Comparable Licenses

All five comparable licenses identified in the Comparable License Analysis (internal C&H memo, July 2024) calculate royalties on a product-level (module-level) revenue base, not on a sub-component basis. This is a critical cross-check because it demonstrates that the industry standard for licensing silicon-photonics patents is a module-level or product-level royalty base — which is exactly the approach we are proposing.

The comparable licenses are summarized below:

Several observations are noteworthy. First, all five licenses use the module/product-level selling price as the royalty base, not a sub-component cost. This confirms that sophisticated parties in the silicon-photonics industry — including OWS itself and its prior licensees — have consistently agreed that the appropriate royalty base for patents covering PIC-level technology is the module-level revenue. Second, the royalty rates in these licenses range from 6.5% to 10.2%, with a median of approximately 8.0%. The proposed rate of 8.5% falls comfortably within this range and is well-supported by the comparable-license evidence. Third, the licenses involving OWS's own patent portfolio (OWS-Terralink, OWS-NovaWave, OWS-Helix) are the most directly comparable because they involve the same licensor, the same technology, and the same industry segment. These licenses carry significant weight under Georgia-Pacific Factor 1 (royalties received by the patentee for licensing the patent-in-suit) and Factor 2 (rates paid by the licensee for comparable patents).

The comparable-license cross-check strongly reinforces the conclusion that the module-level royalty base and the 8.5% royalty rate are appropriate and well-supported.

VI. IMPLICATIONS FOR LOST-PROFITS ANALYSIS

Although the apportionment and EMVR analysis set forth in this memorandum is primarily directed at the reasonable-royalty calculation under Theory 2, it also supports and reinforces the lost-profits analysis under Theory 1.

The lost-profits analysis uses the full module revenue as the measure of lost sales. OWS's competing products — the OWS-400ZR and OWS-800XR transceiver module families — compete directly with Lumicore's accused products at the module level. Customers purchase complete transceiver modules, and when a customer chooses a Lumicore module instead of an OWS module, OWS loses a module-level sale. The 47.3% gross margin that serves as the basis for the lost-profits calculation is calculated on the full module selling price.

If the Court were to require apportionment of lost profits to reflect only the value of the patented features (as opposed to the entire module), the same body of evidence supporting the EMVR in the reasonable-royalty context would support using the full module price and margin as the lost-profits base. Because the patented features drive customer demand for the entire module, the full module selling price represents the economic value at risk when a customer chooses a Lumicore module over an OWS module. The customer is not choosing a waveguide multiplexer or a grating coupler in isolation; the customer is choosing a complete transceiver module whose competitive attributes are defined by the patented technologies.

The Panduit Factor Analysis (internal C&H memo, May 2024) and the Market Share Displacement Analysis (internal C&H memo, June 2024) provide detailed support for the 38.2% market-share displacement figure and the four-factor Panduit test. The apportionment analysis in this memorandum complements those analyses by establishing that the patented features account for the module-level competitive differentiation that underpins OWS's lost-profits claim.

VII. ANTICIPATED DAUBERT CHALLENGES AND PREPARATION

A. Expected Timing and Scope of Lumicore's Challenge

Lumicore is expected to file a Daubert motion challenging Dr. Keating's expert damages testimony.

MEMORANDUM — PRIVILEGED AND CONFIDENTIAL ATTORNEY WORK PRODUCT
This memorandum constitutes attorney work product prepared in anticipation of litigation and is protected by the work-product doctrine and attorney-client privilege. It should not be disclosed to any person outside the Calderwood & Harkness LLP litigation team without prior authorization from Alan Ngo.
TO: Alan Ngo, Responsible Partner; Marcus Delgado, Billing Partner
FROM: Wendy Schultz, Lead Associate; Omar Siddiqui, Working Associate
DATE: August 22, 2024
RE: Apportionment Analysis — Smallest Salable Patent-Practicing Unit (SSPPU) and Entire Market Value Rule (EMVR) as Applied to Accused Lumicore Transceiver Modules
MATTER: Optiwave Semiconductor Corp. v. Lumicore Technologies, Inc., Case No. 2:24-cv-00319-JRG (E.D. Tex.)
Right-click to update Table of Contents
I. EXECUTIVE SUMMARY
This memorandum analyzes whether the appropriate royalty base for calculating reasonable-royalty damages under Theory 2 of our damages framework (applying the Georgia-Pacific fifteen-factor analysis) should be the price of the entire accused transceiver module — invoking the Entire Market Value Rule ("EMVR") — or a disaggregated sub-component thereof, under the Smallest Salable Patent-Practicing Unit ("SSPPU") approach. The memorandum is intended to provide the analytical framework that our retained damages expert, Dr. Lawrence Keating of Meridian Economic Consulting Group, LLC, should employ in his expert damages report, which is due on June 15, 2025.
Primary Conclusion. We conclude that the EMVR is properly applied to the module-level selling price of each of the three accused Lumicore products — the LumiLink 400G-ZR+, the LumiLink 800G-DR8, and the LumiEdge 1600G-LR — because the patented features collectively constitute the basis for customer demand for the accused modules. Specifically, the waveguide multiplexer technology claimed in U.S. Patent No. 10,234,847 (the "'847 Patent"), the coherent transceiver integration architecture claimed in U.S. Patent No. 10,891,193 (the "'193 Patent"), the micro-channel thermal management system claimed in U.S. Patent No. 9,678,562 (the "'562 Patent"), and the polarization-diverse grating coupler claimed in U.S. Patent No. 11,402,015 (the "'015 Patent") are the core functional elements that differentiate the accused products in the marketplace and that customers identify as the primary basis for their purchasing decisions.
Alternative SSPPU Analysis. Even if the Court requires the application of a SSPPU analysis rather than the EMVR, the smallest salable patent-practicing unit is the transceiver module itself — not a sub-component. The patented innovations are not sold or priced separately; they are physically and commercially integrated into the photonic integrated circuit ("PIC") die, which is in turn only commercially available as part of the fully assembled transceiver module. Lumicore does not sell PIC dies on a standalone basis, and the individual patented structures within the PIC die (e.g., the waveguide multiplexer, the grating coupler) are not separately manufactured, sold, or priced in any commercial market.
Quantitative Impact. The proposed royalty rate of 8.5% should be applied to total accused-product net sales revenues. For Fiscal Year 2023, total accused-product revenue was $264.9 million, comprising $94.8 million for the LumiLink 400G-ZR+, $128.4 million for the LumiLink 800G-DR8, and $41.7 million for the LumiEdge 1600G-LR. Projected total accused-product revenue through trial (covering the full infringement period of Q3 2022 through Q4 2025) is approximately $714.6 million. After subtracting the portion of revenue subject to the lost-profits claim under Theory 1 (38.2% market-share displacement), the reasonable-royalty base is approximately $441.6 million, yielding a reasonable royalty of approximately $37.5 million.
Anticipated Defense Strategy. We anticipate that Lumicore and its damages expert, Dr. Natasha Volkov of Pinnacle Analytics Partners LLP, will argue for a disaggregated royalty base — for example, limiting the base to the estimated PIC die cost alone, or to a hypothetical imputed value for the waveguide multiplexer sub-component — in an effort to artificially reduce the damages award. This memorandum provides the analytical foundation to defeat that argument at the Daubert stage, at summary judgment, and at trial.
II. BACKGROUND
A. The Parties and Accused Products
1. Optiwave Semiconductor Corp.
Optiwave Semiconductor Corp. ("OWS" or "Optiwave") is a Delaware corporation headquartered at 4810 Bayside Technology Parkway, Suite 300, Santa Clara, California 95054. OWS is a fabless semiconductor company specializing in silicon photonics chipsets for data-center interconnects. According to audited financial statements prepared by Gresham & Whitfield LLP, OWS reported Fiscal Year 2023 revenue of $487.3 million and net income of $61.4 million. OWS employs approximately 1,250 persons across its Santa Clara headquarters, its Austin design center, and its Singapore operations office.
OWS is the third-largest U.S. supplier of silicon-photonics transceiver modules for hyperscale data centers. The company designs its PIC dies in-house and contracts fabrication to third-party foundries, with final module assembly performed at its contract manufacturing partners' facilities. OWS's intellectual property portfolio consists of 142 issued United States patents and 58 pending applications, spanning silicon photonics, waveguide design, thermal management, optical coupling, and coherent transceiver architecture.
Ridgeline Capital Partners, LP acquired a 55% majority stake in OWS in 2019 for $780 million, and Ridgeline remains the majority owner. The minority ownership is held by a combination of early-stage venture investors and current and former OWS management.
OWS's key competing products — that is, the products that compete directly with the accused Lumicore modules in the marketplace — are the OWS-400ZR, OWS-800XR, and OWS-1600DR transceiver module families. OWS's gross margin on these competing products is 47.3%, a figure that will be significant for the lost-profits analysis under Theory 1.
2. Lumicore Technologies, Inc.
Lumicore Technologies, Inc. ("Lumicore" or "LTC") is a Delaware corporation headquartered at 9200 Spectrum Boulevard, Building C, Austin, Texas 78759. Lumicore completed its initial public offering on the Nasdaq exchange in March 2022 at an offering price of $24.00 per share. As of the filing date of this action, Lumicore's common stock was trading at approximately $37.00 per share, implying a market capitalization of approximately $3.4 billion based on 91.8 million shares outstanding. Per Lumicore's Form 10-K filed with the Securities and Exchange Commission, the company reported Fiscal Year 2023 revenue of $312.6 million. Lumicore employs approximately 870 persons.
Lumicore designs and sells optical transceiver modules for data-center and telecommunications applications. Like OWS, Lumicore is a fabless semiconductor company; it designs its PIC dies in-house and uses foundry partners for fabrication. Lumicore's assembly operations are conducted at a facility in Austin, Texas.
3. Accused Product Lines
The Complaint asserts infringement against three Lumicore product families, each of which is an integrated optical transceiver module sold to hyperscale data-center operators and telecommunications equipment manufacturers. The accused products, their launch dates, Fiscal Year 2023 revenues, and asserted patents are summarized below:
Total accused-product revenue over the infringement period from product launch through Q1 2024 (the most recent quarter for which actual revenue data has been produced in discovery) is approximately $378.2 million, per Dr. Keating's preliminary analysis. Based on Dr. Keating's revenue projections, which incorporate Lumicore's publicly reported growth rates and analyst estimates, projected total accused-product revenue through trial (Q3 2022 through Q4 2025) is approximately $714.6 million.
It bears emphasis that the accused products are among Lumicore's highest-revenue product lines, collectively accounting for approximately 84.7% of Lumicore's total FY 2023 revenue.
B. The Asserted Patents and Patented Technology
1. U.S. Patent No. 10,234,847 (the "'847 Patent")
The '847 Patent is titled "Multi-Channel Silicon Photonic Waveguide Multiplexer with Low Insertion Loss." It was filed on September 14, 2016, and issued on March 19, 2019. The original and current assignee is Optiwave Semiconductor Corp. The Complaint asserts infringement of Claims 1, 3, 5, 9, and 14 of the '847 Patent.
The '847 Patent claims a multi-channel waveguide multiplexer structure fabricated on a silicon photonic substrate that combines multiple optical channels onto a single waveguide with significantly reduced insertion loss compared to prior-art multiplexer designs. The claimed structure employs a novel adiabatic coupler geometry and optimized waveguide taper profiles that minimize optical signal loss at each channel-combining junction. This technology is fundamental to achieving the high data rates (400G, 800G, and 1600G) specified for the accused products.
2. U.S. Patent No. 10,891,193 (the "'193 Patent")
The '193 Patent is titled "High-Bandwidth Coherent Optical Transceiver Integrated on Silicon Substrate." It was filed on April 7, 2017, and issued on January 12, 2021. The '193 Patent shares Provisional Application No. 62/499,117 with the '015 Patent, establishing a common priority date for the two inventions. The Complaint asserts infringement of Claims 1, 3, 7, 10, 15, and 18 of the '193 Patent.
The '193 Patent claims a coherent optical transceiver architecture that is monolithically integrated on a silicon substrate, enabling detection of both the amplitude and phase of optical signals. This coherent detection capability dramatically increases data capacity compared to legacy intensity-modulation / direct-detection ("IM/DD") transceivers. The patented architecture includes on-chip local oscillator generation, balanced photodetector arrays, and digital signal processing interfaces, all integrated within a single silicon photonic die.
3. U.S. Patent No. 9,678,562 (the "'562 Patent")
The '562 Patent is titled "Method for Thermal Management of Photonic Integrated Circuits Using Embedded Micro-Channel Heat Sinks." It was filed on November 22, 2014, and issued on June 13, 2017. The '562 Patent was originally assigned to Solara Optics, Inc. and was acquired by OWS pursuant to an asset purchase agreement dated June 15, 2018. The total purchase price for the asset acquisition was $34.5 million, covering a portfolio of 12 patents, including the '562 Patent. The '562 Patent has a continuation relationship with the '847 Patent, as both derive from the same patent family. The Complaint asserts infringement of Claims 1, 4, 6, and 11 of the '562 Patent.
The '562 Patent claims a method for managing the thermal output of photonic integrated circuits by embedding micro-channel heat sinks directly within or adjacent to the silicon substrate. The claimed micro-channel geometry and fluid-flow characteristics enable highly efficient heat dissipation from the photonic structures without requiring bulky external heat sinks, which would increase the module's form factor beyond industry-standard specifications. This thermal management technology is essential for high-power PIC operation in the thermally constrained environments typical of data-center rack-mount deployments.
4. U.S. Patent No. 11,402,015 (the "'015 Patent")
The '015 Patent is titled "Polarization-Diverse Grating Coupler for Silicon Photonic Transceivers." It was filed on August 30, 2019, and issued on August 2, 2022. As noted above, the '015 Patent shares Provisional Application No. 62/499,117 with the '193 Patent. The Complaint asserts infringement of Claims 1, 5, 8, 12, and 16 of the '015 Patent.
The '015 Patent claims a grating coupler structure that efficiently couples light of both orthogonal polarization states into and out of the photonic integrated circuit. Prior-art grating couplers were polarization-sensitive, meaning they could efficiently couple only one polarization state, resulting in significant optical power loss for the orthogonal polarization. The claimed polarization-diverse design uses a two-dimensional grating pattern and sub-wavelength structuring to achieve near-equal coupling efficiency for both polarization states, which is essential for achieving the optical performance specifications of modern coherent transceivers.
5. Technical Summary
The four asserted patents collectively cover the core functional elements of a silicon-photonics transceiver module: (1) the waveguide multiplexer that combines multiple optical channels with low insertion loss ('847 Patent); (2) the coherent transceiver architecture integrated on a silicon substrate that enables high-bandwidth data transmission ('193 Patent); (3) the embedded thermal management system using micro-channel heat sinks that ensures reliable high-power operation ('562 Patent); and (4) the polarization-diverse grating coupler that efficiently couples light into and out of the photonic integrated circuit ('015 Patent). Together, these innovations enable high-bandwidth, low-power, thermally stable optical transceiver modules suitable for deployment in hyperscale data-center interconnect applications. The four patented technologies are not peripheral or incremental improvements; they define the core architecture and performance characteristics of the accused transceiver modules.
C. Damages Theories Overview
OWS's damages case comprises three theories, as follows:
Theory 1: Lost Profits Under *Panduit*. OWS seeks lost profits on the portion of Lumicore's accused-product sales that displaced OWS's own sales of competing products. The lost-profits claim is projected at $128.9 million through trial, calculated as follows: $714.6 million (projected total accused-product revenue) × 38.2% (market-share displacement factor) × 47.3% (OWS gross margin on competing products) = approximately $128.9 million. The 38.2% market-share displacement figure is derived from the Market Share Displacement Analysis prepared by this team in June 2024, which analyzed the relevant market, the two-supplier market structure, and the absence of non-infringing substitutes.
Theory 2: Reasonable Royalty Under *Georgia-Pacific*. For the portion of Lumicore's accused-product sales not subject to the lost-profits claim — i.e., sales that would not have been captured by OWS even in the absence of infringement — OWS seeks a reasonable royalty. The reasonable-royalty calculation is as follows: $714.6 million (projected total accused-product revenue) minus $272.98 million (the 38.2% of revenue subject to the lost-profits claim) yields a non-lost-profit revenue base of approximately $441.6 million. Applying the proposed royalty rate of 8.5% produces a reasonable royalty of approximately $37.5 million. This memorandum focuses primarily on Theory 2 and the apportionment analysis that underlies the royalty-base determination. However, as discussed in Section VI below, the apportionment analysis is also relevant to Theory 1.
Theory 3: Enhanced Damages for Willful Infringement. OWS will seek enhanced damages under 35 U.S.C. § 284, based on Lumicore's continued infringement after receiving actual notice of the asserted patents. Post-notice lost profits are estimated at approximately $109.2 million; various multiplier scenarios (1.5x, 2x, 3x) will be presented at trial. The willfulness analysis is addressed in a separate memorandum dated July 2024.
Total Damages Demand. The total damages demand, as set forth in the Complaint, is approximately $210.0 million (blended and rounded).
The apportionment analysis set forth in this memorandum will be communicated to Dr. Keating for incorporation into his expert damages report. Dr. Keating's report will present the royalty-base and royalty-rate analysis; this memorandum provides the legal framework and factual support for his methodology.
III. LEGAL FRAMEWORK
A. The Default Rule: The Smallest Salable Patent-Practicing Unit (SSPPU)
The Federal Circuit has established a strong preference for tying patent-damages calculations to the smallest salable unit that embodies the patented invention, rather than to an entire multi-component product. This preference, articulated most forcefully in LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51 (Fed. Cir. 2012), is designed to ensure that reasonable-royalty damages reflect only the value of the patented feature, not the value of unrelated features that contribute to the overall product's market price.
The rationale for the SSPPU doctrine is straightforward: where a patent covers one feature of a complex, multi-component product, using the entire product's selling price as the royalty base risks attributing to the patentee economic value that is properly attributable to non-patented features. As the Federal Circuit explained in Uniloc USA, Inc. v. Kestrel Partners LLC, 632 F.3d 1292, 1320 (Fed. Cir. 2011), damages methodologies that employ the entire value of a multi-component product as a starting point risk misleading the jury into awarding damages based on unpatented features. In Uniloc, the court rejected the so-called "25 percent rule of thumb" as an analytically unsound method of approximating a reasonable royalty, emphasizing the need for a rigorous, evidence-based apportionment of the royalty to the patented contribution.
In VirnetX, Inc. v. Sterling Reach Holdings, 767 F.3d 1308 (Fed. Cir. 2014), the Federal Circuit reinforced the SSPPU requirement, holding that a damages expert must "carefully tie proof of damages to the claimed invention's footprint in the market place" and must "apportion the defendant's profits and the patentee's damages between the patented feature and the unpatented features." Id. at 1326. The VirnetX court vacated a jury verdict where the patentee's damages expert had used the total value of an iPhone or other accused device as the royalty base without adequate evidence that the patented features — VPN and FaceTime functionalities — were the basis for consumer demand for the entire device.
The SSPPU analysis requires identification of the smallest commercial unit that practices the asserted patent claims. In many cases, this will be a sub-component of the accused end product. For example, in LaserDynamics, the Federal Circuit held that the smallest salable patent-practicing unit for a patent covering an optical disc discrimination method was the optical disc drive ("ODD"), not the laptop computer into which the ODD was installed. 694 F.3d at 67-68. Because the ODD was a discrete, separately sold component, using the laptop price as the royalty base was impermissible absent satisfaction of the EMVR requirements.
It is critical to understand, however, that the SSPPU doctrine is not a rigid formula but rather a principle of analytical discipline. The Federal Circuit has recognized that the identification of the SSPPU depends on the facts of each case, including the structure of the accused product, the commercial practices of the parties, and the relationship between the patented technology and the accused product's functionality. As we discuss below, in cases involving highly integrated products — such as semiconductor devices and photonic integrated circuits — the SSPPU analysis may lead to the conclusion that the entire chip or module is the smallest salable patent-practicing unit.
B. The Exception: The Entire Market Value Rule (EMVR)
The EMVR provides an exception to the general SSPPU preference. Under the EMVR, a patentee may use the entire market value of a multi-component product as the royalty base where the patented feature constitutes the "basis for customer demand" for the entire product, or where the patented feature is the "predominant basis for consumer demand" or "substantially creates the value of the component parts."
The EMVR has a long pedigree in patent-damages jurisprudence. In Rite-Hite Corp. v. Kelley, Inc., 56 F.3d 1538 (Fed. Cir. 1995) (en banc), the Federal Circuit sitting en banc permitted entire-market-value recovery for unpatented components sold with a patented device, where the patented invention served as the basis for demand for the overall product. The en banc court recognized that in many commercial settings, customers purchase integrated products precisely because of a particular patented feature, and limiting the damages base to a sub-component would under-compensate the patentee for the full economic impact of the infringement.
In Lucent Technologies, Inc. v. Gateway, Inc., 580 F.3d 1301 (Fed. Cir. 2009), the court clarified that the EMVR requires affirmative proof — not mere assertion — that the patented feature is the driver of customer demand for the entire product. In Lucent, the court vacated a $1.538 billion verdict where the patentee had used the total sales revenue of Hollowell Systems as the royalty base for a patent covering a date-picker tool, without adequate evidence that the date-picker tool drove demand for the outlook software. The court held that "the base used in a running royalty calculation can always be the value of the entire commercial embodiment, as long as the magnitude of the rate is within an acceptable range (as determined by the evidence)," but emphasized that the selection of royalty base and rate must be grounded in evidence, not speculation. Id. at 1338-39.
LaserDynamics further articulated the EMVR standard, holding that the EMVR "requires proof that the patented feature is the basis for customer demand for the entire product," not merely an important feature among several. 694 F.3d at 67. The court explained that applying the EMVR to a multi-component product when the patented feature does not drive demand for the product as a whole risks inflating the damages award by capturing value attributable to unpatented features.
An important refinement was introduced in Commonwealth Scientific & Industrial Research Organisation v. Sterling Reach Holdings, 809 F.3d 1295 (Fed. Cir. 2015) ("CSIRO"). In CSIRO, the Federal Circuit recognized two permissible approaches when the EMVR is invoked:
(1) The patentee may use the entire product price as the royalty base and apply a royalty rate that specifically accounts for the contribution of the patented feature (i.e., apportioning through the rate rather than the base); or
(2) The patentee may directly demonstrate that the patented feature is the basis for customer demand for the entire product, thereby justifying use of the full product price as the base without further reduction.
The CSIRO court emphasized that the overriding requirement is that the damages calculation must "be carefully tied to the claimed invention's footprint in the market place" and must not result in a damages award that exceeds the value attributable to the patented technology. 809 F.3d at 1301. Critically, the CSIRO court held that even when the EMVR is invoked, the royalty rate itself can be adjusted downward to account for the value contribution of non-patented features — the so-called "rate-apportionment" approach. This means that the EMVR does not necessarily lead to inflated damages if the rate is properly calibrated to reflect the patented technology's proportional contribution to the product's value.
This rate-apportionment approach is the methodology that Dr. Keating will employ in this case: using the module-level selling price as the royalty base and applying a royalty rate of 8.5% that has been derived from comparable license analysis and that inherently reflects the proportion of the module's value attributable to the patented features.
C. The Interplay Between SSPPU and EMVR in the Context of Integrated Products
The Federal Circuit has recognized that in cases involving highly integrated products — particularly semiconductor chips, integrated circuits, and other devices where the patented technology is physically inseparable from the product — the SSPPU may in fact be the entire chip or module. This is because the patented technology is not separately sold or separately priceable; it exists only as an integral part of the assembled product.
In Finjan, Inc. v. Blue Coat Systems, Inc., 879 F.3d 1299 (Fed. Cir. 2018), the Federal Circuit permitted the use of revenue from the entire accused product as the royalty base where the patented security feature was "inseparable from the product's function." The Finjan court recognized that when a patented feature is so deeply integrated into a product that it cannot be separated from the product's core functionality, requiring disaggregation of the royalty base to some hypothetical sub-component would introduce speculative and unreliable elements into the damages calculation.
This line of cases is critical to our analysis. The present case involves patented features — a waveguide multiplexer, a coherent transceiver architecture, a micro-channel heat sink system, and a polarization-diverse grating coupler — that are monolithically fabricated into a photonic integrated circuit die, which is in turn assembled into a transceiver module. The patented structures are not discrete components that can be removed from the PIC die and sold or valued independently. They are integral parts of the silicon substrate itself, fabricated during the same lithographic and deposition processes that create the entire PIC. In this respect, the present case is fundamentally distinguishable from cases like LaserDynamics, where the patented feature was embodied in a discrete, separately sold component (the optical disc drive) that was installed into a larger product (the laptop computer). Here, there is no analogous separable component. The patented features and the PIC die are one and the same, and the PIC die and the transceiver module are commercially inseparable.
Accordingly, the SSPPU analysis and the EMVR analysis converge in this case: (1) the SSPPU is the transceiver module because neither the PIC die nor its constituent patented structures are separately sold or priced; and (2) the EMVR is independently satisfied because the patented features are the basis for customer demand for the transceiver module as a whole. Either analytical path leads to the same result: the module-level selling price is the appropriate royalty base.
IV. APPLICATION TO THE ACCUSED LUMICORE TRANSCEIVER MODULES
A. The SSPPU Is the Transceiver Module Itself
1. Structural Overview of the Accused Products
Each accused transceiver module — the LumiLink 400G-ZR+, the LumiLink 800G-DR8, and the LumiEdge 1600G-LR — is an integrated optical transceiver module designed for high-bandwidth data-center interconnect applications. Each module is a self-contained unit that converts electrical data signals to optical signals (and vice versa) for transmission over fiber-optic links. The modules comply with industry-standard form factors (QSFP-DD and OSFP, depending on the product line) and interface specifications.
Each accused module contains, at minimum, the following components:
(a) A photonic integrated circuit (PIC) die, fabricated on a silicon substrate, incorporating the optical waveguide structures, modulators, photodetectors, multiplexers, grating couplers, and thermal management features that perform the optical signal processing functions;
(b) Electronic integrated circuits (ICs), including digital signal processors (DSPs), transimpedance amplifiers (TIAs), and driver ICs, which perform the electronic signal processing functions;
(c) Optical fiber connectors, which provide the physical interface between the module and the external fiber-optic network;
(d) A housing/package, which provides mechanical protection, electromagnetic shielding, and compliance with the applicable form-factor standard; and
(e) Thermal management components, including the PIC-level embedded micro-channel heat sinks (covered by the '562 Patent) and module-level thermal interface materials and heat spreaders.
The PIC die is the core component of each accused module. It is the PIC die that incorporates the waveguide multiplexer structures claimed in the '847 Patent, the coherent transceiver architecture claimed in the '193 Patent, the micro-channel heat sink structures claimed in the '562 Patent, and the polarization-diverse grating coupler structures claimed in the '015 Patent. These patented features are fabricated directly within the silicon substrate of the PIC die during the semiconductor manufacturing process. They are not separately manufactured components that are later assembled onto or into the PIC die; they are monolithically integrated into the die itself.
2. The PIC Die Is Not Separately Sold
The PIC die is not sold separately by Lumicore. Based on the evidence produced in discovery to date, Lumicore fabricates its PIC dies at a foundry partner facility and ships the fabricated dies directly to Lumicore's Austin assembly facility, where they are assembled into the accused transceiver modules. At no point in this supply chain is the PIC die offered for sale as a standalone product. There is no commercial market for Lumicore PIC dies, no published pricing for Lumicore PIC dies, and no customer that purchases Lumicore PIC dies other than Lumicore itself (for internal assembly use).
Even if one were to treat the PIC die as a theoretically separable "unit," it is not a "salable" unit in any commercially meaningful sense. The PIC die is designed specifically for and exclusively used in the accused transceiver modules. It has no alternative commercial application. A bare PIC die, without the accompanying electronic ICs, optical connectors, and packaging, cannot perform any useful function. It cannot be connected to a fiber-optic network, cannot process data, and has no standalone utility.
3. The Patented Sub-Components Are Not Separable
The individual patented structures within the PIC die — the waveguide multiplexer, the coherent transceiver elements, the micro-channel heat sinks, and the grating couplers — occupy an even lower level of the product architecture. These structures are not discrete components that can be removed from the PIC die and sold or valued independently. They are patterned into the silicon substrate during the same photolithographic, etching, and deposition processes that create the rest of the PIC die's optical and electronic structures. Assigning a separate "cost" or "price" to an individual waveguide multiplexer within a PIC die would be analogous to assigning a separate cost to a single transistor within a microprocessor — it is theoretically possible as an accounting exercise, but it has no basis in commercial reality and would introduce a speculative and unreliable element into the damages calculation.
4. Supporting Evidence
The conclusion that the transceiver module is the SSPPU is supported by multiple categories of evidence:
Benchmark Semiconductor Analysis, LLC Teardown Reports. Benchmark Semiconductor Analysis, LLC conducted independent teardown analyses of each of the three accused products. The teardown reports — dated December 12, 2022 (LumiLink 400G-ZR+), June 2023 (LumiLink 800G-DR8), and February 2024 (LumiEdge 1600G-LR) — were prepared by Kevin Ng, Lead Engineer. In each report, Mr. Ng confirmed that the patented features are integrated into the PIC die, that the PIC die is assembled into the transceiver module, and that no standalone PIC die or sub-component is available as a separate product from Lumicore or from any other source. The Benchmark teardown reports provide detailed photographic and spectroscopic evidence of the physical integration of the patented structures within the PIC die.
Deposition of Dr. Sandra Michaud (Lumicore CTO). Dr. Sandra Michaud, Lumicore's Chief Technology Officer, was deposed on November 19, 2024. During her deposition, Dr. Michaud confirmed that Lumicore does not sell the PIC die as a standalone product. She testified that the PIC die is "designed and fabricated exclusively for integration into the LumiLink and LumiEdge modules" and that there is "no separate product offering" for the PIC die. Dr. Michaud further confirmed that the individual photonic structures within the PIC die — including the waveguide multiplexer, the grating couplers, and the micro-channel heat sinks — are not separately manufactured or separately priced. She described the PIC die fabrication process as a "monolithic integration process" in which all optical and thermal structures are formed simultaneously during wafer-level processing.
Lumicore Product Datasheets and Marketing Materials. All Lumicore product datasheets and marketing materials produced in discovery describe the accused products at the module level. Customers order and receive complete transceiver modules, identified by module-level part numbers (e.g., LL-400G-ZR+-C2, LL-800G-DR8-C1, LE-1600G-LR-A1). No Lumicore marketing materials reference the PIC die as a separately available product, and no PIC-die-level specifications or pricing are provided to customers.
Lumicore Financial Data. Lumicore's financial data produced in discovery (designated "Highly Confidential — Attorneys' Eyes Only") confirms that revenue is tracked at the module level only. There are no separate cost or revenue line items for the PIC die or for any of its sub-components. Lumicore's internal accounting allocates manufacturing costs at the module level, with the PIC die cost embedded within the module's overall bill of materials. While Lumicore's internal BOM data does identify the PIC die cost as a line item within the module BOM, this is an internal cost-accounting entry, not a reflection of any external market transaction. The PIC die cost is never exposed to or negotiated with customers.
B. The EMVR Is Independently Satisfied: Patented Features Drive Customer Demand
Even if Lumicore argues that the SSPPU should be something smaller than the module — for example, the PIC die cost alone, or a hypothetical imputed value for the waveguide multiplexer — OWS has strong evidence that the EMVR is independently satisfied. The patented features are the basis for customer demand for the entire accused transceiver module. As set forth below, the four patented technologies are not minor additions or incremental improvements to an otherwise independently valuable product; they are the core innovations that define each accused product's competitive position in the marketplace and that customers specifically seek when selecting a transceiver module for their data-center deployments.
1. Waveguide Multiplexer ('847 Patent)
The multi-channel silicon photonic waveguide multiplexer enables the accused modules to combine multiple optical channels onto a single waveguide with low insertion loss. This is the fundamental capability that allows the modules to achieve their specified data rates of 400 Gbps, 800 Gbps, and 1600 Gbps, respectively. Without the waveguide multiplexer, the module simply cannot perform its primary function of high-bandwidth optical data transmission. The waveguide multiplexer is not an ancillary or optional feature; it is an essential element of the optical signal path.
Customer specifications and requests for quotation ("RFQs") from hyperscale data-center operators provide direct evidence that waveguide multiplexing performance is a primary selection criterion. RFQs produced in discovery from three of Lumicore's largest customers — Pinnacle Data Systems, LLC (headquartered in Longview, Texas), Frontier Cloud Services, Inc. (headquartered in marshall, Texas), and CrossPoint Networking Corp. (headquartered in Plano, Texas) — consistently identify waveguide multiplexing performance metrics, including insertion loss per channel (specified at less than 1.5 dB), channel count (8, 16, or 32 channels, depending on the application), and spectral flatness (specified at less than 0.5 dB variation across channels), as primary mandatory requirements. Modules that do not meet these waveguide multiplexer performance requirements are disqualified from consideration, regardless of other attributes.
2. Coherent Transceiver Architecture ('193 Patent)
The high-bandwidth coherent optical transceiver architecture integrated on a silicon substrate enables detection of both the amplitude and phase of optical signals, dramatically increasing data capacity per wavelength compared to legacy intensity-modulation / direct-detection ("IM/DD") transceivers. Coherent detection is not merely a performance enhancement — it is the architectural paradigm that defines the product category. The accused modules are marketed and sold as "coherent transceiver modules," and coherent detection capability is the foundational reason that customers select these modules over alternative technologies.
The transition from IM/DD to coherent detection in data-center interconnects is one of the most significant technology shifts in the optical networking industry in the past decade. Customers migrating to coherent transceivers are doing so specifically because of the capabilities enabled by the patented architecture — capabilities that include higher spectral efficiency, longer reach, and compatibility with advanced modulation formats (e.g., 16-QAM, 64-QAM). The coherent transceiver architecture is not one feature among many; it is the defining characteristic that creates customer demand for the accused products as a class.
3. Thermal Management ('562 Patent)
The embedded micro-channel heat sink system enables reliable high-power operation of the PIC die without thermal-induced performance degradation. This is a critical requirement for deployment in data-center environments, where transceiver modules operate continuously in thermally constrained rack-mount enclosures with limited airflow. Inadequate thermal management leads to wavelength drift, increased bit-error rates, and, in extreme cases, device failure — any of which would render the module unsuitable for its intended application.
Significantly, Lumicore's own marketing materials — produced in discovery and marked as exhibits during the Michaud deposition — expressly tout the thermal performance of the LumiLink and LumiEdge modules as a key competitive differentiator. A Lumicore product brief for the LumiLink 800G-DR8, dated January 2023 (LTC-PROD-000487), states that the module "features an advanced embedded thermal management architecture that enables sustained high-performance operation across the full industrial temperature range (-5°C to +75°C) without derating." Similarly, a Lumicore white paper titled "Next-Generation Thermal Solutions for Silicon Photonic Transceivers" (LTC-MKT-001294, published Q2 2023) describes the micro-channel heat sink design as "a breakthrough in PIC-level thermal management" that "eliminates the need for external thermoelectric coolers."
Dr. Sandra Michaud confirmed during her November 19, 2024 deposition that thermal management was "critical to achieving the performance specifications" of the accused products and that the micro-channel heat sink design was "one of the key innovations" that enabled Lumicore to achieve its target power dissipation and thermal resistance specifications. (Michaud Dep. Tr. at 147:8-148:3.)
4. Polarization-Diverse Grating Coupler ('015 Patent)
The polarization-diverse grating coupler enables efficient coupling of light of both orthogonal polarization states into and out of the PIC die. This is essential for achieving the specified optical performance because the polarization state of light in a fiber-optic link is inherently random and time-varying. A grating coupler that is polarization-sensitive — i.e., that efficiently couples only one polarization state — would suffer a 3 dB (50%) power penalty for the orthogonal polarization, which would significantly degrade the module's practical data throughput and optical reach.
The polarization-diverse grating coupler is particularly important for the LumiLink 800G-DR8 and LumiEdge 1600G-LR products, which operate at higher data rates and are more sensitive to optical power budget margins. For these products, the grating coupler's polarization-diversity is not a "nice to have" feature but a mandatory requirement to meet the specified optical link budget.
5. Collective Impact
The four patented technologies, taken together, constitute the technological foundation of the accused transceiver modules. They are not independent, severable features that each contributes a discrete increment of value to an otherwise complete product. Rather, they are synergistically integrated elements of a unified photonic architecture: the waveguide multiplexer combines the optical channels, the coherent transceiver architecture processes the optical signals, the thermal management system ensures stable operation, and the grating coupler interfaces the PIC die with the external fiber-optic network. Removing any one of these elements would fundamentally impair the product's ability to meet its specifications and to compete in the market. A transceiver module without a low-loss waveguide multiplexer cannot achieve its data rate. A transceiver module without coherent detection capability cannot compete in the coherent-transceiver market segment. A transceiver module without adequate thermal management cannot operate reliably in a data-center environment. A transceiver module without a polarization-diverse grating coupler suffers unacceptable optical power penalties.
Gregory Falk, who will serve as OWS's industry expert, is expected to testify in his report (due June 15, 2025) that silicon-photonics transceiver modules are sold on the basis of their integrated optical performance, which is determined primarily by the waveguide multiplexer, coherent transceiver architecture, thermal management system, and grating coupler — the very features covered by OWS's asserted patents. Mr. Falk, formerly Vice President of Engineering at Solara Optics, Inc. and now an independent consultant, has over 25 years of experience in the silicon-photonics industry and is well-positioned to testify on industry purchasing practices and the drivers of customer demand.
The fact that OWS's own products — the OWS-400ZR, OWS-800XR, and OWS-1600DR — practice the asserted patents and compete directly with the accused products further confirms that the patented features are the basis for customer demand. The accused Lumicore products and OWS's competing products are direct substitutes in the marketplace: they target the same customers, meet the same specifications, and compete on the same performance parameters. This market-substitutability evidence strongly supports the conclusion that the patented features — which are the features that define both product families — are the features that drive customer purchasing decisions.
C. Evidence from Lumicore's Own Conduct and Marketing
The conclusion that the patented features drive customer demand for the accused modules is further reinforced by Lumicore's own conduct and marketing communications.
Product Launch Materials and Press Releases. Lumicore's product launch materials and press releases for each of the three accused products specifically highlight the optical performance characteristics enabled by the patented technologies. The LumiLink 400G-ZR+ press release (Q3 2022, LTC-PR-000012) describes the product as featuring "a novel low-loss waveguide multiplexer architecture" and "integrated coherent transceiver functionality on a single silicon die." The LumiLink 800G-DR8 press release (Q1 2023, LTC-PR-000031) emphasizes the product's "breakthrough thermal management" and "polarization-insensitive optical coupling." The LumiEdge 1600G-LR press release (Q4 2023, LTC-PR-000058) highlights the "advanced coherent detection architecture" and "high-efficiency grating coupler design." In each case, the features that Lumicore chose to highlight to the market are the very features covered by OWS's asserted patents.
Lumicore's Awareness of OWS's Patent Portfolio. discovery has produced evidence that Lumicore's engineering team reviewed OWS's published patent applications — specifically Published Application No. 2018/0123456, which corresponds to the '193 Patent — during the LumiLink 400G-ZR+ design phase in 2021 and 2022. This evidence is expected to be corroborated by the anticipated testimony of Melissa Chen, a former Lumicore staff engineer who is now employed by Veridian Photonics, Inc. Ms. Chen's anticipated testimony will establish that Lumicore's design team specifically studied OWS's published applications in the course of developing the accused products' PIC die architecture, demonstrating that the patented technologies were central to Lumicore's product design decisions. This evidence is significant not only for the willfulness analysis but also for the apportionment analysis, because it confirms that Lumicore itself recognized the importance of the patented technologies to its product development.
The April 20, 2023 Licensing Meeting. On April 20, 2023, an in-person licensing meeting was held in Austin, Texas, attended by Angela Reeves of FairPoint Licensing Group, LLC (OWS's licensing agent), Rebecca Tanaka (OWS's General Counsel), Raymond Bautista (Lumicore's General Counsel), and Kate Rourke of Hawthorne Vega & Cruz LLP (Lumicore's outside patent counsel). Per the deposition testimony of Angela Reeves (deposed December 10, 2024), Lumicore's representatives at this meeting did not dispute the technological importance of the patented features to the accused products. Instead, Lumicore's representatives focused their arguments on infringement and validity — they disputed whether the accused products actually practice the claimed inventions and whether the asserted claims are valid over the prior art. The fact that Lumicore did not dispute the technological importance of the patented features during this pre-suit licensing meeting is significant: it is an implicit concession that the patented features are central to the accused products' functionality and competitive value.
D. Distinguishing Lumicore's Anticipated Counter-Arguments
1. "The PIC Die Is the SSPPU, Not the Module"
We anticipate that Lumicore's damages expert, Dr. Volkov, will argue that the PIC die — not the transceiver module — is the SSPPU, and that the royalty base should be limited to the estimated cost of the PIC die rather than the module selling price. This argument should be rejected for several reasons.
First, the PIC die is not separately sold or separately priced by Lumicore. As discussed above, there is no commercial market for Lumicore PIC dies, no standalone PIC die product offering, and no external PIC die pricing. The PIC die has no alternative commercial use; it is designed exclusively for the accused transceiver modules. Therefore, even if the PIC die could theoretically be considered a "unit," it is not a "salable" unit in any commercially meaningful sense. The SSPPU doctrine requires identification of the smallest salable unit — i.e., a unit that is actually offered for sale in commercial transactions. A component that exists only as an intermediate work-in-progress within a manufacturer's internal supply chain does not qualify.
Second, even if the PIC die were deemed to be the SSPPU, the cost-allocation approach would not materially change the overall damages result in the manner Lumicore seeks. Per Lumicore's financial data produced in discovery (designated "Highly Confidential — Attorneys' Eyes Only"), the PIC die and associated photonic components account for approximately 65% to 70% of the transceiver module's total bill of materials ("BOM") cost. The remaining 30% to 35% consists of commodity electronic ICs (available from multiple third-party suppliers), standard fiber-optic connectors, and packaging materials — components that are widely available in the commercial market and that are not the basis for competitive differentiation among transceiver module vendors. Thus, even under a PIC-die-cost-based royalty approach, the royalty base would capture the vast majority of the module's BOM cost, and the resulting royalty rate would need to be proportionally higher to yield a reasonable royalty — a point that undercuts the purpose of disaggregation.
2. "The Individual Patented Sub-Components Are the SSPPU"
We further anticipate that Dr. Volkov may attempt to push the disaggregation even further, arguing that the SSPPU is not the PIC die but rather the individual patented sub-components — e.g., the waveguide multiplexer structure alone, or the grating coupler structure alone. This argument is even weaker than the PIC-die argument and should be rejected out of hand.
The individual patented features — the waveguide multiplexer, the coherent transceiver elements, the micro-channel heat sinks, and the grating couplers — are not sold, priced, or capable of being separately valued on any commercial basis. They are monolithically integrated into the PIC die during the semiconductor fabrication process. There is no market for individual waveguide multiplexer structures, no price list for grating couplers, and no independent valuation methodology that could assign a meaningful economic value to a single photonic structure within a monolithically fabricated PIC die. Attempting to impute such a value would be purely speculative and contrary to commercial reality.
The Federal Circuit has consistently rejected such artificial disaggregation. Where a patented feature is so deeply integrated into a product that it has no independent commercial existence, requiring the patentee to construct a hypothetical sub-component price as the royalty base would introduce unreliable, speculative elements into the damages calculation — precisely the outcome the SSPPU doctrine was designed to prevent. The SSPPU doctrine is intended to prevent over-compensation by tying damages to the patented feature's actual market footprint; it is not intended to manufacture under-compensation by requiring a fictitious disaggregation of inseparable product elements.
3. "The Royalty Rate Should Account for Non-Patented Features (Rate Apportionment)"
We anticipate that Lumicore will argue, in the alternative, that even if the module-level selling price is the appropriate royalty base, the royalty rate should be reduced to account for the contribution of non-patented features to the module's market value. This argument has some force as a general matter — the Federal Circuit in CSIRO recognized rate apportionment as a permissible approach — and we should address it forthrightly rather than dismissing it entirely.
Our position is that the proposed royalty rate of 8.5% already accounts for the contribution of non-patented features through the Georgia-Pacific factor analysis. The 8.5% rate is not an arbitrary figure; it is derived from a rigorous analysis of comparable licenses, each of which calculates royalties on a product-level (module-level) revenue base. The comparable licenses include:
• OWS-Terralink Semiconductor (2020): 6.5% running royalty on $45 million in licensed revenue, calculated at the module level;
• OWS-NovaWave Corp. (2021): 9.0% running royalty on $22 million in licensed revenue, calculated at the module level;
• OWS-Helix Integrated Photonics (2022): 7.75% running royalty plus a $2.5 million upfront payment on $68 million in licensed revenue, calculated at the module level;
• Stratos Photonics v. ArcLight Systems (2019 settlement): 8.0% royalty, calculated at the product level; and
• Coreline Technologies v. Prisma Semiconductor (2021 cross-license): 10.2% effective rate, calculated at the product level.
Each of these licenses was negotiated between sophisticated parties in the silicon-photonics industry, and each employs a module-level or product-level royalty base. The royalty rates in these licenses inherently reflect the negotiated allocation of value between patented and non-patented features at the product level. By deriving the 8.5% rate from these comparable licenses, Dr. Keating's methodology ensures that rate apportionment is already embedded in the damages calculation.
This is the approach Dr. Keating will employ in his expert report: a module-level royalty base multiplied by a royalty rate that is itself apportioned to reflect the patented contribution. This methodology satisfies the Federal Circuit's apportionment requirement under both VirnetX and CSIRO and is fully consistent with the EMVR framework.
V. QUANTITATIVE ANALYSIS
A. Royalty Base Calculation Using Module-Level Revenue
The following table summarizes the accused-product revenue data by product line, based on Lumicore's financial data produced in discovery and Dr. Keating's preliminary revenue projections:
The non-lost-profit sales base for the reasonable-royalty calculation is derived by subtracting the portion of revenue subject to the lost-profits claim under Theory 1. The market-share displacement factor of 38.2%, supported by the Market Share Displacement Analysis (internal C&H memo, June 2024), yields the following:
Total projected accused-product revenue through trial: $714.6M
Less: Revenue displaced to OWS (38.2%): $714.6M × 38.2% = $272.98M
Non-lost-profit revenue base: $714.6M − $272.98M = $441.62M
Reasonable royalty at 8.5%: $441.62M × 8.5% = $37.54M
This yields a reasonable-royalty damages figure of approximately $37.5 million, which is the amount Dr. Keating will present under Theory 2.
B. Comparison: Hypothetical Disaggregated Royalty Base (for Rebuttal Purposes)
To anticipate and rebut Lumicore's expected disaggregation arguments, we have analyzed the hypothetical impact of using a PIC-die-cost-based royalty base instead of the module-level revenue base.
PIC-Die-Cost Approach. Based on Lumicore's BOM data produced in discovery, the estimated PIC die cost ranges from approximately $55 to $75 per module, depending on the product line (the LumiEdge 1600G-LR, being the newest and most complex product, has the highest PIC die cost). Over the projected infringement period, the total PIC die cost for all accused products would represent approximately 25% to 30% of total module revenue. If Dr. Volkov uses PIC die cost as the royalty base, she would be working with a base that is roughly one-quarter to one-third of the module-level base. To yield the same total reasonable-royalty damages of $37.5 million on this reduced base, the royalty rate would need to be approximately 25% to 30% — a rate that is far above any rate observed in comparable licenses in the silicon-photonics industry and that would be difficult to defend as reasonable.
Further Disaggregation to Individual Patented Sub-Components. If Dr. Volkov attempts to further disaggregate the royalty base to the level of individual patented sub-components (e.g., the waveguide multiplexer alone, or the grating coupler alone), the resulting base would be even smaller and the required royalty rate even more extreme. More fundamentally, as discussed in Section IV.D.2 above, there is no market data, no separately manufactured component, and no independent valuation methodology that could support a cost allocation to a single photonic structure within a monolithically fabricated PIC die. Any such allocation would be speculative and should be excluded under Daubert.
Key Point. The disaggregation exercise demonstrates that artificially reducing the royalty base below the module level does not change the total amount of reasonable-royalty damages that is economically justified; it merely shifts the apportionment from the rate to the base, producing mathematically equivalent results but introducing speculative elements that make the analysis less reliable. The module-level base with an 8.5% rate is the more straightforward, more transparent, and more defensible approach.
C. Cross-Check Against Comparable Licenses
All five comparable licenses identified in the Comparable License Analysis (internal C&H memo, July 2024) calculate royalties on a product-level (module-level) revenue base, not on a sub-component basis. This is a critical cross-check because it demonstrates that the industry standard for licensing silicon-photonics patents is a module-level or product-level royalty base — which is exactly the approach we are proposing.
The comparable licenses are summarized below:
Several observations are noteworthy. First, all five licenses use the module/product-level selling price as the royalty base, not a sub-component cost. This confirms that sophisticated parties in the silicon-photonics industry — including OWS itself and its prior licensees — have consistently agreed that the appropriate royalty base for patents covering PIC-level technology is the module-level revenue. Second, the royalty rates in these licenses range from 6.5% to 10.2%, with a median of approximately 8.0%. The proposed rate of 8.5% falls comfortably within this range and is well-supported by the comparable-license evidence. Third, the licenses involving OWS's own patent portfolio (OWS-Terralink, OWS-NovaWave, OWS-Helix) are the most directly comparable because they involve the same licensor, the same technology, and the same industry segment. These licenses carry significant weight under Georgia-Pacific Factor 1 (royalties received by the patentee for licensing the patent-in-suit) and Factor 2 (rates paid by the licensee for comparable patents).
The comparable-license cross-check strongly reinforces the conclusion that the module-level royalty base and the 8.5% royalty rate are appropriate and well-supported.
VI. IMPLICATIONS FOR LOST-PROFITS ANALYSIS
Although the apportionment and EMVR analysis set forth in this memorandum is primarily directed at the reasonable-royalty calculation under Theory 2, it also supports and reinforces the lost-profits analysis under Theory 1.
The lost-profits analysis uses the full module revenue as the measure of lost sales. OWS's competing products — the OWS-400ZR and OWS-800XR transceiver module families — compete directly with Lumicore's accused products at the module level. Customers purchase complete transceiver modules, and when a customer chooses a Lumicore module instead of an OWS module, OWS loses a module-level sale. The 47.3% gross margin that serves as the basis for the lost-profits calculation is calculated on the full module selling price.
If the Court were to require apportionment of lost profits to reflect only the value of the patented features (as opposed to the entire module), the same body of evidence supporting the EMVR in the reasonable-royalty context would support using the full module price and margin as the lost-profits base. Because the patented features drive customer demand for the entire module, the full module selling price represents the economic value at risk when a customer chooses a Lumicore module over an OWS module. The customer is not choosing a waveguide multiplexer or a grating coupler in isolation; the customer is choosing a complete transceiver module whose competitive attributes are defined by the patented technologies.
The Panduit Factor Analysis (internal C&H memo, May 2024) and the Market Share Displacement Analysis (internal C&H memo, June 2024) provide detailed support for the 38.2% market-share displacement figure and the four-factor Panduit test. The apportionment analysis in this memorandum complements those analyses by establishing that the patented features account for the module-level competitive differentiation that underpins OWS's lost-profits claim.
VII. ANTICIPATED DAUBERT CHALLENGES AND PREPARATION
A. Expected Timing and Scope of Lumicore's Challenge
Lumicore is expected to file a Daubert motion challenging Dr. Keating's expert damages testimony.

MEMORANDUM — PRIVILEGED AND CONFIDENTIAL ATTORNEY WORK PRODUCT

I. EXECUTIVE SUMMARY

II. BACKGROUND

III. LEGAL FRAMEWORK

IV. APPLICATION TO THE ACCUSED LUMICORE TRANSCEIVER MODULES

V. QUANTITATIVE ANALYSIS

VI. IMPLICATIONS FOR LOST-PROFITS ANALYSIS

VII. ANTICIPATED DAUBERT CHALLENGES AND PREPARATION

Record

Parties

Dates

Amounts