# Huawei Optical Capability and Data-Center Optics Risk: Implications for Lumentum (LITE) and Marvell (MRVL)

**Assessment date:** July 9, 2025  
**Purpose:** Fundamental competitive-risk analysis; not personalized investment advice, a price target, or a buy/sell recommendation.

## Executive answer

**Huawei is a credible optical-telecommunications, data-center-networking, and datacom-adjacent competitor.** The public record available by July 9, 2025 shows Huawei-branded coherent/DCI capability, 400GE/800GE switching, named 400G short-reach pluggables, and a dated Huawei portfolio that names 800G datacom modules. Those facts make it incorrect to dismiss Huawei as “telecom only.”

**But capability is not the same thing as a demonstrated global merchant-component threat.** The same public record does not establish that Huawei owned the relevant lasers/PICs or Ethernet DSPs, produced modules at merchant scale, had SKU-specific independent interoperability/qualification, or had a named non-Huawei OEM, hyperscaler, distributor, host platform, or deployed customer route for the cited products. It also does not identify a Huawei product/channel that displaced a Lumentum product or a named Marvell DSP/interconnect/switch/custom-silicon program. This is an evidence limitation—not proof that private qualification or sales did not exist.

**Bottom line for the two companies:**

- **Lumentum (LITE):** Huawei matters more directly as a historical customer/export-control and potential localized-optics/price-competition issue than as a proven current merchant-module displacer. Lumentum is materially exposed to cloud/AI/DCI optical demand at a broad reporting level, and its historic Huawei relationship demonstrates sensitivity to a direct channel being cut off. The magnitude of any current Huawei-competition impact is not publicly quantifiable from the disclosed data.
- **Marvell (MRVL):** Huawei is a conditional localization, pricing, architecture, and China-customer risk—not a documented direct displacement threat to named Marvell programs on this record. Marvell’s optical DSP, interconnect, Ethernet, and data-center portfolio is adjacent to Huawei systems and modules, but a Huawei system/module catalogue does not prove substitution of Marvell silicon. China-related localization risk is economically relevant, but shipment geography is not end-demand geography and no Huawei-to-Marvell program mapping is disclosed.

The supportable conclusion is therefore **credible China-centric/integrated-system/datacom-adjacent risk, potentially material through local customers, bundles, and price competition—but not a proven globally scalable, externally qualified merchant replacement threat to LITE’s lasers/components or MRVL’s named DSP/interconnect programs as of the assessment date.**

---

## 1. What Huawei can demonstrably do

### 1.1 Capability map: separate product layers before drawing a competitive conclusion

| Product layer | What the eligible record shows | Fact / inference status | What it does **not** show by itself |
|---|---|---|---|
| Telecom / coherent DCI | Huawei announced a 200–800G tunable OptiXtrans module for backbone, metro, and data-center interconnection in 2020. | **Fact** [H6] | It is not evidence of 800GbE short-reach Ethernet PMD, an AI-fabric pluggable, a merchant Ethernet DSP, or a non-Huawei host qualification. |
| Data-center switching | Huawei announced a 128×800GE CloudEngine XH9330 data-center switch in March 2025. | **Fact** [H5] | Switch-port capacity does not identify a compatible third-party or merchant 800G module, its BOM, or its customer route. |
| Named 400G short-reach optics | Huawei documentation identifies 02313KST / `QSFP-DD-400G-850-0.1km` and 02314WGY / `QSFP-DD-400G-SR8-AI`, with MMF/MPO-16/850nm/70–100m characteristics; the latter is documented in a CloudEngine context. | **Fact (issuer product documentation)** [H1] [H2] | These facts do not prove Huawei’s ownership of laser/PIC/DSP/package technology, external interoperability, customer use, volume, or a LITE/MRVL displacement. |
| Named 400G/800G datacom portfolio | Huawei’s April 24, 2025 short-distance DCN portfolio names `QSFP-DD-400G-SR8-D`, `OSFP-800G-SR8-D`, and `OSFP-800G-2*FR4-D`. The 400G SKU is labelled “Direct Sales Model.” | **Fact (issuer portfolio)** [H3] | The label does not identify a named reseller, purchaser, non-Huawei host, sale, deployment, compatible OEM/NIC/switch, volume, or geography. |
| Data-center module positioning | Huawei’s March 2025 StarryLink announcement positioned optical modules for data-center networks and stated that the launch session had 100+ unnamed customers/partners from several countries. | **Fact for launch/audience; issuer claims for claimed performance** [H4] | An unnamed launch audience is not a named customer, qualification, order, or deployment. |
| Short-reach competitive class | The cited 400G SR8/MMF/MPO-16 70–100m profiles are technically in the class of short-reach 400GbE data-center fabric optics. | **Bounded inference from product specifications** [H1] [H2] [H3] | Technical adjacency is not commercial substitution or external merchant parity. |

### 1.2 Named product evidence and the key 400G/800G distinction

The pre-cutoff record is strongest for **product identity**. Huawei’s October 31, 2024 document specifies 02313KST as a QSFP-DD 400G short-reach MMF module with MPO-16 connectivity, 850nm optics, and stated 70m/100m reaches; its May 8, 2025 CloudEngine documentation specifies 02314WGY as a 400GBASE-SR8 module. [H1] [H2] Huawei’s April 24, 2025 DCN portfolio additionally names one 400G direct-sale model and two 800G models. [H3]

The appropriate conclusion is therefore not “Huawei has no data-center optics.” It is: **Huawei had public named datacom-pluggable products by the cutoff, including 400G short-reach items and named 800G DCN items.**

However, the record must not combine distinct technology layers:

- Huawei’s 2020 800G tunable OptiXtrans product is credible coherent/transport/DCI evidence, but does not demonstrate 800GbE client-side short-reach module parity. [H6]
- A 128×800GE switch is evidence of Huawei system capability, not a substitute for a named external 800G pluggable qualification or merchant DSP design win. [H5]
- IEEE 802.3bs and QSFP-DD form-factor materials establish standards/form-factor context, not certification of a Huawei SKU or multivendor interoperability. [S1] [S2]

### 1.3 Source table for Huawei capability

| ID | Original source (publisher; date) | Use in this report | Evidence type / limitation |
|---|---|---|---|
| H1 | Huawei, *NetEngine 8000 F1A and 8000E F2C … Hardware Description*, EDOC1100410972; October 31, 2024 | 02313KST 400G short-reach attributes | Primary issuer product table; not independent performance, channel, or customer proof. |
| H2 | Huawei, *CloudEngine 9800, 8800, and 6800 Series Switches Hardware Description*, EDOC1100369504; May 8, 2025 | 02314WGY / 400GBASE-SR8 short-reach record | Primary issuer product table; CloudEngine context, not external qualification. |
| H3 | Huawei, *Data Center Network Short-Distance Optical Module Portfolio*; April 24, 2025 | Named `QSFP-DD-400G-SR8-D`, `OSFP-800G-SR8-D`, `OSFP-800G-2*FR4-D`; direct-sale label | Primary issuer portfolio; no buyer, external host, volume, or deployment. |
| H4 | Huawei Enterprise, *Huawei Unveils StarryLink Optical Modules…*; March 4, 2025 | Data-center module positioning and unnamed international audience | Primary marketing announcement; no SKU-linked sale, test, or named customer. |
| H5 | Huawei Enterprise, *Huawei Recognized as a Leader…*; March 31, 2025 | 128×800GE switch context | Issuer marketing/system evidence; not a module datasheet or independent test. |
| H6 | Huawei, *Huawei Launches Industry’s First 800G Tunable…*; February 26, 2020 | Coherent/DCI boundary | Primary product announcement; explicitly a different product layer from short-reach Ethernet. |

---

## 2. Why capability does—and does not—equal competitive threat

### What capability can mean economically

Huawei’s integrated position can matter even before one can prove global merchant-component parity. It can package switches, transport/DCI equipment, optical modules, systems integration, service, and local procurement relationships. That can create risk through bundled offers, customer preference for a locally integrated supplier, local support, price discipline, or architecture choices in China-facing accounts. The named module record makes the possibility of **datacom-adjacent pricing/localization competition** more concrete than a telecom-only thesis.

### Why the direct merchant-threat conclusion is not established

A competitive threat to LITE or MRVL needs more than a product catalogue. The public record lacks affirmative evidence for several bridges:

| Required bridge | Eligible evidence | Result as of July 9, 2025 |
|---|---|---|
| Merchant component ownership / scale | Huawei tables give module attributes; they do not identify laser source, PIC ownership, DSP/firmware part or design ownership, package architecture, electrical power, yield, output, price, or merchant volume. | **Not established.** Missing disclosure is not evidence of absence. |
| External qualification | The inspected public Ethernet Alliance/OIF records do not identify a Huawei 400G/800G short-reach product demo or qualification path; Huawei-system tests do not establish module qualification outside Huawei systems. | **Not established on the reviewed public record.** Event nonappearance is a bounded null result, not proof that private tests did not occur. [E1] [E2] |
| Named external channel / customer | A “Direct Sales Model” label and a global launch are outward-market indications, but no named external distributor, purchaser, OEM, cloud customer, host, order, or deployment is disclosed for the cited products. | **Not established.** [H3] [H4] |
| Non-Huawei host compatibility | No named non-Huawei switch, NIC, OEM, hyperscaler, or AI-fabric qualification is tied to a Huawei SKU. | **Not established.** |
| Company-specific substitution | No public evidence maps a Huawei module to a Lumentum component/program or to a material named Marvell DSP/interconnect/switch/custom-silicon program. | **Not established.** |
| Reachable market | Huawei’s base is demonstrably China-centered and ex-China restrictions/friction are material but jurisdiction-, buyer-, product-, and transaction-specific. The reviewed record does not support either a universal ex-China ban or a globally addressable merchant-optics conclusion. | **Segment- and geography-specific, not globally resolved.** |

**Inference:** Huawei’s systems capability raises a real competitive option value, especially in China/localized networks. The absence of the bridges above prevents that option value from being converted into a supported claim of a globally scalable merchant replacement threat.

---

## 3. Implications for Lumentum (LITE)

### 3.1 Direct business channels

Lumentum is not merely a telecom-optics company. Its Cloud & Networking segment supplies optical/photonics chips, components, modules, and subsystems to cloud-data-center operators, AI/ML infrastructure providers, network-equipment manufacturers, DCI, communications-service-provider, and enterprise customers. Lumentum’s Cloud Light acquisition also brought advanced optical modules for data-center interconnect applications. [L1] [L2]

In the quarter ended March 29, 2025, Cloud & Networking represented $365.2 million, or 85.9%, of net revenue; in the nine-month period it represented $986.7 million, or 84.7%. These are useful **broad exposure envelopes**, not data-center-only, module-only, China-end-demand, Huawei-exposed, or Huawei-competition-exposed revenue. [L1]

### 3.2 Historical Huawei restriction channel

Huawei was historically a material direct Lumentum customer: Lumentum reported Huawei at 15.2% of FY2019 revenue; applying that rounded percentage to disclosed FY2019 revenue yields an approximate $237.9 million, clearly a derived—not issuer-reported—dollar figure. [L3] In May 2019 Lumentum said it stopped Huawei shipments and reduced FY2019 Q4 revenue guidance from $405–425 million to $375–390 million, citing the export restriction, discontinued sales, and capacity-repurposing time. The guidance change should not be represented as a Huawei-only realized loss. [L4]

By the beginning of calendar 2024, Lumentum had stopped all Huawei shipments; its FY2025 Q3 filing described the relationship as completely restricted and said it could not work with Huawei on future product development. The same filing says the restriction had negatively affected Huawei revenue and financial results. A FY2023 $2.7 million excess/obsolete inventory charge was expressly linked to U.S. trade restrictions and lower Huawei demand. [L1] [L2]

This is important: **LITE’s clearest direct Huawei exposure in the public record is a historical customer/export-control relationship, not evidence that Huawei has won an external competing module design against Lumentum.**

### 3.3 Sensitivity and limits on quantification

Lumentum disclosed that Cloud & Networking gross margin improved 3.1 percentage points in FY2025 Q3, primarily due to mix toward higher-margin cloud/AI/ML and transport products, while noting that the market is competitive and price sensitive. [L1] That supports a qualitative conclusion that product mix and pricing matter to profitability.

It does **not** support a numeric Huawei competition haircut. Lumentum disaggregates revenue principally by segment and shipment geography, not by data-center product, customer, end market, China end demand, Huawei-linked channel, or product margin. Initial shipment location can be a contract-manufacturer location rather than end-customer location. [L1] Therefore Asia-Pacific, Hong Kong, or Thailand shipment figures cannot be used as a proxy for China demand or Huawei exposure.

### 3.4 LITE assessment

| Consideration | Assessment |
|---|---|
| Relative sensitivity to Huawei-related risk | **Moderate and more direct than MRVL in components/modules and historical channel exposure, but not quantifiable.** |
| Most credible risk channel | China/local customer preference, module/component price pressure, and a future externally qualified Huawei merchant route; separately, export/customer restrictions and stranded customized inventory. |
| What is not proven | Any present Huawei module displacement of a Lumentum product, a current Huawei revenue figure, a data-center-only exposure denominator, or a Huawei-specific margin effect. |
| Evidence that would worsen the thesis | Dated proof that a Huawei 400G/800G SKU is qualified in non-Huawei hardware and wins a named cloud/OEM or China-local account that uses Lumentum-relevant optics; disclosed pricing/margin pressure or loss of named Lumentum program. |
| Counterarguments | Lumentum’s Cloud/AI/DCI activity and mix can remain driven by non-Huawei customers, and the global customer/qualification bridge for Huawei is not established. |

---

## 4. Implications for Marvell (MRVL)

### 4.1 Portfolio adjacency is real, but it is not direct displacement evidence

Marvell’s FY2025 10-K describes optical PAM/coherent-lite DSPs, drivers, TIAs, silicon photonics, pluggable DCI, CPO, LPO chipsets, AEC DSPs, PCIe retimers, Ethernet controllers/adapters/PHYs, and switches. It says PAM DSPs enable interconnection among servers, routers, switches, storage, and related infrastructure inside data centers. [M1]

That creates a genuine product-layer adjacency: Huawei can sell systems, switches, modules, and DCI equipment in fields where Marvell supplies silicon and interconnect technology. But the layers are not interchangeable:

- A Huawei switch or pluggable does not establish a replacement for a Marvell PAM4 DSP, coherent-lite DSP, TIA, laser driver, PHY, switch ASIC, retimer, custom compute product, or CPO/LPO building block.
- A Huawei module that may contain an “oDSP” in marketing terminology does not identify the DSP make, ownership, architecture, process node, firmware ownership, or whether it competes with a specified Marvell part. [H4]
- No eligible evidence names a material Marvell program, a Huawei external customer/channel, or a non-Huawei deployment in which technical and commercial substitution can be compared.

### 4.2 China, localization, and pricing risk

Marvell reported $4.164 billion of FY2025 revenue from the data-center end market (72% of total), demonstrating meaningful exposure to data-center demand. Its FY2025 10-K also warns that China customers may develop indigenous solutions or replace products affected by export controls; it reports China destination-of-shipment revenue of $2.508 billion (43% of FY2025 revenue), while expressly cautioning that a substantial majority of China shipments relate to non-China customers with factories or contract manufacturing in China. [M1]

The correct inference is not “43% of MRVL demand is China end demand.” It is that **China/localization and export-control dynamics can be economically important to Marvell, while the disclosed shipment figure cannot be treated as an end-demand or Huawei-exposure figure.** A Huawei-led local bundle or local architecture could contribute to pricing/substitution pressure in qualifying accounts, but the record does not identify that event or quantify its impact.

### 4.3 MRVL assessment

| Consideration | Assessment |
|---|---|
| Relative sensitivity to Huawei-related risk | **Conditional / indirect relative to LITE.** Marvell has broad data-center/interconnect exposure and China/localization risk, but no direct named Huawei-to-Marvell program overlap. |
| Most credible risk channel | China customer localization, price competition, local system architecture choices, and export-control-induced substitution toward indigenous alternatives. |
| What is not proven | Huawei displacement of a named Marvell DSP, interconnect, switch, CPO/LPO, custom-silicon, or DCI program; a Huawei-owned merchant Ethernet-DSP business; or an externally qualified Huawei route that reaches a material Marvell customer. |
| Evidence that would worsen the thesis | Named Huawei external module/silicon compatibility plus a disclosed win at a non-Huawei host/customer that intersects a material Marvell product category; disclosed China/local customer replacement of a Marvell program; loss of a named design win or pricing pressure tied to local alternatives. |
| Counterarguments | Marvell’s portfolio is multi-layered and customer-/architecture-specific; systems/modules can coexist with Marvell silicon. System adjacency alone cannot identify direct component substitution. |

---

## 5. Risk matrix and scenario conditions

These are business-risk conditions, not forecasts of revenue, earnings, valuation, or share price.

| Scenario | Conditions that would support it | LITE implication | MRVL implication | Evidence status at July 9, 2025 |
|---|---|---|---|---|
| **Base case: credible but bounded Huawei risk** | Huawei sells/uses integrated datacom systems and named modules primarily in its ecosystem and China-centered channels; no public non-Huawei qualification/customer/volume bridge appears. | Qualitative pricing/localization risk plus ongoing legacy restriction/inventory sensitivity; no supported numeric impact. | Localization/export-control/architecture risk remains conditional; no direct program displacement. | **Best supported.** |
| **Downside case for incumbents: merchant route becomes credible** | A dated pre- or post-cutoff record identifies an exact Huawei 400G/800G SKU, named distributor/buyer/non-Huawei host, independent interop or qualification, production/volume, and an account/product overlap. | Competitive modules/components could pressure a disclosed or identifiable LITE-relevant product/customer; margin sensitivity would be more relevant. | A named Huawei product/channel could be mapped to a material MRVL DSP/interconnect/networking program; China localization could become more direct. | **Not established at cutoff.** |
| **Downside case: China/local bundle and price pressure without global merchant scale** | China/local customers procure Huawei-integrated switches/modules/systems; credible tender, pricing, or deployment evidence shows bundle-driven substitution or price discipline. | More plausible localized module/component pricing risk, even if ex-China merchant scale is absent. | More plausible local architecture and price risk, especially where customers seek indigenous alternatives. | **Plausible inference; not quantified or directly evidenced for either issuer.** |
| **Lower-Huawei-risk / incumbent-upside case** | No material non-Huawei qualification, customer, volume, or external channel emerges; buyers retain multivendor qualification barriers, ex-China restrictions/friction, and incumbent supply relationships. | Huawei remains an ecosystem/local competitor rather than a broad merchant replacement; LITE’s drivers remain primarily its broader cloud/AI/DCI mix. | Huawei hardware does not translate into displacement of Marvell silicon; MRVL remains driven by its own design wins, architecture mix, and customer demand. | **Consistent with present evidence but not a guarantee.** |

---

## 6. Decision-relevant monitoring list

The following evidence would materially change the conclusion; broad marketing claims alone would not.

1. **Named non-Huawei qualification:** OEM/NIC/switch compatibility matrix, hyperscaler qualification, independent test report, or production deployment linking an exact Huawei 400G/800G SKU to a non-Huawei environment.
2. **Named commercial route:** Dated authorized distributor listing, purchase disclosure, tender award, customer announcement, or shipment/volume evidence tied to a specific Huawei SKU.
3. **Independent interoperability:** Huawei product participation in a credible 400G/800G short-reach multivendor demonstration with an identifiable SKU/host path—not merely a standards membership, event attendance, or coherent/DCI demonstration.
4. **Component ownership and scale:** Disclosed laser/PIC/DSP/firmware/packaging ownership, electrical power, production capacity/yield, pricing, and merchant sales volume for Huawei pluggables.
5. **Lumentum issuer disclosure:** Data-center/product/customer-level revenue or margin, disclosed China end-demand exposure, a named program loss, a Huawei-competition comment, or recurring inventory/write-down consequences tied to a specified product.
6. **Marvell issuer disclosure:** Named China/local substitution of optical DSP/interconnect/switch/custom-silicon programs, loss of a design win, customer pricing pressure from indigenous alternatives, or product-specific revenue/margin data.
7. **Geography and regulation:** Product- and buyer-specific official rules or procurement actions. Generic bans, generic localization policy, and shipment destination data are insufficient to determine a serviceable market or company-specific impact.

---

## 7. Evidence quality, contradictions, uncertainty, and falsification

### Evidence quality

- **High for Huawei product identity and stated specifications:** dated Huawei support tables and portfolio documents are direct evidence of what Huawei publicly documented, though still issuer evidence. [H1] [H2] [H3]
- **High for Lumentum reporting and historical Huawei restrictions:** SEC filings and contemporaneous investor-relations material establish disclosed reporting limits and historic direct relationship facts. [L1] [L2] [L3] [L4]
- **High for Marvell portfolio and aggregate market/geography disclosures:** Marvell’s FY2025 10-K is primary issuer/SEC evidence, but it is not Huawei overlap evidence. [M1]
- **Limited for external qualification:** the public event records inspected are useful only for their document-specific contents; no Huawei listing in one event record cannot prove no qualification elsewhere. [E1] [E2]

### Material contradictions resolved

1. **“No named 800G SKU” versus named 800G portfolio products:** the module-parity source set did not itself identify a named 800G short-reach SKU, whereas the separate April 2025 Huawei portfolio did. This is a source-scope refinement, not a conflict: the broader eligible record supports named 800G product identity, but not customer/channel/qualification. [H3]
2. **“Direct Sales Model” versus no merchant parity:** the label supports a Huawei-described standalone outward-sales classification. It does not prove a named channel, sale, host compatibility, qualification, volume, or merchant scale. [H3]
3. **Huawei system tests versus external module qualification:** a Huawei-system/platform test supports system maturity, not a merchant-module qualification in a named non-Huawei host/customer environment.
4. **Large Asia/China shipment figures versus China end demand:** Lumentum and Marvell both describe shipment-location limitations. Shipment destinations cannot be converted into China end demand, Huawei demand, or direct competitive exposure. [L1] [M1]

### What would falsify or materially change this conclusion

The conclusion would be upgraded from a bounded competitive risk to a demonstrated direct threat if credible evidence showed **all or most** of: (1) a named Huawei 400G/800G product; (2) non-Huawei host compatibility or independent interop; (3) a named external distributor/customer/production deployment; (4) production scale/pricing or component ownership; and (5) a specific LITE or MRVL product/customer/design-win overlap.

Conversely, evidence that Huawei’s relevant modules remained confined to Huawei systems, lacked a viable external channel, or could not meet independent host/qualification requirements would reduce the merchant-threat case—but the current absence of public proof cannot be treated as proof of those negative propositions.

---

## Bottom-line answer

**Yes—Huawei is a risk factor for data-center optics, but the risk must be specified correctly.** Huawei’s publicly documented 400G short-reach products, named 800G DCN portfolio entries, 800GE switching, and DCI/telecom capability establish that it is more than a telecom-only vendor. Huawei can matter through China/local customers, integrated systems, bundled modules, and price/localization pressure.

**No—the July 9, 2025 public record does not prove that Huawei is already a globally scalable, externally qualified merchant replacement for Lumentum’s lasers/components or Marvell’s named DSP/interconnect programs.** The record lacks the external-channel, non-Huawei qualification, component/scale, and company-specific substitution evidence necessary for that stronger conclusion. The lack of public evidence is a meaningful uncertainty, not proof of no private capability.

For **LITE**, the more direct supported issue is historical Huawei customer/export restriction exposure plus a qualitative future pricing/localization risk in broad cloud/AI/DCI optics. For **MRVL**, the supported issue is conditional China/localization and architecture/price risk across a broad data-center/interconnect portfolio; direct Huawei displacement of named MRVL programs is not demonstrated. Neither conclusion provides a stock-price forecast or a trading recommendation.

---

## Source appendix

All links below identify original sources used for facts in this report. Documents were retrieved after the historical cutoff solely to inspect sources published on or before July 9, 2025; no later operating or product claims are used as evidence for the assessment-date conclusion.

| ID | Publisher | Original publication / filing date | URL or document ID | How used |
|---|---|---:|---|---|
| H1 | Huawei | 2024-10-31 | https://support.huawei.com/enterprise/en/doc/EDOC1100410972/46b8c83c/400gbps-qsfp-dd-optical-module | Named 02313KST 400G short-reach product and stated attributes. |
| H2 | Huawei | 2025-05-08 | https://support.huawei.com/enterprise/en/doc/EDOC1100369504/e12996d3/qsfp-dd-400g-sr8-ai | Named 02314WGY 400GBASE-SR8 product and CloudEngine context. |
| H3 | Huawei | 2025-04-24 | https://e.huawei.com/marketingcloud/pep/asset/20000001/Material/e939c63850d344d69077245292e7b005/M3T1A590N1122226626417115403/Huawei%20Data%20Center%20Network%20Short-Distance%20Optical%20Module%20Portfolio.pdf | Named 400G/800G DCN portfolio entries and `Direct Sales Model` label. |
| H4 | Huawei Enterprise | 2025-03-04 | https://e.huawei.com/eu/news/2025/solutions/enterprise-network/starrylink-optical-modules-3s | StarryLink data-center launch and unnamed international audience; issuer claims labelled as such. |
| H5 | Huawei Enterprise | 2025-03-31 | https://e.huawei.com/eu/news/2025/solutions/enterprise-network/2025-gartner-magic-quadrant | 128×800GE CloudEngine data-center switch context. |
| H6 | Huawei | 2020-02-26 | https://www.huawei.com/en/news/2020/2/800g-tunable-ultra-high-speed-optical-module | Coherent/transport/DCI boundary. |
| S1 | IEEE Standards Association | 2017-12-12 | https://standards.ieee.org/ieee/802.3bs/6748/ | 200/400GbE standards context; not Huawei certification. |
| S2 | QSFP-DD MSA | 2024-06-25 | https://www.qsfp-dd.com/wp-content/uploads/2024/07/QSFP-DD-Hardware-Rev7.1.pdf | QSFP-DD/QSFP-DD800 form-factor context; not Huawei certification. |
| E1 | Ethernet Alliance | 2024-04-01 | https://ethernetalliance.org/wp-content/uploads/2023/08/EA_OFC24Handout.pdf | Bounded inspection of public 400G/800G multivendor interop record. |
| E2 | Optical Internetworking Forum (OIF) | 2025-03-12 | https://www.oiforum.com/oif-advances-interoperability-at-ofc-2025-with-live-demos-expert-insights-and-cross-industry-collaboration/ | Bounded inspection of public OFC 2025 interop announcement; scope limitation noted. |
| E3 | CAICT / IMT-2020 (5G) Promotion Group | 2019-01-21 | https://www.caict.ac.cn/kxyj/qwfb/bps/201901/P020190123587216630553.pdf | Telecom/transport test boundary; not Huawei external 400G/800G module qualification. |
| L1 | Lumentum / U.S. SEC | 2025-05-06 | SEC accession 0001628280-25-022788; https://www.sec.gov/Archives/edgar/data/1633978/000162828025022788/lite-20250329.htm | FY2025 Q3 segment/mix/margin/geography/restriction disclosure. |
| L2 | Lumentum / U.S. SEC | 2024-08-21 | SEC accession 0001628280-24-038024; https://www.sec.gov/Archives/edgar/data/1633978/000162828024038024/lite-20240629.htm | Product scope, Cloud Light, and FY2023 named Huawei inventory charge. |
| L3 | Lumentum / U.S. SEC | 2019-08-27 | SEC accession 0001633978-19-000069; https://www.sec.gov/Archives/edgar/data/1633978/000163397819000069/lumentumfy19-10k.htm | Historical Huawei 15.2% customer concentration. |
| L4 | Lumentum | 2019-05-20 | https://investor.lumentum.com/financial-news-releases/news-details/2019/Lumentum-Provides-Update-on-U-S--Department-of-Commerce-Entity-List-Designation-of-Huawei-and-Affiliates-and-the-Impact-to-Fiscal-Fourth-Quarter-Ending-June-29-2019-Outlook-2019-5-20/default.aspx | 2019 shipment stop and revised guidance. |
| M1 | Marvell Technology / U.S. SEC | 2025-03-12 | SEC accession 0001835632-25-000057; https://investor.marvell.com/sec-filings/all-sec-filings/content/0001835632-25-000057/mrvl-20250201.htm | Marvell optical/interconnect portfolio, FY2025 data-center exposure, China/localization risk, and shipment-location caveat. |

### Excluded look-ahead material

Huawei materials dated July 25, October 23, November 14, and December 15, 2025 that used more expansive 800G, independent-sales, or multivendor language were identified but excluded. They cannot establish the state as of July 9, 2025.