DCI Patent Atlas — Data Center Interconnect patent corpus, value chain and citation network

Data Center Interconnect (DCI) carries traffic between data centres over coherent optical links. This project assembles the patent record behind that technology, places every application on the DCI value chain, and reconstructs how the field cites itself. Everything needed to reproduce the result — data, code, queries and the report — is published here.

745,624Applications

1980–2025Priority years

2,435,797Citation edges

149Applicant countries

What the corpus contains

The corpus is built in two layers. A seed of 203,711 applications was retrieved by classification, using the IPC and CPC code groups that define coherent optical transmission, lasers and modulators, photonic integration, and fibre and cable. Its citation columns then define a neighbour layer: every application the seed cites or is cited by, 541,913 in total. Both layers carry the same 21 descriptive fields, so the corpus can be analysed as one table.

Table 1. Corpus composition. Unique applications, no duplicate identifiers.
Layer	Applications	Fields
Seed (classification-based)	203,711	21 + 3 citation lists
Citation neighbours	541,913	21
Unified corpus	745,624	21 + 3 + provenance

Bibliographic fields are complete for every record. IPC codes are present for 99.1% of applications, applicant names for 96.4%, abstracts for 94.6% and applicant countries for 68.4% after imputation. Country coverage is the binding constraint on country-level analysis, and it is not missing at random: it correlates with jurisdiction, so results are checked against the complete-country subsample.

DCI patent applications by earliest priority year, 1980 to 2025 — Figure 1. Applications by earliest priority year. The dashed grey line marks the annual mean. Filings rise through the late 1990s, plateau across the 2000s and peak in the 2015–2019 window; the fall after 2020 is publication lag, not a decline in filing.

Where the patents are

The United States holds 269,718 applications by applicant country, ahead of Japan, China and Korea. The gap between filings at the Chinese office (143,942) and applications with a Chinese applicant country (34,321) reflects both domestic filing behaviour and the weaker country coverage of Chinese person records.

Table 2. Leading applicant countries and patent offices.
Applicant country	Applications	Patent office	Applications
United States	269,718	US	360,874
Japan	71,482	CN	143,942
China	34,321	JP	88,100
Korea	27,824	WO	51,152
Germany	22,682	EP	41,535
Total (149 countries)	537,506	Total (85 offices)	745,624

Applications by leading applicant country across five-year windows — Figure 2. Leading applicant countries across five-year windows, against the all-country mean.

Position on the value chain

Each application is placed on the DCI value chain by two routes. Classification codes decide position from the technology recorded in IPC and CPC; the applicant list decides position from the filer's known place in the industry. Membership is multiple, and the two routes overlap, so the total is a union rather than a sum.

Why design and architecture needs the applicant route. System and interconnect design is not expressed in the classification scheme, so the reference table assigns it no codes at all. Its 5,318 applications — Broadcom, Avago, Marvell, NVIDIA, Celestial AI — are visible only through applicants. Classification alone would show this layer as empty.

Table 3. Value-chain positions. Multiple memberships permitted.
Position	Classification	Applicant	Both	Total
0 Materials and substrate	17,105	6,878	1,246	22,737
1 Design and architecture	0	5,318	0	5,318
2 Optical core components	143,853	23,392	6,198	161,047
3 Module assembly	89,024	4,032	2,299	90,757
4 Packaging (CPO, SiPh)	62,426	16,460	2,014	76,872
5 Systems, submarine, operations	217,685	122,665	62,974	277,376
Unique applications in scope				370,301

Value-chain positions: totals by identification route and trajectories over time — Figure 3. Value-chain positions. a, totals by identification route. b, trajectories across five-year windows, against the mean across positions.

The upstream is almost invisible

Linking the trade codes of the reference framework to the corpus through their classification codes exposes a sharp asymmetry. The two transceiver headings stand on large patent stocks, while the materials chokepoint that dominates supply-chain discussion barely registers: gallium, germanium and indium link to 136 applications, doped wafers to 1,557. Control over these inputs is exercised through capacity and export licensing, not through patents, which is precisely why patent data alone cannot see it.

Table 4. Applications linked to HS headings through the classification concordance. These are patent counts, not trade volumes.
HS heading	Traded good	Linked applications
`8517.62`	Transmission and switching equipment	118,075
`8517.79`	Parts of 8517 apparatus	97,453
`8541.49`	Laser diodes, photodiodes, APDs	35,983
`9001.10`	Optical fibres and bundles	23,276
`8542.31`	Processors and controllers	15,135
`3818.00`	Doped wafers (InP, GaAs)	1,557
`8112.92`	Gallium, germanium, indium	136

How the field cites itself

The seed's citation columns expand into 2,435,797 directed edges over 698,834 applications, and node attributes resolve for effectively every endpoint. Flows between countries are markedly asymmetric: China cites the United States 38,391 times against a far smaller reverse flow, the signature of a follower position in the optical layer, whereas the United States and Japan exchange citations in both directions.

Figure 4. Citation flows between the ten most active applicant countries, on a logarithmic scale, with outgoing citations per country.

Reproducing this

The published dataset is the output of collection and cleaning; every figure and table above is regenerated by the notebook from that file.

git clone https://github.com/deep1003/dci-patent-atlas.git
cd dci-patent-atlas
pip install -r requirements.txt
jupyter lab notebooks/dci_patent_atlas_analysis.ipynb

The notebook is organised as tasks with steps underneath, one operation per cell. Collection from PATSTAT is documented rather than executed, since it needs credentials; the SQL and the collector are published in sql/ and scripts/. A note on method: retrieving all fields in one statement is refused by the PATSTAT cost estimator under the 2026 Spring edition, and a reduced variant returns two records in 352 seconds. Splitting the retrieval into four set-based statements joined locally returns 850 applications in 0.87 seconds.

data/ — analysis-ready corpus (745,624 applications), citation edges, country flows, reference workbook
notebooks/ — the analysis notebook, executed end to end
scripts/ — collector, framework mapping, figure style
sql/ — the four set-based PATSTAT statements
tables/, figures/ — published aggregates and figures
report/ — technical report with full appendices

Data note. The published file omits titles, abstracts and person names to stay within repository size limits; it retains identifiers, dates, offices, families, countries, full IPC and CPC codes and value-chain flags. Keyword-based counts, which need the text fields, ship as pre-computed tables. Full records can be rebuilt from PATSTAT with the published SQL.