Apple Inc. — Patent Filing Strategy 2020–2024

Executive Summary

Across filing years 2020 to 2024, Apple Inc. filed 13,934 distinct patent families from 32,851 individual applications in 27 patent offices — roughly 2.4 applications per invention. The portfolio is anchored in the United States: 82% of all families (11,464) contain a US filing. China is the second destination with 4,967 families (36%), ahead of the international PCT route with 4,215 (30%) and the European Patent Office with 2,474 (18%). South Korea follows at 1,217 families, then Germany (636), Japan (488) and Australia (357).

Technologically the portfolio rests on two pillars of almost identical weight: digital communication (5,480 families) and computer technology (5,476). Audio-visual technology (2,396) and telecommunications (2,299) form the second tier, optics (1,286), measurement (1,024) and electrical machinery (1,022) the third. At IPC subclass level the same picture sharpens: electric digital data processing (G06F, 4,448 families) and wireless communication networks (H04W, 4,408) lead by a wide margin, followed by transmission of digital information (H04L, 2,927), image data processing (G06T, 1,482) and transmission (H04B, 1,452). The wireless weight is mirrored in the inventor data, where the most frequently recorded names sit in cellular-standards engineering.

Three structural features stand out. First, Apple files essentially alone: only 28 families (0.2%) carry an institutional applicant other than Apple, and the partners behind them are small US laboratories and recently acquired technology firms. Second, the annual volume is flat rather than growing — 6,381, 6,732 and 6,705 applications in 2021, 2022 and 2023, within ±6% of one another. Third, the company keeps its oldest inventions in force: its most-cited families — multipoint touch screens, the intelligent automated assistant, unlocking by gesture — have earliest filings between 2002 and 2014 and are still being extended by applications filed inside this window.

Apple's filing strategy in this period reads as consolidation rather than expansion: a stable annual output, concentrated in two technology pillars, filed almost exclusively in its own name, and defended through continuations of inventions first filed up to two decades earlier. The one visibly moving part is geography — China absorbs 4,967 families, second only to the United States.

Filing Trend 2020–2024

Annual applications and patent families across all 27 filing authorities.

Annual output is flat rather than trending: 2021, 2022 and 2023 sit at 6,381, 6,732 and 6,705 applications, a spread of under 6%. 2020 is the largest year at 7,756 applications, 2024 the smallest at 5,277 — but the 2024 figure is depressed by the publication lag and should not be read as a decline. The family series follows the same shape at roughly 40% of the volume, which is the direct consequence of Apple filing about 2.4 applications per invention.

* 2024: incomplete data due to the ~18-month publication delay (PATSTAT Spring 2026 edition, patstat_2026a). † Each family is counted once, in the earliest year it was filed inside the window, so the five yearly values sum exactly to the 13,934 total. Families whose first application predates 2020 and that were extended during the window therefore land in the 2020 bucket, which is one reason 2020 is the largest family year.

Data table: Filing Trend 2020–2024

Filing year	Applications	YoY change	Patent families †	Granted applications
2020	7,756	—	3,695	5,530
2021	6,381	−17.7%	2,666	3,797
2022	6,732	+5.5%	2,694	3,110
2023	6,705	−0.4%	2,768	2,235
2024	5,277	−21.3%*	2,111	746
Total	32,851	—	13,934	15,418

Technology Profile — WIPO Fields

Patent families per WIPO technology field (35-field schema). A family can belong to several fields, so the values are non-exclusive.

Two fields carry the portfolio in near-equal measure: digital communication (5,480 families, 39%) and computer technology (5,476, 39%). Adding audio-visual technology and telecommunications, four fields account for the bulk of the activity. The remaining weight is spread thinly and deliberately: optics (1,286), measurement (1,024), medical technology (398), control (346) and transport (268) each describe a product surface — cameras and displays, sensing, health, and vehicle interfaces.

Field values are non-exclusive: one family can be assigned to several WIPO fields, so the bars sum above the 13,934 total. Shares are calculated against that total.

Data table: WIPO Technology Fields

WIPO field	Patent families	Share of 13,934
Digital communication	5,480	39.3%
Computer technology	5,476	39.3%
Audio-visual technology	2,396	17.2%
Telecommunications	2,299	16.5%
Optics	1,286	9.2%
Measurement	1,024	7.3%
Electrical machinery, apparatus, energy	1,022	7.3%
Basic communication processes	445	3.2%
Semiconductors	419	3.0%
Medical technology	398	2.9%
Control	346	2.5%
IT methods for management	342	2.5%
Transport	268	1.9%
Other consumer goods	180	1.3%
Furniture, games	128	0.9%

Patent Classes in Detail

Top 15 IPC subclasses by patent family, read from the classifications of all family members. A family usually carries several subclasses, so the values are non-exclusive.

The two leading subclasses are separated by 40 families: G06F (4,448) for digital data processing and H04W (4,408) for wireless communication networks — each touching roughly a third of the portfolio. Below them the picture splits into a communications branch (H04L 2,927, H04B 1,452, H04M 631, H04R 503) and an imaging and display branch (G06T 1,482, H04N 1,200, G02B 1,062, G06V 643, G09G 515). Speech (G10L, 347) and computational models such as neural networks (G06N, 338) sit at the edge of the top 15 rather than at its centre.

Subclass values are non-exclusive: one family can carry several IPC subclasses, so the bars sum above the 13,934 total. Titles are the subclass titles of the IPC edition in force since 1 January 2026.

Data table: Top IPC Subclasses

IPC	Title	Families
`G06F`	Electric digital data processing	4,448
`H04W`	Wireless communication networks	4,408
`H04L`	Transmission of digital information	2,927
`G06T`	Image data processing or generation, in general	1,482
`H04B`	Transmission	1,452
`H04N`	Pictorial communication, e.g. television	1,200
`G02B`	Optical elements, systems or apparatus	1,062
`G06V`	Image or video recognition or understanding	643
`H04M`	Telephonic communication	631
`G09G`	Arrangements or circuits for control of indicating devices	515
`H04R`	Loudspeakers, microphones, gramophone pick-ups	503
`H05K`	Printed circuits; casings or constructional details of electric apparatus	371
`G10L`	Speech analysis techniques or speech synthesis; speech recognition	347
`G06Q`	Information and communication technology [ICT] specially adapted for specific fields	342
`G06N`	Computing arrangements based on specific computational models	338

Geographic Analysis

Where Apple files: distribution across offices, and how the mix moves year by year.

The United States carries 82% of all families (11,464), and no other office comes close. The genuinely notable position is second: China holds 4,967 families (36%), ahead of the PCT route (4,215) and the EPO (2,474). Year by year the US share of applications rises from 43% in 2020 to 49% in 2023, while EP filings fall from 764 to 499 and Chinese filings from 1,586 to 1,142 — but both of those authorities publish later than the USPTO, so part of that movement is publication lag rather than strategy. 2024 is left out of this comparison entirely: it is the least complete column, and the office that publishes fastest is necessarily over-represented in it.

* 2024: incomplete data due to the ~18-month publication delay (PATSTAT Spring 2026 edition, patstat_2026a); offices differ in how quickly they publish, so the 2024 mix is the least reliable column. Family shares per office sum above 100% because one family is normally filed at several offices; the yearly application columns sum exactly to the annual totals.

Data table: Filing Offices 2020–2024

Office	Patent families	Share of 13,934	Applications
US — United States	11,464	82.3%	15,455
CN — China	4,967	35.6%	5,919
WO — PCT	4,215	30.2%	4,237
EP — European Patent Office	2,474	17.8%	2,860
KR — Korea	1,217	8.7%	1,609
DE — Germany	636	4.6%	641
JP — Japan	488	3.5%	717
AU — Australia	357	2.6%	768
GB — United Kingdom	160	1.1%	198
TW — Taiwan	108	0.8%	155

Data table: Geographic Shift by Filing Year (applications)

Filing year	US	CN	WO (PCT)	EP	KR	Other offices	Total
2020	3,328	1,586	961	764	418	699	7,756
2021	2,921	1,250	738	662	330	480	6,381
2022	3,125	1,280	698	683	431	515	6,732
2023	3,316	1,142	997	499	233	518	6,705
2024*	2,765	661	843	252	197	559	5,277

Grant Status

Share of applications recorded as granted, by filing year and by office. Grant is measured at application level — the office that grants is the office the application was filed at.

Across the whole window 15,418 of 32,851 applications (46.9%) are recorded as granted. The year curve is a pendency curve, not a quality curve: 2020 filings stand at 71.3% and each later cohort is lower simply because fewer of its applications have been examined yet. The same caution applies across offices — the spread from 83.5% (AU) down to 5.8% (DE) mixes genuinely different examination practice with genuinely different processing speed, and cannot be read as a ranking of outcomes.

Grant rates for filings from 2022 onwards are artificially low due to examination pendency; the 2020 cohort is the only one close to complete. PCT (WO) applications are excluded from both views — an international application does not grant. Data basis: PATSTAT Spring 2026 edition (patstat_2026a).

Data table: Grant Status by Filing Year

Filing year	Applications	Granted	Granted share
2020	7,756	5,530	71.3%
2021	6,381	3,797	59.5%
2022	6,732	3,110	46.2%
2023	6,705	2,235	33.3%
2024	5,277	746	14.1%
Total	32,851	15,418	46.9%

Data table: Grant Status by Office

Office	Applications	Granted	Granted share
AU — Australia	768	641	83.5%
JP — Japan	717	548	76.4%
US — United States	15,455	10,295	66.6%
KR — Korea	1,609	774	48.1%
CN — China	5,919	1,562	26.4%
EP — European Patent Office	2,860	536	18.7%
DE — Germany	641	37	5.8%

Inventor Analysis

Distinct inventor names recorded on Apple applications, and the names appearing on the most families.

22,190 distinct inventor names appear on the 13,931 families that carry an inventor record. The annual count rises from 8,145 in 2020 to 10,223 in 2023 while annual application volume does not rise — the pool of names recorded on Apple filings widens over the window. The names at the top of the list cluster in cellular and wireless engineering, which is consistent with the H04W and H04L weight of the classification profile.

Data Quality Note — Inventor Names

Inventor names are normalised by stripping everything except Latin letters, so ZHANG DAWEI and Zhang, Dawei merge into one entry. Two consequences follow, and both inflate the top of the list: different people who share a romanised name are counted as one, and names recorded in CJK script collapse into a single empty bucket, which is excluded from the ranking below. The table therefore shows how often a name appears, not how prolific an individual is.

Names on the most families

Inventor name (as recorded)	Country	Families	Share of 13,934
ZHANG DAWEI	—	2,007	14.4%
ZENG WEI	—	1,416	10.2%
SUN HAITONG	CN	1,306	9.4%
HE HONG	CN	1,292	9.3%
YE, Chunxuan	CN	1,078	7.7%
CUI JIE	CN	871	6.3%
ZHANG YUSHU	CN	848	6.1%
YANG WEIDONG	CN	789	5.7%
Niu, Huaning	CN	761	5.5%
HU HAIJING	CN	736	5.3%
OTERI OGHENEKOME	—	702	5.0%
TANG YANG	—	672	4.8%

Country is the country code on the person record; a dash means PATSTAT carries none. The per-year family column below counts a family in every year it filed, so those five values sum above the 13,931 total; the distinct-name totals are deduplicated within each row. * 2024: incomplete data due to the ~18-month publication delay (PATSTAT Spring 2026 edition, patstat_2026a) — the 2024 column is not used for any statement in this section.

Data table: Distinct Inventor Names per Filing Year

Filing year	Distinct inventor names	Families with an inventor record
2020	8,145	3,694
2021	8,335	3,254
2022	9,317	3,787
2023	10,223	4,402
2024	9,331	3,710
Whole window	22,190	13,931

Collaboration & Acquisitions

Applications on which someone other than Apple is also named as applicant.

Co-applicants are read at application level, not at family level: a family-level join would pair Apple with its own name variants and return Apple as its own largest partner. All Apple spellings — Latin, mis-typed and Chinese or Japanese script — are excluded from the partner list. On that basis 1,073 of 13,934 families (7.7%) carry an applicant other than Apple. Almost all of those are the inventors themselves: under United States practice an inventor is frequently recorded in the applicant role as well. Restricting the list to applicants that are not also named as inventor on the same application leaves 28 families (0.2%) — and several of those 28 are still individuals rather than organisations.

Co-applicants that are not also inventors

Co-applicant	Type	Country	Shared families
Sterling Labs LLC	Company	US	3
Dathomir Laboratories LLC	Company	US	2
Xnor AI Inc	Company	US	2
OwnSurround Ltd	Company	FI	2
Intel IP Corp.	Company	US	1
Mammou, Khaled	Individual	CA	2
Flynn, David	Individual	DE	2
Robinet, Fabrice A.	Individual	FR	2

Shown: every organisation on this list, plus the individual applicants that appear on two families each. Counts are so small that no share or trend can be derived from them.

Apple's portfolio in this window is effectively unshared. With 28 of 13,934 families carrying an applicant that is not an Apple inventor, there is no collaboration network to map — no university programme, no supplier consortium and no standards joint venture appears in the applicant field. The organisations that do appear are small US and Finnish technology firms, on one to three families each; one possible explanation is a filing pipeline inherited through acquisition, but the data shows the co-filing, not its cause. Patents held by acquired companies under their own names are outside this scope entirely.

Foundational Patents Kept Alive

Apple families with a filing inside 2020–2024 that are cited by the largest number of other patent families.

Every one of the twelve most-cited families has an earliest filing before the window — between 2002 and 2014 — yet each carries at least one application filed in 2020 or later. The top entry, a multipoint touch screen family first filed in 2004, is cited by 3,493 other DOCDB families. The list reads as a map of the platform Apple built between 2004 and 2007: multi-touch, the hand-held device, unlocking by gesture, image sharing, and the intelligent automated assistant. Extending these families through continuations is a different activity from filing new inventions, and it is invisible in a count of new families.

Citation counts are the family-level counter nb_citing_docdb_fam and are cumulative over the whole life of the family — older families have had longer to accumulate citations. Titles are the English title recorded on a family member.

Data table: Most-Cited Active Families

DOCDB family	Representative title	Earliest filing	Citing families
`35159775`	Multipoint touch screen	2004	3,493
`39092692`	Device, method, and medium for sharing images	2006	2,565
`44304930`	Active input elicitation by intelligent automated assistant	2006	2,536
`80588058`	Multi-functional hand-held device	2005	2,355
`38008079`	Unlocking a device by performing gestures	2005	2,132
`38326964`	Multipoint touch surface controller	2006	1,433
`46124117`	Electronic device having display and surrounding touch sensor	2002	846
`39595357`	Device, method and graphical user interface for zooming in	2007	760
`44760640`	Compositing pairs of image frames from different cameras	2010	722
`54932017`	Intelligent automated assistant for TV user interactions	2014	688
`40429680`	Authentication system embedded in electronic device	2007	675
`50193584`	Voice trigger for a digital assistant	2013	673

Methodology

Search strategy, data source, and known limitations

Data basis. The figures in this report were computed on 15 August 2026 against EPO PATSTAT Global, Spring 2026 (dataset patstat-mtc.patstat_2026a). Classification labels come from the IPC edition in force since 1 January 2026 (edition 20260101) and, where CPC is referenced, the CPC edition 202608 in force since 1 August 2026. Two portfolios read from different editions are not comparable, which is why the editions are named here rather than assumed.

Search strategy, counting rules and data source — in full

Search strategy — applicant harmonisation

The scope of a company profile comes from the applicant name list, not from a classification code set. It was built in four steps:

Variant discovery. The 120 most frequent person records containing the word Apple were listed from tls206_person and reviewed by hand.
Variant review. Of the 26 name-and-country rows with at least one filing in 2020–2024, 16 are corporate Apple spellings and were kept (APPLE INC., APPLE INC, APPLE, INC., APPLE. INC., APPLE INC ., APPLE LNC., APPLE INE., APPLE COMPANY and their country variants). 10 rows were excluded as unrelated organisations that merely carry the word: a flavour and fragrance group, a faucet manufacturer, a chemicals company, a clothing brand, a design studio, two Chinese apple-growing bodies and three small Asian companies. Individuals surnamed Apple were excluded as well. Historic spellings (APPLE COMPUTER, INC. and its variants) were carried in the filter for completeness; none has a filing inside the window.
Non-Latin renderings. A keyword or prefix search in Latin script cannot reach a name written in another script. Four renderings were found and verified by family overlap with the Latin set: 苹果公司 (CN, 1,505 families, 1,448 shared), 美商蘋果公司 and 美商蘋果公司 (美國) (TW, 57 and 18 families, all shared) and the Japanese アップルインコーポレイテッド (51 families, 40 shared). Together they add 68 families that no Latin spelling reaches.
Negative test. A search for near-miss spellings returned four mis-typed forms — APPL INC., APPLI INC., AOOLE INC. and APLLE INC — each on a single family, and every one of those families is already inside the scope. One further spelling, APPLIE INC. (1 family), shares no family with the confirmed set and was left out.

Was han_name usable? Only as a cross-check, not as the scope key. Read in the direction "which names sit under this harmonised key", han_name = 'APPLE INC' is clean: every person record under it is an Apple spelling, and no foreign organisation is mixed in. Read in the other direction it fails — the single spelling APPLE INC. is harmonised to three different keys (APPLE INC, APPLE INC. and Apple Inc.), so a filter on one key is incomplete. Measured directly: a scope built on han_name = 'APPLE INC' alone returns 13,404 families and misses 530 of the 13,934 in this report (3.8%), while contributing nothing the name-based scope does not already contain. The scope is therefore defined on person_name.

Family counting

All counts use DOCDB patent families (docdb_family_id) so that the same invention filed in several countries is counted once.
Consolidation evidence: adding the per-spelling family counts of the 16 Latin rows gives 18,986; counting them together with COUNT(DISTINCT docdb_family_id) gives 14,926. The 4,060-family gap is the cross-variant overlap that consolidation removes. Adding the four non-Latin renderings raises the deduplicated total to 14,994.
Restricting to patents of invention (ipr_type = 'PI') removes 3,235 design applications (1,061 families) and 43 utility-model applications (37 families). Because a family can contain both kinds, the family total falls by 1,060 rather than by the sum, leaving the 13,934 families this report is built on.
Applicant role only (applt_seq_nr > 0); inventor figures use invt_seq_nr > 0.
The annual family series assigns each family to the earliest year it was filed inside the window, so the five yearly values sum exactly to 13,934. Per-office and per-year application counts are separate series and are labelled as such.
Co-applicants are joined on appln_id, never on docdb_family_id: a family-level join pairs a company with its own name variants and returns it as its own largest partner.

Multi-assignment note

A patent family can carry several IPC subclasses and belong to several WIPO technology fields. The technology and IPC charts are therefore non-exclusive and their bars sum above the 13,934 family total; the same applies to the per-office family counts, because one family is normally filed at several offices. This is expected, not an error — but it means those charts must not be read as parts of a whole, which is also why none of them is drawn as a pie or doughnut.

Classification scope

This report's patent set does not come from a classification code set, so the three code-set completion steps — subtree expansion, union of predecessor symbols and retention of deprecated symbols — do not apply and were not performed. IPC subclasses are used only to describe the portfolio that the applicant scope produced, and are read directly from tls209_appln_ipc across all family members without any code filter. The IPC and CPC editions used for the labels are named at the top of this section.

Data source

EPO PATSTAT Global, Spring 2026 edition (patstat-mtc.patstat_2026a) on Google BigQuery.
Classification views patstat-mtc.classification.dpt_ipc and dpt_edition.
All SQL for this report is in queries.sql; every figure shown is traceable to one of its eleven queries.

Scope Limitations

Design patents (3,235 applications, 1,061 families) and utility models (43 applications, 37 families) are excluded — patents of invention only.
Only applications on which an Apple spelling appears in the applicant role are included. Patents filed by subsidiaries or acquired companies under their own names are outside the scope.
One near-miss spelling, APPLIE INC. (1 family), was left out because it shares no family with the confirmed set.
2024 filings are visibly incomplete: the ~18-month publication delay of the Spring 2026 edition means later filings of that year are not yet in the data. No growth statement in this report is based on 2024.
Grant rates for filings from 2022 onwards are depressed by examination pendency and are not comparable with the 2020 cohort.
Inventor names are normalised to Latin letters: identically romanised names of different people merge, and names in CJK script collapse into a single bucket that is excluded from the ranking.
Citation counts are cumulative over a family's life, so older families are structurally advantaged in the most-cited list.

Glossary — patent terms explained

Patent family (DOCDB): A group of applications protecting the same invention in different countries. Counting families instead of applications avoids counting one invention several times.
Application vs. family: Apple filed 32,851 applications for 13,934 families in this window — about 2.4 filings per invention.
IPC: International Patent Classification, a hierarchical scheme for the technical content of a patent. A four-character code such as H04W is a subclass.
WIPO technology field: A 35-field schema that maps IPC codes onto broader technology areas such as "digital communication".
PCT / WO: An international application filed under the Patent Cooperation Treaty. It reserves the right to enter national phases later; it is never granted itself.
Applicant vs. inventor: The applicant owns the right applied for; the inventor is the natural person who made the invention. Under US practice the same person is frequently recorded in both roles.
han_name: PATSTAT's harmonised applicant name. It groups spellings, but the grouping is imperfect in both directions and must be verified before use.
ipr_type: The kind of right applied for: PI patent of invention, DP design patent, UM utility model.
Publication delay: An application is normally published 18 months after filing, so the most recent filing years are always incomplete in any PATSTAT edition.

Report files

This report consists of meta.json (metadata), cover.html, kpi-strip.html, eleven section files under sections/, and queries.sql with all SQL, the edition header and the scope definition. The rendering, listing and preview images are produced by the publishing application, not by the report folder.

Like what you see?

This report was built with a fully reproducible pipeline: EPO PATSTAT Global on BigQuery, a custom MCP server, and Claude AI for analysis and visualization.

Custom Patent Report Need a similar analysis for your technology area, company, or research question? We build tailored patent intelligence reports.

The Toolchain Want to run your own AI-powered patent analyses? We help you set up PATSTAT on BigQuery with MCP server integration.

Arne Krueger — LinkedIn Book a Call depa.tech mtc.berlin