Retrieval-Augmented Generation Patent Landscape 2026
Retrieval-Augmented Generation Patent Landscape in 2026
Retrieval-Augmented Generation has entered a sharp growth phase, with the corpus reaching 11,873 patent families and annual filings accelerating dramatically from 2022 onward. Microsoft Technology Licensing leads by family count, but the field is fragmented enough that no single applicant dominates — the top five filers account for only 17% of the hundred largest filers’ combined total.
Microsoft leads a fragmented field with no dominant incumbent
Microsoft Technology Licensing LLC holds the top position with 198 patent families, followed closely by Alipay (Hangzhou) Information Technology Co., Ltd. at 119, Oracle International Corporation at 117, Salesforce Inc. at 110, and. China Mobile Communications Group Co., Ltd. at 106. The ranking reflects strong participation from both U. S. platform vendors and Chinese technology and telecoms players.
The top five filers’ 17% share of the hundred largest filers’ combined total signals a highly distributed competitive structure. No single applicant has established a commanding lead, and the gap between the first-ranked applicant and the tenth (Citibank, N. A. at 84 patent families) is narrower than is typical in more mature technology fields, suggesting the race remains genuinely open.
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 1 | Microsoft Technology Licensing LLC | 198 | |
| 2 | Alipay (Hangzhou) Information Technology Co., Ltd. | 119 | |
| 3 | Oracle International Corporation | 117 | |
| 4 | Salesforce Inc. | 110 | |
| 5 | China Mobile Communications Group Co., Ltd. | 106 | |
| 6 | NVIDIA Corporation | 101 | |
| 7 | Dell Products LP | 90 | |
| 8 | China Telecom Corporation Limited | 87 | |
| 9 | BEIJING BAIDU NETCOM SCI & TECH CO LTD | 85 | |
| 10 | Citibank, N.A. | 84 |
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 11 | Intuit Inc. | 76 | |
| 12 | Amazon Technologies Inc. | 71 | |
| 13 | Google LLC | 64 | |
| 14 | Tencent Technology (Shenzhen) Co., Ltd. | 64 | |
| 15 | Inspur Suzhou Intelligent Technology Co., Ltd. | 59 | |
| 16 | Lenovo (Beijing) Ltd. | 58 | |
| 17 | IBM Corporation | 57 | |
| 18 | Huawei Technologies Co., Ltd. | 55 | |
| 19 | Zhejiang University | 53 | |
| 20 | The Toronto-Dominion Bank | 53 |
The breadth of the top-20 — spanning cloud platforms, enterprise software, financial services, telecoms infrastructure, and academic research — implies that RAG is being claimed simultaneously as a foundational AI technique and as a domain-specific application layer. Entrants positioning in any of these verticals face overlapping prior art across at least three distinct applicant communities.
Filing counts for the most recent 18–24 months are systematically understated due to publication lag; apparent growth should be read as a lower bound, not a ceiling. Longer-window growth, applicant concentration, and technology-route coverage are therefore more reliable signals than the latest-year bar alone.
Filings accelerating sharply since 2022; digital data processing dominates the IPC mix
Two views of the corpus reveal a technology still in rapid expansion: annual filing volume inflected sharply upward beginning in 2022, and the IPC class distribution is heavily concentrated in general digital data processing and AI model classes, with a long tail of applied-domain branches.
Annual filing trend
Volume was negligible through 2021 (27 families in 2017, 73 in 2021), then jumped to 170 in 2022, 632 in 2023, and 4,235 in 2024. The 2025 figure of 5,632 represents the largest single-year count on record. The 2026 figure of 867 is a partial year affected by publication lag and should not be read as a slowdown. The lifecycle evidence confirms the field is in a Growth stage with annual filings still rising.
↗ Hover for values · click a bar to ask EurekaTechnology composition
G06F (Electric digital data processing) is the dominant IPC class, reflecting RAG’s core dependence on indexing, retrieval, and language processing infrastructure. G06N (Computing based on AI models) is the second-largest class, confirming that neural-model integration is a primary patent claim axis. G06Q (Business, commerce and administrative data processing), G06V (Image/video recognition), and H04L (Digital information transmission) are the next-largest classes but at substantially lower counts, indicating that enterprise and multimodal applications are filed but not yet as densely claimed as core infrastructure.
↗ Hover for values · click a bar to ask EurekaHighly cited patent families surfaced by the query
Citation-heavy patent families returned by the query. Use this section as citation context, not as a curated list of the most topic-specific patents.
System and method for use of in-memory data grid a…
In accordance with an embodiment, described herein are systems and methods for use of an in-memory data grid as a vector database, with linearly-scalable data ingestion, for use in generative artificial intelligence (AI), data visualization, or other applications that include the use of a large language model (LLM) or a retrieval-augmented generation (RAG)… (excerpt from the patent abstract)
| # | Patent | Citations |
|---|---|---|
| 1 | Automatically assigning medical codes using natura… | 656 |
| 2 | Method and apparatus for tracking vehicle location | 419 |
| 3 | Resolution independent vector display of internet … | 372 |
| 4 | Context vector generation and retrieval | 363 |
| 5 | Method and apparatus for tracking vehicle location | 340 |
| 6 | Composite symbolic and non-symbolic artificial int… | 321 |
| 7 | Method and apparatus for tracking vehicle location | 192 |
| 8 | Convergent Intelligence Fabric for Multi-Domain Or… | 169 |
Ranked by total forward citations. Citation counts favour older and broadly cited patent families, and broad or adjacent patents may appear when they match the search scope. Treat this section as citation context, not as a curated list of the most topic-specific patents.
What the competitive structure means for R&D strategy
United States defines both the opportunity and the risk landscape for new entrants.
Growth stage: filings accelerating, foundational claims still being filed
The lifecycle evidence classifies RAG as a Growth-stage technology, with annual filings still rising and recent years understated by publication lag. Foundational retrieval-index and model-integration claims are actively being filed by a broad set of applicants, meaning the window for staking core positions has not closed but is narrowing. R&D investment made now will face a more crowded prior-art environment within two to three years.
Growth · Active filing windowFragmented top tier: 17% top-5 share leaves room for challengers
The top five filers hold only 17% of the hundred largest filers’ combined total, a low concentration ratio for a technology this young. The first-to-tenth rank gap is narrow. This structure means freedom-to-operate analysis should cover at least the top 20 applicants, but also that a focused filer entering now can realistically build a competitive position without facing a single dominant blocker.
Low concentrationChina Telecom and China Mobile subsidiaries are the most active co-filers
The densest co-filing relationships are within Chinese telecoms groups. China Telecom Corp. Ltd. and its Technology Innovation Center co-filed 45 patent families — the largest single co-filing relationship in the corpus. China Mobile Communications Group Co., Ltd. maintains active co-filing relationships with at least seven subsidiaries and research units, with pairings ranging from 5 to 14 families each. These intra-group filings suggest coordinated, portfolio-building strategies rather than opportunistic single filings, and may indicate prior-art clusters that external applicants should map carefully.
Intra-group co-filing dominantChina is the primary filing jurisdiction; U.S. and PCT filings signal global protection intent
China accounts for the largest share of filings at the patent-record level, followed by the United States and WIPO PCT. India ranks fourth, suggesting that South Asian enterprise-software markets are being designated for protection. Europe (EPO) and Japan trail significantly. Applicants building a global portfolio should treat China and the United States as mandatory jurisdictions and evaluate PCT entry for cost-efficient multi-territory coverage.
China-first, U.S. secondGo beyond the landscape: Eureka’s TRIZ Solution agent breaks down an R&D problem and returns patented concept solutions, each with a technical approach and cited patent & literature evidence.
| Applicant | Collaborator | Co-filings |
|---|---|---|
| China Telecom Corporation Limited | China Telecom Corporation Limited Technology Innovation Center | 45 |
| China Mobile Communications Group Co., Ltd. | China Mobile Information Technology Co., Ltd. | 14 |
| China Mobile Communications Group Co., Ltd. | China Mobile Research Institute | 14 |
| China Mobile Communications Group Co., Ltd. | China Mobile Group Design Institute Co., Ltd. | 9 |
| China Mobile Communications Group Co., Ltd. | China Mobile Xiongan Information and Communication Technology Co., Ltd. | 7 |
| China Mobile Communications Group Co., Ltd. | China Mobile System Integration Co., Ltd. | 7 |
| China Mobile Communications Group Co., Ltd. | China Mobile Information Systems Integration Co., Ltd. | 7 |
| China Mobile Communications Group Co., Ltd. | China Mobile Information Technology Co., Ltd. (subsidiary) | 7 |
| China Mobile Communications Group Co., Ltd. | China Mobile IoT Co., Ltd. | 5 |
| China Mobile Communications Group Co., Ltd. | China Mobile (Hangzhou) Information Technology Co., Ltd. | 5 |
Co-filing pairs, ranked by the number of jointly-filed patent families.
Microsoft and Dell lead on volume; NVIDIA differentiates on language-processing focus
The top applicants entered RAG patent filing as new entrants in the recent window, confirming that most volume has been filed within the current growth phase. Technology emphasis diverges meaningfully across leaders, creating differentiated claim landscapes.
Microsoft Technology Licensing LLC
Microsoft leads the ranking with 198 patent families and shows a trajectory classified as a new entrant in the recent filing window, with 122 families filed in the recent period alone. Its technology emphasis concentrates on G06F 16 (digital data processing, information retrieval) at 120 filings, followed by G06F 40 (natural language processing) at 55 and G06N 3 (neural computing models) at 55. This tripartite focus — retrieval infrastructure, NLP, and neural models — maps directly to the core RAG architecture stack, suggesting Microsoft is pursuing broad foundational coverage rather than a single-application niche.
patent families: 198Dell Products LP
Dell Products LP ranks seventh with 90 patent families and enters the recent window as a new entrant with 85 families filed recently, indicating that virtually its entire RAG portfolio has been built in the current growth surge. Its technology emphasis is heavily weighted toward G06F 16 (information retrieval, 78 filings), with comparatively lighter coverage of G06F 40 (12) and G06N 3 (9). This profile suggests Dell is focusing on retrieval-layer infrastructure — consistent with enterprise server and storage use cases — rather than model-integration or NLP-generation claims.
patent families: 90| Applicant | Recent (3 yrs) | Trend |
|---|---|---|
| Microsoft Technology Licensing LLC | 122 | ▲ new entrant |
| Alipay (Hangzhou) Information Technology Co., Ltd. | 27 | ▲ new entrant |
| Oracle International Corporation | 47 | ▲ new entrant |
| Dell Products LP | 51 | ▲ new entrant |
| China Mobile Communications Group Co., Ltd. | 35 | ▲ new entrant |
| NVIDIA Corporation | 58 | ▲ new entrant |
| Dell Products LP | 85 | ▲ new entrant |
| China Telecom Corporation Limited | 28 | ▲ new entrant |
Under-served branches in healthcare informatics and speech processing
Several IPC branches adjacent to the dominant G06F/G06N core show patent-record counts that are low relative to their technical plausibility as RAG application domains, representing areas where existing RAG infrastructure could be extended with comparatively less prior-art competition.
G16H · Healthcare informatics
G16H covers clinical decision support, patient record retrieval, and medical knowledge management — all natural application domains for RAG systems that retrieve from structured clinical corpora. With 400 patent records, this branch is notably sparse relative to the broader corpus and relative to the demonstrated demand for evidence-based clinical AI. Entry paths include retrieval over electronic health record databases, drug-interaction knowledge bases, and clinical-trial registries. Applicants with existing G06F 16 or G06N 3 portfolios would have a plausible technical bridge into this branch.
Search this in Eureka →G10L · Speech and audio analysis/synthesis
G10L covers automatic speech recognition, speaker verification, and audio synthesis — components increasingly integrated with RAG pipelines in voice-assistant and call-centre automation contexts. With 353 patent records, this branch is sparse given the commercial momentum of voice-enabled LLM applications. The entry path is technically direct: spoken-query transcription feeds a retrieval step, and synthesised speech delivers the grounded response. Applicants in enterprise communications or contact-centre infrastructure are best positioned to extend existing RAG claims into this branch.
Search this in Eureka →How leaders differ by IPC technology route
Strength of each leader across the main technology routes.
| Player | G06F 16 · Electric digital data processing | G06F 40 · Electric digital data processing | G06N 3 · Computing based on AI models | G06N 5 · Computing based on AI models | G06F 18 · Electric digital data processing |
|---|---|---|---|---|---|
| Alipay (Hangzhou) Information Technology Co., Ltd. | Strong · 92 | Moderate · 46 | Strong · 57 | Strong · 76 | Moderate · 26 |
| China Mobile Communications Group Co., Ltd. | Strong · 92 | Moderate · 41 | Moderate · 42 | Strong · 52 | Moderate · 21 |
| Microsoft Technology Licensing LLC | Strong · 120 | Moderate · 55 | Moderate · 55 | Absent | Absent |
| China Telecom Corporation Limited | Strong · 74 | Strong · 41 | Moderate · 27 | Strong · 49 | Moderate · 16 |
| Dell Products LP | Strong · 73 | Strong · 62 | Moderate · 25 | Absent | Absent |
| MADISETTI VIJAY | Strong · 76 | Strong · 76 | Absent | Absent | Absent |
| Beijing Baidu Netcom Science & Technology Co., Ltd. | Strong · 72 | Absent | Moderate · 27 | Moderate · 34 | Absent |
Frequently asked questions
The corpus covers 11,873 patent families in scope. Annual filing volume remained below 200 families per year through 2021 before accelerating to 632 in 2023 and 4,235 in 2024, confirming that the technology has entered a rapid-growth phase.
Microsoft Technology Licensing LLC leads with 198 patent families, followed by Alipay (Hangzhou) Information Technology Co., Ltd. at 119 and Oracle International Corporation at 117. The top five filers account for 17% of the hundred largest filers’ combined total, indicating a fragmented rather than oligopolistic field.
China is the leading jurisdiction at the patent-record level, followed by the United States and WIPO PCT. India ranks fourth, ahead of Europe (EPO) and Japan, reflecting both the geographic distribution of active filers and their multi-territory protection strategies.
The evidence classifies RAG as a Growth-stage technology, with annual filings still rising. The most recent 18–24 months are understated due to publication lag, so current counts represent a lower bound on actual filing activity.
All eight applicants tracked in the momentum analysis are classified as new entrants in the recent filing window, meaning they have little or no prior-period history in the corpus. Dell Products LP shows the largest absolute recent count at 85 patent families in the recent period, followed by NVIDIA Corp at 58 and Microsoft Technology Licensing LLC at 122.
Healthcare informatics (G16H, 400 patent records) and speech and audio analysis (G10L, 353 patent records) are the sparsest branches with clear technical relevance to RAG pipelines. Business and commerce data processing (G06Q, 1,678 records) and image/video recognition (G06V, 737 records) are larger but still substantially smaller than the dominant G06F class, suggesting room for domain-specific RAG claim development.
Ready to map your own Retrieval-Augmented Generation landscape?
Join 18,000+ innovators using PatSnap Eureka to map any technology landscape: search 2B+ patents and papers, surface key assignees, and generate a report like this in minutes.
Disclaimer. This page is generated from PatSnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. PatSnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.