Edge AI Inference Patent Landscape 2026
Edge AI Inference Patent Landscape in 2026
The Edge AI Inference space is in an active Growth phase, with annual filings rising sharply and a 98% increase over the recent window. Google LLC holds a commanding lead, but a wave of new entrants from Samsung, Qualcomm, and Rain Neuromorphics signals a rapidly intensifying competitive field.
Google leads a concentrated but increasingly contested field
Google LLC sits at the top of the top five filers together account for 49% of the combined total across the hundred largest filers, signalling meaningful concentration at the apex.
A clear tier gap exists between the top two incumbents — Google and Samsung — and the remaining ranked applicants. Vellore Institute of Technology at 30 patent families represents a notable academic cluster, but falls well short of the commercial leaders. Beyond that, Qualcomm (16) and L’Oréal (13) form a thin second commercial tier before filings drop to single digits.
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 1 | Google LLC | 87 | |
| 2 | Samsung Electronics Co., Ltd. | 63 | |
| 3 | Vellore Institute of Technology | 30 | |
| 4 | Qualcomm Incorporated | 16 | |
| 5 | L’Oréal S.A. | 13 | |
| 6 | Intel Corporation | 9 | |
| 7 | Vellore Institute of Technology Chennai | 8 | |
| 8 | Vaibhav Laxman Dhasal (Director) | 8 | |
| 9 | Jio Platforms Ltd. | 7 | |
| 10 | Tata Consultancy Services Ltd. | 6 |
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 11 | Rain Neuromorphics Inc. | 6 | |
| 12 | SRM University | 5 | |
| 13 | Sage University | 5 | |
| 14 | Adobe Inc. | 5 | |
| 15 | SR University | 5 | |
| 16 | OpenAI OpCo LLC | 4 | |
| 17 | Easwari Engineering College | 4 | |
| 18 | DDAIM Inc. | 4 | |
| 19 | Vallurupalli Nageswara Rao Vignana Jyothi Institute of Engineering and Technology | 4 | |
| 20 | Lovely Professional University | 3 |
Google’s position, grounded in on-device model deployment and AI computing frameworks, reflects years of sustained investment. Samsung’s recent momentum (3.1× growth versus its prior three-year period) suggests an accelerating challenge to Google’s lead, particularly in neural-network and video-inference workloads.
The most recent 18–24 months of filing data are subject to publication lag and will be revised upward as applications are published; apparent activity levels for 2025–2026 should be treated as provisional lower bounds. Longer-window growth, applicant concentration, and technology-route coverage are therefore more reliable signals than the latest-year bar alone.
Sharp acceleration since 2022, dominated by AI-computing and data-processing branches
Annual filing volumes show a multi-year upward trajectory with a pronounced surge in 2025, while the technology composition confirms that AI model computing is the overwhelming structural backbone of the field, with adjacent application verticals still relatively sparse.
Annual filing trend
Filings grew from 14 in 2017 to a sustained plateau in the low-to-mid forties between 2022 and 2024, then jumped sharply to 215 in 2025. The 2025–2026 figures carry publication lag and will rise further; the underlying growth trend is real and ongoing.
↗ Hover for values · click a bar to ask EurekaTechnology composition
G06N (Computing based on AI models) dominates with 555 records, dwarfing all other branches. G06F (Electric digital data processing) is the second pillar at 189, followed by G06V (Image/video recognition) at 92 and G06T (Image data processing) and H04L (Digital information transmission) each at 74. Healthcare branches A61B and G16H appear at 65 and 62 respectively, indicating emerging vertical application but still modest relative to the core computing stack.
↗ 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.
Method and system for on-device inference in a dee…
The disclosure relates to method and system for on-device inference in a deep neural network (DNN). The method comprises: determining whether one or more layers of the DNN satisfy one of a first, a second and a third condition, the one or more layers including one or more convolution layers and one or more resampling layers; performing the on-device… (excerpt from the patent abstract)


| # | Patent | Citations |
|---|---|---|
| 1 | Application Development Platform and Software Deve… | 243 |
| 2 | Smart replies using an on-device model | 177 |
| 3 | Systems and methods for self-learning artificial i… | 113 |
| 4 | On-device neural networks for natural language und… | 108 |
| 5 | Smart replies using an on-device model | 87 |
| 6 | Application development platform and software deve… | 84 |
| 7 | On-Device Machine Learning Platform to Enable Shar… | 80 |
| 8 | Augmented Reality Microscope for Pathology | 55 |
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. Some patent titles may be shown in their original, non-English language where an accurate translation could not be guaranteed.
What the structure means for R&D strategy
The combination of an active Growth lifecycle, a concentrated top tier, and a broad spread of emerging verticals creates a landscape where first-mover advantage in core AI inference is largely established — but application-layer differentiation remains genuinely open.
Active Growth: annual filings still climbing
The lifecycle stage is confirmed as Growth, with annual filings still rising and the recent-window growth rate at 98%. The field has not yet reached a consolidation phase, which means claim scope and foundational positions are still being established. Entrants who delay may face a rapidly closing window for broad foundational patents.
Growth stageTop-heavy with a thin second commercial tier
The top five filers hold 49% of the combined total across the hundred largest filers, with Google and Samsung alone accounting for a large share of that. The second commercial tier — Qualcomm (16 patent families), L’Oréal (13), and Intel (9) — is thin, suggesting that challengers have not yet built defensive depth comparable to the leaders. New entrants such as Qualcomm, Rain Neuromorphics, and Jio Platforms are all recorded as new entrants in the recent period, indicating the competitive map is still forming below the top tier.
Concentrated apexMinimal co-filing activity; ecosystem largely siloed
Only one co-filing relationship is recorded in evidence: Samsung Electronics and SNU R&DB Foundation (Seoul National University’s research and development foundation) have filed jointly once. This near-absence of collaboration suggests the field is still operating in a competitive rather than consortium mode, with most applicants building proprietary portfolios independently. Monitoring emerging academic-industry partnerships may provide early signals of technology transfer activity.
Low co-filingIndia leads filing volumes; US is the primary commercial jurisdiction
India accounts for 306 patent records, making it the lead filing office — driven largely by academic and institutional applicants such as Vellore Institute of Technology, SRM University, and numerous engineering colleges. The United States follows at 118 records, reflecting the commercial core where Google, Qualcomm, and Intel concentrate enforcement-grade protection. EPO (48) and WIPO PCT (46) coverage indicates that leading commercial players are pursuing multi-jurisdictional strategies, while China at only 5 records appears markedly underrepresented relative to its role in AI hardware manufacturing.
India-led, US-enforcedGo 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 |
|---|---|---|
| Samsung Electronics Co., Ltd. | SNU R&DB Foundation | 1 |
Co-filing pairs, ranked by the number of jointly-filed patent families.
Google stable at the top; Samsung accelerating fast
Two commercial players — Google LLC and Samsung Electronics — sit clearly above the rest of the ranked field, but with meaningfully different recent trajectories and technology emphases.
Google LLC
Google leads with 87 patent families and a stable recent trajectory (flat, 0% change versus its prior three-year period), suggesting a mature, well-defended portfolio rather than an aggressive current expansion. Its technology emphasis concentrates on G06N 20 (machine learning models, 57 records) and G06N 3 (neural networks, 36 records), complemented by G06F 9 (processing execution, 16 records). The most-cited patents in the corpus — covering on-device models for smart replies and natural language understanding — are associated with Google’s foundational on-device inference approach, reinforcing its structural position.
patent families: 87Samsung Electronics Co., Ltd.
Samsung ranks second with 63 patent families and is the fastest-growing incumbent, with recent filings running at 3.1× its prior three-year rate. Its portfolio is heavily weighted toward G06N 3 (neural networks, 61 records) and extends into H04N 19 (video/pictorial communication, 10 records), reflecting a strategic push into on-device video inference — a differentiated angle from Google’s language-and-search focus. The collaboration with SNU R&DB Foundation adds an academic research pipeline. At this trajectory, Samsung is the most credible near-term challenger to Google’s top position.
patent families: 63| Applicant | Recent (3 yrs) | Trend |
|---|---|---|
| Google LLC | 22 | ▬ 0% |
| Samsung Electronics Co., Ltd. | 44 | ▲ 3.1× vs prior 3-yr |
| Qualcomm Incorporated | 8 | ▲ new entrant |
| Rain Neuromorphics Inc. | 10 | ▲ new entrant |
| L’Oréal S.A. | 4 | ▲ new entrant |
| Intel Corporation | 4 | ▲ new entrant |
| Jio Platforms Ltd. | 7 | ▲ new entrant |
Under-served routes worth watching in Edge AI Inference
Several IPC branches adjacent to the dominant G06N core carry meaningful filing counts but remain at low share relative to the corpus, indicating areas where edge AI inference techniques are being applied but where patent density has not yet hardened into defensive thickets.
A61B / G16H · Medical diagnosis and healthcare informatics at the edge
A61B (Diagnosis and surgery) and G16H (Healthcare informatics) together account for 65 and 62 records respectively — modest shares of roughly 4% each — despite the substantial cited patent activity around ‘Augmented Reality Microscope for Pathology’ already appearing in the top-cited corpus. This suggests that on-device AI inference for point-of-care diagnostics, surgical guidance, and real-time patient monitoring is technically active but still under-patented relative to its apparent commercial potential. Applicants with edge hardware and clinical validation experience would find relatively open claim space here.
Search this in Eureka →H04L / H04W · Inference over digital and wireless transmission networks
H04L (Digital information transmission, 74 records) and H04W (Wireless communication networks, 18 records) together represent edge AI inference applied to network infrastructure — latency-sensitive decisions at base stations, intelligent routing, and distributed inference across wireless links. At roughly 5% and under 2% of records respectively, and with Qualcomm (a new entrant with wireless-inference overlap in its portfolio) only recently entering, this branch is technically plausible for 5G/6G edge deployments yet remains relatively sparse. It represents an adjacent observation rather than a validated commercial gap, pending further applicant entry signals.
Search this in Eureka →How leaders differ by technology route
Strength of each leader across the main technology routes.
| Player | G06N 3 · Computing based on AI models | G06N 20 · Computing based on AI models | G06V 10 · Image/video recognition | G06N 5 · Computing based on AI models | A61B 5 · Diagnosis & surgery |
|---|---|---|---|---|---|
| Google LLC | Strong · 36 | Strong · 57 | Absent | Moderate · 14 | Absent |
| Samsung Electronics Co., Ltd. | Strong · 61 | Emerging · 11 | Absent | Emerging · 7 | Emerging · 2 |
| Vellore Institute of Technology | Strong · 26 | Moderate · 10 | Moderate · 7 | Moderate · 7 | Emerging · 5 |
| L’Oréal S.A. | Strong · 8 | Moderate · 4 | Strong · 8 | Moderate · 2 | Absent |
| Qualcomm Incorporated | Strong · 15 | Absent | Absent | Absent | Absent |
| Vaibhav Laxman Dhasal (Director) | Strong · 7 | Strong · 4 | Absent | Absent | Absent |
| Vellore Institute of Technology Chennai | Absent | Strong · 4 | Strong · 4 | Moderate · 2 | Absent |
Frequently asked questions
The corpus in scope contains 563 patent families. This total spans filings from 2017 through 2026, with the most recent 18–24 months subject to publication lag and therefore likely to be revised upward.
Google LLC leads with 87 patent families, ahead of Samsung Electronics Co., Ltd. at 63 patent families. The top five filers collectively hold 49% of the combined total across the hundred largest filers.
India is the lead filing office with 306 patent records, driven substantially by academic and engineering-college applicants. The United States follows at 118 records and represents the primary jurisdiction for commercially enforced protection.
The field is assessed as being in a Growth stage. Annual filings are still rising, and the recent-window growth rate stands at 98%. The surge to 215 records in 2025 — noting publication lag — reinforces that activity is accelerating rather than plateauing.
G06N (Computing based on AI models) is overwhelmingly dominant at 555 records. G06F (Electric digital data processing) is the second pillar at 189 records, followed by G06V (Image/video recognition) at 92, and G06T (Image data processing) and H04L (Digital information transmission) each at 74 records.
Co-filing activity is minimal. The only recorded collaboration in evidence is between Samsung Electronics and SNU R&DB Foundation (one joint filing). The field is operating predominantly in competitive rather than consortium mode, with applicants building independent proprietary portfolios.
Ready to map your own Edge AI Inference 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.