Neuromorphic Computing Patent Landscape 2026
Neuromorphic Computing Patent Landscape in 2026
Neuromorphic computing is a concentrated, post-peak field: NVIDIA, IBM, and Samsung together hold commanding positions among 7,478 patent families in scope, and annual filing volume has eased from its 2021 high. The top five filers account for 39% of the hundred largest filers’ combined output, signaling that a small tier of technology companies and portfolio holders define the competitive frontier.
NVIDIA leads a tightly concentrated field with two close challengers
NVIDIA Corporation ranks first with 679 patent families, followed closely by IBM at 632 and Samsung Electronics at 564 — a near-three-way contest at the top before a steep drop to fourth-placed Qualcomm at 411.
The top five filers account for 39% of the hundred largest filers’ combined output, confirming that neuromorphic computing IP is concentrated in a short-list of large technology companies and specialized portfolio holders. A secondary tier — Intel, Micron, Innatera Nanosystems, and several universities — holds meaningful but substantially smaller positions.
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 1 | NVIDIA Corporation | 679 | |
| 2 | International Business Machines Corporation | 632 | |
| 3 | Samsung Electronics Co Ltd | 564 | |
| 4 | Qualcomm Inc | 411 | |
| 5 | Strong Force IoT Portfolio 2016 LLC | 261 | |
| 6 | Intel Corporation | 239 | |
| 7 | Strong Force TX Portfolio 2018 LLC | 178 | |
| 8 | Micron Technology Inc | 149 | |
| 9 | Innatera Nanosystems BV | 131 | |
| 10 | Zhejiang University | 118 |
| # | Applicant | Patent families | Share |
|---|---|---|---|
| 11 | HRL Laboratories LLC | 115 | |
| 12 | Brain Corp | 105 | |
| 13 | Lightelligence Pte Ltd | 95 | |
| 14 | Lynxi Technologies Co Ltd | 86 | |
| 15 | Tsinghua University | 76 | |
| 16 | Peking University | 73 | |
| 17 | Applied Brain Research Inc | 72 | |
| 18 | UNIV OF ELECTRONICS SCI & TECH OF CHINA | 71 | |
| 19 | Syntiant Corp | 70 | |
| 20 | GDM Holding LLC | 68 |
NVIDIA’s leadership reflects a broad AI-model and vision-processing portfolio, while IBM and Samsung have reinforced their positions with significant memory-device and semiconductor-device filings alongside neural-network software patents, suggesting vertically integrated strategies spanning chip architecture through algorithm.
The most recent 18–24 months of filing data are subject to publication lag and should be treated as undercounts; the apparent drop in 2024–2026 figures does not reflect a definitive activity level. Longer-window growth, applicant concentration, and technology-route coverage are therefore more reliable signals than the latest-year bar alone.
Filing activity peaked in 2021 and has eased; AI-model computing dominates the technology mix
The annual trend chart and IPC technology breakdown together show a field that expanded sharply from 2017 to 2021 and has since pulled back, while remaining overwhelmingly concentrated in AI-model computing methods (G06N).
Annual filing trend
Filings rose steeply from 340 families in 2017 to a peak of 1,134 in 2021, then eased to 910 in 2022, 829 in 2023, and 699 in 2024, with a partial 2025 figure of 842 reflecting incomplete publication. The 2026 count of 143 is heavily undercounted due to publication lag and should not be read as a meaningful data point. The multi-year window still shows substantial accumulated activity relative to the pre-2019 baseline.
↗ Hover for values · click a bar to ask EurekaTechnology composition
G06N (Computing based on AI models) dominates by a wide margin, reflecting the centrality of neural-network algorithm patents. G06F (Electric digital data processing) and G11C (Static and digital memories) are the next largest branches, with G06V (image/video recognition) and G06K (data recognition) adding application-layer depth. Hardware-oriented branches — H01L (Semiconductor devices) and H10N (Other electric solid-state devices) — are present but comparatively sparse, pointing to a field that is still more software- and algorithm-heavy than device-heavy at the patent level.
↗ 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.
Neuromorphic processor structure for layer unit ev…
Proposed are a neuromorphic processor structure for layer unit event routing of a spiking neural network, and a control method therefor. A layer unit event routing method of a spiking neural network, proposed in the present invention, comprises the steps of: optimizing a data structure for performing layer unit event routing by using a neuron address index… (excerpt from the patent abstract)
| # | Patent | Citations |
|---|---|---|
| 1 | Methods and systems for data collection, learning,… | 960 |
| 2 | Platform for facilitating development of intellige… | 720 |
| 3 | Intelligent vibration digital twin systems and met… | 690 |
| 4 | Systems and methods for crowdsourcing information … | 622 |
| 5 | Methods, systems, kits and apparatuses for monitor… | 546 |
| 6 | Methods and systems for data collection, learning,… | 506 |
| 7 | Methods and systems for detection in an industrial… | 406 |
| 8 | Systems and methods for policy automation for a da… | 363 |
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 patent structure means for R&D investment decisions
Four structural observations — maturity stage, concentration, collaboration patterns, and geographic spread — shape where new entrants and incumbents can most efficiently direct neuromorphic computing R&D.
Post-peak: field is maturing after a 2021 high
Annual filing volume has eased back from the 2021 peak of 1,134 families, placing neuromorphic computing in a declining phase of the IP cycle. This does not mean the technology is exhausted — the accumulated base of 7,478 families represents a dense prior-art landscape — but new entrants will find fewer under-patented foundational positions than were available in 2018–2020. R&D investment should focus on differentiated architectural or application-specific angles rather than broad platform filings.
Post-peak · 2021 highTop three players hold outsized positions; tier gap is real
NVIDIA (679), IBM (632), and Samsung (564) together represent the dominant tier, with Qualcomm (411) and Strong Force IoT Portfolio 2016 (261) completing the top five. Below rank five, counts drop rapidly into the sub-200 range, signaling a fragmented second tier. New entrants differentiating on specific memory-compute integration or edge-inference architectures may find more competitive headroom than in the general neural-network software space where the leaders are deeply entrenched.
High concentrationCorporate-academic co-filing is active, led by Samsung and IBM
Zhejiang University and Zhejiang Lab co-filed 34 patent families together, the most active pairing in the dataset. IBM co-filed 27 families with its UK subsidiary and 23 with its China entity, reflecting internal cross-jurisdictional coordination rather than external partnerships. Samsung Electronics co-filed with University of Zurich (17), POSTECH (17), Seoul National University R&DB Foundation (15), Sungkyunkwan University (13), and Harvard University (12), assembling a broad academic network spanning Europe, Korea, and the US — a strategy that diversifies research inputs across neuromorphic device physics and algorithm design.
Academic co-filing activeUS leads; China is a strong second; Europe and Korea are present but smaller
The United States accounts for the largest share of patent records, followed by China, with WIPO (PCT) and the EPO as the main international filing routes. India, the UK, South Korea, Germany, and Japan round out the active jurisdictions. The US–China duality reflects the two countries’ parallel investments in neuromorphic hardware and AI accelerator R&D, and organizations seeking broad protection should prioritize both offices alongside PCT or EPO filings for transatlantic coverage.
US & China dominantGo 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 |
|---|---|---|
| Zhejiang University | Zhejiang Lab | 34 |
| International Business Machines Corporation | IBM United Kingdom Ltd | 27 |
| International Business Machines Corporation | IBM China Co Ltd | 23 |
| Samsung Electronics Co Ltd | University of Zurich | 17 |
| Samsung Electronics Co Ltd | 浦项工科大学校产学协力团 | 17 |
| Samsung Electronics Co Ltd | Seoul National University R&DB Foundation | 15 |
| Samsung Electronics Co Ltd | Sungkyunkwan University Industry-Academic Cooperation Foundation | 13 |
| Samsung Electronics Co Ltd | President and Fellows of Harvard College | 12 |
| International Business Machines Corporation | IBM DEUTSCHLAND GMBH | 9 |
| International Business Machines Corporation | IBM ISRAEL SCI & TECH LTD | 8 |
Co-filing pairs, ranked by the number of jointly-filed patent families.
NVIDIA and IBM lead on volume; Samsung differentiates through memory and academic partnerships
The top three applicants share a common emphasis on AI-model computing (G06N 3) but diverge significantly in secondary technology focus, signaling different product and research strategies within neuromorphic computing.
NVIDIA Corporation
NVIDIA leads with 679 patent families, concentrated in AI-model computing (G06N 3: 677 filings), image data processing (G06T 7: 134), and data recognition (G06K 9: 106). This profile reflects NVIDIA’s GPU-centric approach to neural inference, extending into vision and perception workloads. Recent-period filings show a trend of -58% versus the prior three years, consistent with the broader field easing from its 2021 peak rather than a company-specific retreat.
families: 679IBM (International Business Machines Corporation)
IBM holds 632 patent families with a vertically integrated profile: AI-model computing (G06N 3: 481) is the primary focus, but significant secondary filings in static memories (G11C 13: 169) and semiconductor devices (H01L 45: 113) signal investment in the physical substrate of neuromorphic chips, not just algorithms. IBM’s recent-period trend of -70% is steep, though its large accumulated base and active international co-filing with IBM UK and IBM China maintain broad jurisdictional coverage. Qualcomm, ranked fourth at 411 families, is flagged as a new entrant in the recent window, suggesting emerging competitive pressure from the mobile-silicon side.
families: 632| Applicant | Recent (3 yrs) | Trend |
|---|---|---|
| NVIDIA Corporation | 189 | ▼ -58% |
| International Business Machines Corporation | 96 | ▼ -70% |
| Samsung Electronics Co Ltd | 151 | ▼ -28% |
| Qualcomm Inc | 10 | ▲ new entrant |
| Strong Force IoT Portfolio 2016 LLC | 32 | ▼ -71% |
| Intel Corporation | 17 | ▼ -29% |
| Strong Force TX Portfolio 2018 LLC | 31 | ▼ -78% |
| Micron Technology Inc | 22 | ▼ -83% |
Under-served adjacent branches worth monitoring in neuromorphic computing
Several IPC branches appear at lower relative share alongside the dominant G06N class, representing areas where the neuromorphic computing patent literature is comparatively sparse. These are observations of relative sparsity; technical and commercial viability must be assessed independently.
H01L · Semiconductor Devices
H01L (Semiconductor devices) accounts for 574 patent records — a 3% relative share within the technology composition — despite neuromorphic hardware being fundamentally dependent on novel device physics such as memristors, phase-change memory cells, and resistive switching elements. The sparsity relative to the algorithm-heavy G06N branch suggests that device-level neuromorphic IP remains under-developed compared to software methods. Organizations with materials science or advanced CMOS capabilities may find this branch offers a less crowded entry path, particularly in device structures that enable in-memory or near-memory computation.
Search this in Eureka →G06E · Optical Computing
G06E (Optical computing) appears with 111 patent records, placing it among the lower-share branches in the neuromorphic computing corpus. Photonic neuromorphic systems — which use light-based components to implement spiking or analog neural computations at high speed and low energy — represent a technically distinct route from purely electronic implementations. The combination of sparse existing IP and plausible energy-efficiency advantages makes G06E an adjacent branch worth monitoring, though meaningful progress depends on the availability of silicon-photonics fabrication infrastructure and integration expertise, which limits the realistic entry path to organizations with photonics or integrated optics competency.
Search this in Eureka →How leading applicants differ across technology routes
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 | G06K 9 · Data recognition & presentation |
|---|---|---|---|---|---|
| Strong Force IoT Portfolio 2016 LLC | Strong · 264 | Strong · 230 | Moderate · 68 | Strong · 259 | Strong · 216 |
| NVIDIA Corporation | Strong · 677 | Emerging · 38 | Emerging · 89 | Emerging · 46 | Emerging · 106 |
| Strong Force TX Portfolio 2018 LLC | Strong · 178 | Strong · 159 | Moderate · 44 | Strong · 124 | Moderate · 81 |
| International Business Machines Corporation | Strong · 481 | Emerging · 17 | Absent | Absent | Absent |
| Samsung Electronics Co Ltd | Strong · 399 | Emerging · 22 | Absent | Emerging · 10 | Emerging · 12 |
| Qualcomm Inc | Strong · 359 | Absent | Absent | Absent | Emerging · 12 |
| Intel Corporation | Strong · 209 | Absent | Absent | Absent | Emerging · 9 |
Frequently asked questions
The dataset covers 7,478 patent families in scope globally, spanning the period from 2017 through 2026, with the most recent 18–24 months subject to publication lag.
NVIDIA Corporation leads with 679 patent families, narrowly ahead of IBM at 632 and Samsung Electronics at 564. These three form a distinct top tier before a step-down to Qualcomm at 411.
Annual filing volume peaked in 2021 at 1,134 families, up from 340 in 2017. Filings have eased since then, reaching 910 in 2022, 829 in 2023, and 699 in 2024, though the 2025 and 2026 figures are undercounted due to publication lag.
The United States leads in patent records, followed by China, with WIPO (PCT) and the EPO as the primary international filing routes. India, the United Kingdom, South Korea, Germany, and Japan are also active jurisdictions.
G06N (Computing based on AI models) dominates by a substantial margin, reflecting the centrality of neural-network algorithm patents. G06F (Electric digital data processing) and G11C (Static and digital memories) are the next largest branches, followed by application-layer classes such as G06V (image/video recognition).
Yes. Zhejiang University (118 patent families) and Tsinghua University (76) appear in the top-20 ranked applicants, and the collaboration data shows Samsung Electronics maintaining active co-filing relationships with the University of Zurich, POSTECH, Seoul National University, Sungkyunkwan University, and Harvard University.
Ready to map your own neuromorphic computing 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.