Spiking Neural Processor Patents: Leaders & White Space 2026
Filing growth compares 2021 (54 records) with 2024 (68) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 718 records in scope (CR5), not by the ranked leaders only.
What the spiking neural processor patent record shows
Spiking neural processors translate biological neuron behaviour — discrete spikes rather than continuous activations — into silicon and firmware. The patent record for this niche, 718 records filed between 2015 and 2026, sits inside the broader neuromorphic computing field but is narrow enough that filing activity by a handful of assignees moves the whole picture. Publication lags filing by roughly 18 months, so the 2025-2026 counts in any trend line are still filling in and should not be read as a slowdown.
Filing concentrates around core inference architecture — neuron arrays, synaptic weight storage, spike-encoding schemes — with thinner but present activity in speech, radar and memory-adjacent classes. The receiving-office spread points to the US, Europe and India as the three jurisdictions carrying the bulk of filing strategy, with China and PCT filings still comparatively light for a compute architecture field.
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
Filing trend and technology composition
Two views of the same 718-record dataset: how filing activity has moved year over year, and how those records split across IPC subclasses.
Filing trend, 2017-2026
Filings ran from 54 in 2017 to a peak of 99 in 2022, then eased to 68 by 2024 — the last year the dataset treats as complete — before the visibly partial 2025-2026 counts. The +26% move from 2021 (54) to 2024 (68) reads as sustained interest rather than a spike-and-fade cycle.
Technology composition by IPC subclass
G06N (AI-model computing) touches 82.2% of the 718 records, confirming this is fundamentally an AI-architecture patent set. G06F (18.2%) and G06V (9.6%) trail well behind, and application-adjacent classes — speech (G10L, 4.2%), memory (G11C, 3.8%), radar/positioning (G01S, 3.3%), digital transmission (H04L, 3.1%) — each cover under one in ten records, since a record can carry more than one class these shares add to over 100%.
Shares are the percentage of the 718 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Neuromorphic Computing: Spiking Neural Processor Patent Landscape with Eureka
This page is one run against one query. Ask Eureka your own question about neuromorphic computing: spiking neural processor patent landscape and every answer comes back with the patent numbers behind it.
Try EurekaRepresentative and most-cited filings
Sequential neural machine for memory optimized inference (US20260044722A1)
A spiking neural processor built from neurons interconnected by synaptic elements, where a subset of neurons receives input signals and each neuron generates its own output signal. A storage unit captures selected neuron outputs, and augmented input circuits feed back into selected neurons through the same synaptic network, aimed at cutting memory overhead during inference.Filed by Innatera Nanosystems B.V., published 2026-02-12 — one of the most recent records in scope.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | WO2019028269A2 | Methods and systems for detection in an industrial internet of things data collection environment with large … | 392 |
| 2 | US20190115011A1 | Detecting keywords in audio using a spiking neural network | 132 |
| 3 | US20170229117A1 | Low power neuromorphic voice activation system and method | 86 |
| 4 | US20210049480A1 | Predictive maintenance of automotive battery | 75 |
| 5 | US20190042920A1 | Spiking neural network accelerator using external memory | 74 |
| 6 | US20160034812A1 | Long short-term memory using a spiking neural network | 73 |
| 7 | US20170024644A1 | Neural processor based accelerator system and method | 72 |
| 8 | US20140019392A1 | Intelligent modular robotic apparatus and methods | 72 |
| 9 | US20210049445A1 | Predictive maintenance of automotive tires | 57 |
| 10 | US20210049444A1 | Predictive maintenance of automotive engines | 56 |
Citation counts favour older filings that have had more time to accumulate citations inside the searched corpus; treat them as a signal of influence, not of current technical importance.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the numbers mean for filing strategy
Three findings a claims strategist should take from this dataset before deciding where to file or where to design around.
The front of the field is tight, the rest is a long tail
Five assignees hold 280 of the 718 records in scope (39.0%), and the top 10 combined reach 54.6% (392 records). Below that the ranking drops fast — the tenth-ranked filer holds 16 records against the leader's 74 — meaning most of the remaining 90 ranked companies are single-digit filers defending narrow positions rather than broad portfolios.
Interest is still building, not cooling
Filings rose from 54 in 2021 to 68 in 2024, the last year with a complete data window, after peaking at 99 in 2022. The 2025-2026 dip in the raw count is a publication-lag artefact, not a real drop in filing activity.
Core AI-architecture claims dominate; application classes are thin
G06N (AI-model computing) touches the overwhelming majority of records, while speech (G10L), memory (G11C), radar (G01S) and transmission (H04L) each sit under 5% of the 718 records. That gap is either unclaimed opportunity or unmet technical fit — the dataset alone cannot say which, but it flags where to look first.
Filing strategy still centres on the US, Europe and India
The United States receives 295 of the filings tracked, ahead of Europe (119) and India (95); WIPO/PCT filings sit at 80 and China at 39, with South Korea at 19. For a compute-architecture field this heavy an American and European skew suggests China-side filing has not yet caught up to the underlying R&D activity.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to neuromorphic computing: spiking neural processor patent landscape, with the prior art for and against each one.
| Assignee | Co-assignee | Shared families |
|---|---|---|
| Samsung Electronics Co., Ltd. (Korea) | University of Zurich | 9 |
| Samsung Electronics Co., Ltd. (Korea) | Sungkyunkwan University Foundation for Corporate Collaboration | 4 |
| Samsung Electronics Co., Ltd. (Korea) | SNU R&DB Foundation | 2 |
| International Business Machines Corporation | École Polytechnique Fédérale de Lausanne (EPFL) | 1 |
| International Business Machines Corporation | IBM (China) Co., Ltd. | 1 |
| International Business Machines Corporation | IBM DEUTSCHLAND GMBH | 1 |
Only 6 co-assignee pairs appear in the dataset, and joint filing is concentrated around one recurring partner combination — most assignees in this field file solo rather than through joint ventures.
Where to take this analysis
The dataset points to specific next steps depending on whether the goal is filing, freedom-to-operate, or partnership scouting.
Map the concentrated leaders in detail
With 39.0% of records held by five assignees, a claim-by-claim review of their portfolios will surface which architectural approaches are locked up and which are merely staked out broadly.
Explore assignee portfolios in EurekaCheck the under-claimed adjacent classes
Speech, radar and memory-adjacent IPC classes each cover under 10% of records despite sitting next to a dense core — worth a focused prior-art search before committing claim language there.
Run a white-space search in EurekaWatch the 2025-2026 filings as they resolve
Because publication lags filing by roughly 18 months, the most recent two years will keep filling in; re-checking the trend in six to twelve months will sharpen the momentum picture.
Track new filings in EurekaCommon questions on spiking neural processor patents
The ranked leader in this dataset holds 74 of the 718 records in scope, with the field dropping off quickly after the top five assignees, who together hold 39.0% of all records. Beyond the top ten (54.6% combined), the remaining ranked companies each hold comparatively few filings, which points to a long tail of single-digit entrants rather than a broad second tier of major players. Anyone doing competitive tracking should watch the top handful closely and treat the rest of the ranking as a landscape of smaller, more specific bets.
Filings grew from 54 in 2021 to 68 in 2024, a 26% increase over that span, after an earlier peak of 99 filings in 2022. The apparent decline in 2025-2026 counts is a publication-lag artefact — patent publication typically trails filing by around 18 months, so the newest years always look thinner than they will once fully published. Based on the complete years available, the trend through 2024 still points upward rather than down.
The dominant class is G06N (computing based on AI models), which touches 82.2% of the 718 records in scope, confirming this is primarily an AI-architecture patent field. Secondary classes include G06F (electric digital data processing, 18.2%) and G06V (image/video recognition, 9.6%), with application-specific classes like speech (G10L, 4.2%), static memory (G11C, 3.8%), radar/positioning (G01S, 3.3%) and digital transmission (H04L, 3.1%) each covering under one in ten records. Because a single record can carry multiple IPC classes, these percentages add to more than 100%.
The United States is the largest receiving office with 295 filings, followed by Europe (EPO) at 119 and India at 95. WIPO/PCT filings account for 80, with China at 39 and South Korea at 19. The relatively light China filing count, given how active AI hardware R&D is there generally, is worth checking against parallel non-patent research activity rather than assumed as low interest in the technology.
One of the most recent records in the dataset, US20260044722A1 filed by Innatera Nanosystems B.V. and published 2026-02-12, covers a spiking neural processor with neurons connected by synaptic elements, a storage unit that captures selected neuron outputs, and augmented input circuits that feed those outputs back into the network to reduce memory overhead during inference. It illustrates a broader pattern in the dataset: incremental architectural refinements to core neuron-synapse-storage designs rather than entirely new categories of claim. Anyone filing nearby should check this claim scope carefully given how recent and specific it is.
Research Neuromorphic Computing: Spiking Neural Processor Patent Landscape in depth with Eureka
Go past this page: query the whole neuromorphic computing: spiking neural processor patent landscape corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
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.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.