Neuromorphic hardware in 2026: overview
Neuromorphic hardware uses specialized circuits and devices to process information in ways inspired by neural systems. Spiking neural networks communicate through discrete events, while neuromorphic architectures can combine memory and computation to reduce data movement. This makes them relevant to edge systems with tight power or latency constraints, but efficiency depends on the workload, device behavior and software mapping. The field spans digital event-driven processors, analog and mixed-signal circuits, and emerging in-memory and photonic devices.
Recent work is tackling the gap between promising architectures and deployable systems. In October 2026, Zhejiang University researchers introduced ConSwap for congestion-aware SNN mapping and FlowMap for scheduling concurrent networks. Also in October, a University of Southern California study examined moisture effects on diffusive memristors, highlighting device stability as a design concern. In March 2026, a reported photonic two-chip system used light for real-time spiking reinforcement learning. Together, these directions show growing attention to hardware-software co-design, scaling and reliable operation, alongside new materials and computing substrates.
- 1Mapping matters: Zhejiang University’s October 2026 work targets congestion and concurrent workloads, showing that routing and scheduling can constrain neuromorphic systems as much as compute elements.
- 2Devices face scrutiny: Research on moisture effects in diffusive memristors points to stability and environmental sensitivity as practical hurdles for device-based neuromorphic computing.
- 3Research outpaces patents: Publications grew 23.7% while patent activity fell 34%, suggesting research momentum is not yet translating into the same pace of recent patenting.
Top organizations working on neuromorphic hardware
Ranked by Linknovate activity score · last 24 monthsThe activity score adds up the relevance of every patent, publication, grant and news item linking an organization to neuromorphic hardware. Showing 7 of 2,118 organizations.
| # | Organization | Approach | Signals (24 mo) | Momentum | Why it matters |
|---|---|---|---|---|---|
| 1 | 
Intel Corporation
Corporate · United States |
Digital event-driven processors | PAT 3 · PUB 17 GRANTS 1 · NEWS 336 |
-46% | Intel develops digital neuromorphic processors, including the Loihi family, for research into energy-efficient spiking neural networks and event-driven computing. |
| 2 | 
Chinese Academy of Sciences CAS
Research · China |
Digital event-driven processors | PUB 281 · NEWS 30 | +73% | CAS research spans spiking-neural-network processors with on-chip learning, visual neuromorphic architectures, and optoelectronic synapse devices. |
| 3 | 
IBM
Corporate · United States |
Digital event-driven processors | PAT 2 · PUB 5 NEWS 265 |
-42% | IBM has developed digital neuromorphic hardware, including its TrueNorth research chip, which uses neuron and synapse circuits for brain-inspired computation. |
| 4 | 
Zhejiang University
Research · China |
Digital event-driven processors | PAT 4 · PUB 184 NEWS 6 |
-10% | Zhejiang University researchers develop neuromorphic processor architectures and methods for mapping and scheduling spiking neural networks across multicore systems. |
| 5 | 
University of Chinese Academy of Sciences
Research · China |
Digital event-driven processors | PUB 207 · NEWS 9 | +18% | UCAS research includes RISC-V neuromorphic processing elements for spiking networks and low-power routing for network-on-chip hardware. |
| 6 | 
Tsinghua University
Research · China |
Photonic neuromorphic systems | PAT 6 · PUB 148 NEWS 22 |
-13% | Tsinghua researchers work on programmable event-driven neuromorphic processors and photonic neural hardware, including a reported processor efficiency above 100 GOPS/W/mm². |
| 7 | 
Fudan University
Research · China |
Memristive and in-memory devices | PAT 3 · PUB 168 NEWS 4 |
+1% | Fudan research includes CMOS-compatible hafnium-zirconium oxide memristive devices that combine artificial neuron and synapse functions for hybrid neuromorphic computing. |
| 8 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 9 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
| 10 | ··Hidden organizationCompany | Approach | PAT 00 · PUB 00 GRANTS 0 · NEWS 00 | +00% | Available to registered users. |
Startups to watch and leading research centers
Startups to watch 30 total
Applied Brain ResearchCanadaApplied Brain Research develops neuroscience-inspired AI and low-power on-device computing, including neural models designed for real-time edge applications.
Innatera NanosystemsNetherlandsInnatera develops low-power spiking-neural-network processors that process events locally for real-time edge applications.
SynSenseChinaSynSense develops neuromorphic processors and integrated sensing-computing systems for low-power, low-latency edge applications.- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
- ··Hidden startupCountryAvailable to registered users, with funding and founding year.
Leading research institutions 1,824 total
- Chinese Academy of Sciences CASChinaCAS research spans spiking-neural-network processors with on-chip learning, visual neuromorphic architectures, and optoelectronic synapse devices.
- Zhejiang UniversityChinaZhejiang University researchers develop neuromorphic processor architectures and methods for mapping and scheduling spiking neural networks across multicore systems.
- University of Chinese Academy of SciencesChinaUCAS research includes RISC-V neuromorphic processing elements for spiking networks and low-power routing for network-on-chip hardware.
- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
- ··Hidden research institutionCountryAvailable to registered users, with their key researchers.
Where neuromorphic hardware innovation happens
Share of documents with at least one organization from each country.
Open the interactive map with all 2,118 organizationsMain technical approaches
Last 24 months| Approach | What it is | Documents | Leading organizations |
|---|---|---|---|
| Digital event-driven processors | Digital processors route discrete spikes between neurosynaptic cores, offering programmable execution while requiring careful mapping and communication management. | 1,392 |
Intel Corporation, IBM, SNN Inc. |
| Analog and mixed-signal circuits | Analog and mixed-signal circuits implement neural dynamics directly in hardware, trading compact, low-power operation for sensitivity to variation and calibration. | 309 |
University of Zürich, Intel Corporation, Cc (Creative Conspiracy) |
| Memristive and in-memory devices | Memristive devices combine storage and computation in the same elements, reducing data movement while facing challenges in variability, endurance and integration. | 5,795 |
Intel Corporation, Chinese Academy of Sciences CAS, IBM |
| Photonic neuromorphic systems | Photonic systems use light to perform neural computation, offering fast signal propagation while requiring practical integration with electronic control and memory. | 363 |
Tsinghua University, Intel Corporation, ANN INC. |
| Event-based sensory hardware | Event-based sensors report changes rather than full frames, reducing redundant data while requiring compatible algorithms and downstream processing. | 294 |
Intel Corporation, IBM, Cc (Creative Conspiracy) |
Emerging trends
Last 12 months vs previous 36 monthsTerms whose frequency in the last 12 months grew the most compared with the previous 36 months.
Funding and investment
Recent funding rounds
Grant data totals $743,158,791 across 392 awards since 2020; the listed funding rounds are not clearly attributable to neuromorphic hardware projects.
Patent landscape
Top patent holders · last 24 monthsRecent developments
Latest news, duplicates removed- Oct 2026Abstract Spiking neural networks (SNNs) are a promising computational paradigm for resource- and energy-constrained edge applications. However, their implementation on field-programmable gate arrays (FPGAs) remains a…Frontiers in Neuroscience | New and Recent Articles
- Sep 2026Right now, your brain is recognising the shapes of these letters, analysing their meaning, and holding a stream of thought together – all on about 20 watts of power. That’s roughly what it takes to run a lightbulb. Yet…English – The Conversation
- Sep 2026Abstract Event cameras provide fast, sparse visual measurements, but their output alone does not supply the persistent state required for closed-loop tracking. This study presents a hybrid event-driven…Frontiers in Neuroscience | New and Recent Articles
- Aug 2026Currently, AI and edge computing are hitting a wall when it comes to power budgets, latency, and cloud dependence, hindering real-world deployments. This issue is complicated by the need to integrate a great deal of…Home | Electronic Design
- Month 20XXHidden developmentAvailable to registered users with alerts enabled.
Frequently asked questions about neuromorphic hardware
What companies are working on neuromorphic hardware?
Intel Corporation and IBM are established corporate participants, while Innatera Nanosystems and SynSense are named startups. Intel develops the Loihi processor family for research into spiking networks and event-driven computing. IBM has developed the TrueNorth research chip. The companies span digital processor design and low-power edge computing, while research organizations also contribute heavily to the field.
Which neuromorphic hardware startups are the most promising in 2026?
Innatera Nanosystems, SynSense and Applied Brain Research are identified as startups working on neuromorphic or brain-inspired computing. Innatera and SynSense focus on low-power spiking processors and integrated sensing-computing for edge applications. Applied Brain Research develops neuroscience-inspired AI and on-device computing. The available facts do not provide comparative performance or commercial traction, so they do not establish a definitive ranking.
Which universities and research centers lead neuromorphic hardware research?
The Chinese Academy of Sciences, Zhejiang University and the University of Chinese Academy of Sciences are prominent research organizations in the available figures. Zhejiang University works on processor architectures and SNN mapping, while CAS research spans on-chip learning and visual neuromorphic systems. Tsinghua University and Fudan University also contribute, including work on photonic hardware and memristive devices.
How much funding has neuromorphic hardware received?
The figures record $743,158,791 in grants across 392 awards since 2020. This is the grant total associated with the topic in the available data, rather than a verified total of all public and private investment. Listed funding rounds are not clearly attributable to neuromorphic hardware projects, so they should not be added to the grant figure as topic-specific investment.
Who holds the most neuromorphic hardware patents?
Innatera Nanosystems leads the listed patent holders with 15 patents in the window. Seoul National University R&DB Foundation has 11, and Tata Consultancy Services Ltd. has 9. Tsinghua University and Micron Technology Inc. are also among the listed holders. These counts reflect the patent-holder data provided and do not establish the total size or legal status of each organization’s portfolio.
Is neuromorphic hardware patent activity growing?
No: patent activity fell 34% in the latest 12-month comparison, with 269 patents in the 24-month window. Publication activity moved in the opposite direction, rising 23.7% to 4,074 publications in the last year. This divergence suggests that technical research is expanding faster than recent patenting, though the figures alone do not explain the causes or indicate commercial readiness.
What are the latest developments in neuromorphic hardware?
Recent developments include Zhejiang University’s ConSwap and FlowMap work on SNN mapping and scheduling in October 2026, and a University of Southern California study of moisture effects on diffusive memristors that month. In March 2026, a reported photonic system performed real-time spiking reinforcement learning using light. These examples address system scaling, device behavior and alternative computing substrates.
What are the main approaches in neuromorphic hardware and how do they differ?
The main approaches are digital event-driven processors, analog and mixed-signal circuits, memristive and in-memory devices, photonic systems, and event-based sensory hardware. Digital designs offer programmable spike processing, while analog circuits implement dynamics directly. Memristors bring storage and computation together; photonics uses light for computation; and event-based sensors reduce redundant input. Each approach has different integration, reliability and software-mapping trade-offs.
Which countries lead neuromorphic hardware?
The United States leads the listed country data with 11,484 documents, followed by China with 9,088 and South Korea with 3,617. The United Kingdom and Germany are also significant contributors, with 3,287 and 3,180 documents respectively. These figures indicate broad research and innovation activity, but document totals do not directly measure product deployment or market leadership.
What are the applications of neuromorphic hardware?
The listed applications include edge AI and IoT, robotics and autonomous systems, machine vision and perception, industrial monitoring and control, and biomedical and neural applications. Edge AI and IoT has 1,216 associated documents, while machine vision has 506. These systems are suited to workloads that use sparse, time-varying signals or require local processing, though practical fit depends on the task and hardware.