What Applied Brain Research does

Applied Brain Research (ABR) is a Waterloo, Ontario-based private company that invented and commercialized State Space Model (SSM) architectures for deep learning, beginning with the Legendre Memory Unit (LMU) introduced at NeurIPS 2019. The company builds on-device voice AI — specifically automatic speech recognition (Niagara), text-to-speech (Nith), speech-to-intent, language translation, and biosensing — using SSM-dominant neural network architectures that achieve linear O(N) computational scaling versus O(N²) for transformers, enabling compact models (5M–38M parameters) to run entirely on edge devices with sub-30mW power consumption and ~150ms first-token latency. ABR's product portfolio includes a Python-based SDK for integrating these models on Linux x86-64, Linux ARM64, and Android ARM64 platforms, plus the TSP1 silicon — the world's first dedicated SSM accelerator chip — which is in pre-production with full manufacturing targeted for Q2 2026.

The company monetizes through a software-IP licensing and royalty model, charging paid non-recurring engineering fees for proof-of-concept engagements with OEMs, Tier-1 suppliers, and device manufacturers, then converting successful POCs into per-device licensing and royalty agreements. Its neural network IP is deployed as embedded software on third-party edge processors from six named silicon partners: Nvidia, NXP, Qualcomm, Alif Semiconductor, STMicroelectronics, and Renesas. ABR targets use cases in augmented reality glasses, wearable devices, robotics, smart home appliances, medical devices, and automotive systems — all categories where cloud-dependent voice AI is constrained by latency, privacy, power, and per-interaction cost. The company holds 11 US patents, 5 Canadian patents, and additional granted patents in China, Europe, and Japan covering SSM implementations, and has raised over $3.6M USD across multiple seed rounds and government funding, with an oversubscribed seed round closed in January 2026 led by Two Small Fish Ventures. ABR is preparing for a Series A fundraise as the TSP1 chip moves into volume production.

Name

Applied Brain Research

Legal name

Applied Brain Research Inc.

Website

https://appliedbrainresearch.com

Company type

Private

Founded year

2012

Operating status

Operating

Headcount range

11–50 employees

Short description

Applied Brain Research invents and licenses State Space Model (SSM) neural network IP for on-device voice AI — ASR, TTS, and biosensing — to OEMs and device manufacturers, paired with the TSP1 SSM accelerator chip, targeting AR, wearables, robotics, and edge products.

Ownership category

akta.pro rank

Product category

Edge AI / Neural Network IP

NAICS

Semiconductor and Other Electronic Component Manufacturing (3344), Software Publishers (513210)

SIC

Electronic Computers (3571), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

Edge AI Hardware & NPUs (SoCs, modules, accelerators) (HDAAAJAC)

akta.pro secondary industries

AI/Neural Processing Units (NPUs) & ML Accelerators (HDAHAJAF), On-Device Inference Runtimes & SDKs (mobile/embedded) (HDAAAJAB), AI Hardware IP & Chiplet/Packaging for Accelerators (EDA/IP, advanced packaging) (HDAAAAAL)

Keywords

• Edge AI software
• Neural network IP licensing
• On-device speech recognition
• State space models
• Voice AI accelerator

Headquarters

HQ city

Waterloo

HQ country

Canada

HQ region

North America

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Personnel, Technology or R&D, Operations, Marketing or Sales, Infrastructure

Revenue model

1. Software Licensing and Royalty Revenue: ABR licenses its neural network IP (ASR and TTS models) to OEMs, device manufacturers, and silicon vendors. Models are deployed as embedded software running on third-party edge processors. Revenue is generated through NRE (Non-Recurring Engineering) payments for proof-of-concept engagements, IP licensing fees, and per-device royalty payments as products ship. The company sells best-in-class neural network IP, not hardware.
2. Proof-of-Concept (POC) Engagements: Paid POC and pilot engagements with OEMs and device manufacturers that demonstrate ABR's performance advantages on customers' target hardware and use cases. POCs are structured, negotiated engagements with defined success criteria, tracked milestones, and go/no-go outcomes that lead to licensing and royalty commitments.

Go-to-market motion3 records

Distribution channels2 records

Marketing channels6 records

Core offering

Applied Brain Research (ABR) develops and licenses State Space Model (SSM) based neural network IP for real-time on-device voice AI. The company sells ASR (Niagara), TTS (Nith), and related SSM models as embedded software for deployment on third-party edge processors, complemented by the TSP1 silicon accelerator. Revenue is generated through NRE payments, IP licensing fees, and per-device royalties to OEMs, device manufacturers, and silicon vendors.

Product overview

Applied Brain Research offers a portfolio of edge AI products centered on State Space Model (SSM) technology for low-power, on-device voice intelligence. The core offering consists of Niagara automatic speech recognition and Nith text-to-speech models, both delivered via the ABR SDK as application packages that run entirely on-device without cloud dependency. These are complemented by the TSP1 hardware accelerator chip—the world's first state space model silicon—which enables sub-30mW operation for always-on voice applications. The company operates a software licensing and royalty revenue model, with models deployed as embedded IP on third-party edge processors. Additional SSM capabilities include speech-to-intent, language translation, and biosensing.

Differentiator

Problem solved

Functional benefit

Brands

• Niagara: Automatic Speech Recognition (ASR) model family named after the river near ABR's hometown of Waterloo, Ontario.
• Nith
• TSP1
• ABR SDK

Products and services

• Niagara Automatic Speech Recognition Real-time on-device speech-to-text transcription using State Space Model architecture. Delivers streaming transcription with ~150ms first-token latency, low power consumption, and accuracy competitive with much larger transformer models. Available in multiple sizes (38M, 19M, 10M parameters) and supports custom vocabulary, capitalization, punctuation, and multiple languages.
• Nith Text-to-Speech On-device text-to-speech synthesis using SSM layers. Low-latency vocalization from text buffer input with support for prosody and emotion control via SSML markup. Available in English, Spanish, and Chinese (Mandarin) with 5M parameter models.
• TSP1 (Time Series Processor) World's first state space model accelerator chip. Delivers streaming voice AI at under 30mW power consumption, enabling all-day operation on a single battery charge for edge devices including AR glasses, wearables, robotics, and medical devices.
• ABR SDK Python library for on-device ASR and TTS providing high-level wrappers for automatic speech recognition and text-to-speech, with application packages for Linux x86-64, Linux ARM64, and Android ARM64. All inference runs locally with no internet required at runtime.
• Other State Space Models Additional SSM-based capabilities including speech-to-intent processing, language translation, and biosensing signal processing. Models designed for real-time, low-power operation on edge devices.

Quantifiable outcome

• 10–100x power reduction vs cloud voice AI alternatives — TSP1 operates at under 30mW
• +6 more outcomes

Segments6 records

Ideal customer profiles4 records

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

AI capability4 records

Feature6 records

Partnerships

Nine partnerships are on record, tiered partner, silicon partner and funding partner.

• VentureLabpartnerStrategic or Co-development PartnerVentureLab is listed as a partner of Applied Brain Research, supporting the company's growth and ecosystem development.
• CommunitechpartnerStrategic or Co-development PartnerCommunitech is listed as a partner of Applied Brain Research, supporting the Waterloo-based tech ecosystem. Communitech helps technology companies scale in the Kitchener-Waterloo region.
• Nvidiasilicon partnerTechnology or IntegrationNvidia is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on Nvidia edge processors.
• NXPsilicon partnerTechnology or IntegrationNXP is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on NXP edge processors.
• Qualcommsilicon partnerTechnology or IntegrationQualcomm is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on Qualcomm edge processors.
• Alif Semiconductorsilicon partnerTechnology or IntegrationAlif Semiconductor is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on Alif Semiconductor processors.
• STMicroelectronicssilicon partnerTechnology or IntegrationSTMicroelectronics is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on STMicroelectronics edge processors.
• Renesassilicon partnerTechnology or IntegrationRenesas is listed as one of the edge processor ecosystem partners whose platforms ABR's models are deployed on. ABR's software IP runs as embedded software on Renesas edge processors.
• Government of Canada / FedDev Ontariofunding partnerOthersABR acknowledges government funding from the Government of Canada through the Federal Economic Development Agency for Southern Ontario (FedDev Ontario), which funded the company's projects.

Scale indicators3 records

Recent moves6 records

Expansion highlights7 records

Broad incumbents

• Qualcomm: Mobile and IoT AI platforms (Hexagon NPU, Snapdragon) plus an embedded voice assistant stack. Functions as a silicon distribution partner for ABR while also being a broader incumbent whose voice+AI roadmap could expand into ABR's niche.
• NVIDIA: Jetson edge-AI platform that hosts ABR's models and is therefore both a distribution partner and a broader incumbent in on-device AI. Comparable through overlapping enterprise OEM customers and a much wider hardware/software portfolio.
• QuickLogic: Edge AI platforms combining FPGAs, AI-enabled SoCs, and sensor processing. Broader portfolio than ABR but competes in low-power on-device AI and serves similar industrial and edge markets.

Direct peers

• Hailo: Hailo-8 and Hailo-15 edge AI inference accelerators targeting low-power real-time AI on cameras and edge devices. Shares ABR's accelerator-chip business model and edge-device OEMs as customers, though focused on vision more than voice.
• BrainChip: Akida neuromorphic edge-AI processor targeting low-power on-device inference for voice and vision. Comparable to ABR's TSP1 in the always-on, low-mW position and in selling chip-plus-IP rather than raw silicon.
• XMOS: Voice-control processors and software stack (Xcore.ai) for smart home and edge audio. Closely comparable in targeting low-latency on-device voice interaction and selling silicon-plus-firmware to OEMs.
• Sensory: Embedded speech recognition, voice biometrics, and wake-word IP licensed to OEMs and silicon vendors - directly comparable to ABR's ASR IP and OEM-licensing motion in consumer electronics and appliances.
• Syntiant: Ultra-low-power Neural Decision Processors for always-on edge AI, with voice/audio wake-word and command recognition as a flagship use case. Syntiant pairs custom silicon with embedded models - the same hardware-plus-model stack that ABR pursues with TSP1.
• Ambiq Micro: Apollo-series ultra-low-power SoCs and AI-enabled MCUs with sub-mW class always-on AI targets. Closely matches ABR's <30mW voice AI positioning and its emphasis on enabling all-day battery operation on tiny devices.
• Picovoice: Direct on-device speech AI competitor offering Leopard/Cheetah (ASR) and Orca (TTS) models targeting edge and embedded platforms. Picovoice's privacy-first, offline voice stack most directly parallels ABR's Niagara/Nith models and per-device licensing model.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/abr_inc

Revenue estimate

Valuation estimate

Number of profiles

Profiles8 records

Funding overview

Funding rounds1 record

Investors5 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Applied Brain Research