Overview
Syntiant Corp., founded in 2017 and based in Irvine, California, is a technology company specializing in end-to-end deep learning solutions for edge AI applications. The company's focus is on developing ultra-low-power, high-performance deep neural network processors that enable machine learning directly on devices, reducing reliance on cloud computing. Key Products and Technologies:
- Digital neural decision processors mimicking the human brain for efficient workload processing
- Use-case specific machine learning models for edge processing in sensor, voice, and video applications
- Low-power, high-performance solutions suitable for compute-constrained environments Partnerships and Funding:
- Backed by prominent investors including Intel Capital, Microsoft's M12, Applied Ventures, Robert Bosch Venture Capital, Amazon Alexa Fund, and Atlantic Bridge Capital
- Total funding of $122.93 million, with the latest being Series C - III Applications and Use Cases:
- Utilized in various sectors including consumer electronics, industrial automation, and automotive
- Notable application in Ring Alarm Glass Break Sensor for efficient, private edge detection Growth and Recognition:
- Over 100 employees across the United States, Asia, and Europe
- Recognized as one of the "Best Places to Work in Orange County" for five consecutive years
- More than 50 million devices deployed globally using Syntiant's technology Vision and Philosophy: Syntiant aims to create seamless human-technology interactions using natural interfaces like voice or gestures, processing data at the edge to enhance privacy and user experience while reducing data congestion between devices and cloud servers.
Leadership Team
Syntiant's leadership team comprises experienced professionals with diverse backgrounds in technology, business, and academia: CEO:
- Kurt Busch: Known for crisis management, team building, and fostering innovation. Expertise in IoT, Networking, SAAS, and Semiconductor Industries. Chief Technology Officer (CTO):
- Jeremy Holleman, Ph.D.: Expert in ultra-low power integrated circuits, directs the Integrated Silicon Systems Laboratory at UNC Charlotte. Founder and Chief Operating Officer (COO):
- Pieter Vorenkamp: Background includes roles at Cadence, Broadcom, NXP Semiconductors, and Philips. Other Key Executives:
- Robert Saman
- Greg Doll
- Paul Henderson
- Dr. Stephen Bailey Advisory Board:
- Kunle Olukotun, Ph.D.: Chief technologist and co-founder of SambaNova Systems, Stanford University professor
- Greg Fischer: Board member at Semtech Corporation
- Magnus Egerstedt, Ph.D.: Dean of Engineering at UC Irvine
- Sander Arts: Founder and CEO of Orange Tulip Consultancy The leadership team focuses on advancing machine learning and neuromorphic computing innovations, driving growth and maintaining Syntiant's position in intelligent edge devices through effective cross-departmental collaboration.
History
Syntiant, a leader in edge-AI deployments, has experienced rapid growth and innovation since its founding in 2017: Founding and Early Vision (2017):
- Founded by four co-founders, including CEO Kurt Busch
- Inspired by the potential of AI and machine learning in voice-activated systems Initial Development and Partnerships (2017-2018):
- Began researching machine learning, crafting audio wake words, and designing first product
- Leveraged resources from Evonexus startup incubator
- Partnered with Arm for processor core and IP First Product and Achievements (2018-2020):
- Completed design of NDP100 based on Syntiant Core 1 in March 2018
- Shipped over 10 million units
- Certified on Amazon Alexa
- Raised over $65 million in venture capital Expansion of Product Line (2020-2022):
- Introduced NDP120 based on Syntiant Core 2
- Developed NDP102 for sensor processing and NDP200 for vision and image recognition Growth and Global Reach (2022):
- Shipped over 20 million Neural Decision Processors worldwide
- Worked with approximately 80 customers globally Acquisitions and Partnerships (2022-Present):
- Acquired Pilot AI Labs in October 2022
- Formed strategic partnerships with Avnet and TDK Corporation Current Focus and Impact:
- Continues to innovate in moving AI from cloud to edge
- Enables local AI tasks on devices, ensuring privacy and security
- Involved in voice digital biomarkers for health monitoring and industrial IoT solutions Syntiant's history reflects its commitment to advancing edge AI technology and expanding its applications across various industries.
Products & Solutions
Syntiant specializes in edge-AI deployments, offering innovative products and solutions designed to bring deep learning capabilities to a wide range of devices. Their offerings include:
Neural Decision Processors (NDPs)
Syntiant's NDPs are custom silicon solutions that efficiently run deep learning models on edge devices:
- Offer 100x efficiency and 10-30x higher throughput compared to existing low-power microcontrollers (MCUs)
- Utilize at-memory compute to reduce power consumption and latency
- Achieve high efficiency (often >80%) in processing neural network layers
Hardware-Agnostic Machine Learning Models
Syntiant's deep learning models are designed to run on various hardware platforms, from large GPUs to small MCUs:
- Provide production-ready, off-the-shelf models for audio, speech, sensor, and computer vision applications
- Ensure compatibility with both legacy and modern compute architectures
Edge AI Solutions
Tailored for always-on, battery-powered edge devices, enabling real-time data processing with near-zero latency:
- Audio and Vibration Sensors: High-performance MEMS microphones and vibration sensors for premium audio and active noise cancellation
- Applications across Smart Home, Automotive, Personal Devices, Government, and Industrial & Commercial sectors
Multimodal AI Solutions
Syntiant collaborates with companies like Renesas to offer solutions combining vision and voice processing:
- NDP120 Neural Decision Processor, when paired with Renesas' RZ/V2M vision AI microprocessor unit, provides advanced voice and image processing capabilities at the edge
Deployment and Development Support
Syntiant offers streamlined solutions to help developers quickly deploy deep learning models:
- Optimized training pipelines for edge applications
- Support for rapid time-to-market Syntiant's products and solutions are designed to enhance the performance, efficiency, and user experience of edge devices through powerful deep learning capabilities.
Core Technology
Syntiant's Core technology comprises a series of programmable deep learning architectures designed for ultra-low-power, high-performance edge AI applications:
Syntiant Core 1
- First-generation neural network processor
- Optimized for very low energy consumption, particularly in audio keyword-spotting applications
- Features a five-layer fully connected network with 4-bit weights and biases
- Operates at 16 MHz with 8 parallel MACs
- Tailored for near-field wake word detection and simple audio processing tasks
Syntiant Core 2
- Significant advancement over Core 1
- Offers a highly flexible and configurable neural network runtime
- Provides up to 25x the processing power of Core 1
- Supports a wide range of neural network architectures
- Can execute multiple layers simultaneously without compilers
- Demonstrated outstanding performance in the MLPerf Tiny v1.1 benchmark suite
Syntiant Core 3
- Latest generation, integrated into the NDP250 Neural Decision Processor
- Delivers 5x the tensor throughput of the previous generation
- Designed for various imaging, speech, and sensor applications
- Capable of handling complex tasks like person detection, object classification, automatic speech recognition, text-to-speech, and motion tracking
- Particularly suited for real-time speech interfaces for large language models and battery-powered, always-on vision applications
Key Features
- Ultra-Low Power Consumption: Suitable for battery-powered devices
- High Performance: High throughput and low latency for real-time processing
- Flexibility and Versatility: Support for a wide range of neural network architectures
- Multi-Modal Capabilities: Can handle multiple applications simultaneously, including edge AI vision, speech, audio, and sensor processing Syntiant's core technologies focus on delivering high-performance, ultra-low-power solutions for edge AI applications, positioning the company as a leader in edge AI deployment.
Industry Peers
Syntiant operates in the competitive fields of edge AI and neuromorphic computing. Here are some notable industry peers and competitors:
Edge AI and AI Processors
- Hailo: Focuses on edge AI processors for embedded deep learning applications in automotive, security, and industrial automation sectors
- MemryX: Develops AI accelerator chips for edge devices in automotive, robotics, and machine vision
- Kneron: Provides integrated software and hardware solutions for on-device AI processing
Neuromorphic and Low-Power Computing
- Innatera: Develops low-power intelligence for sensors using neuromorphic processors that mimic brain mechanisms
- Perceive: Specializes in edge inference chips for various edge devices, offering high accuracy and performance at low power
AI Inference and Accelerators
- Analog Inference: Focuses on analog in-memory computing technology for AI inference accelerators
- Blaize (formerly ThinCI): Offers full-stack hardware architecture and no-code software platform for AI applications
Semiconductor and AI Hardware
- POLYN Technology: A fabless semiconductor company developing application-specific Tiny AI chips
- STMicroelectronics: Creates semiconductor technologies, including intelligent and energy-efficient solutions
Other Competitors
- GreenWaves: Known for their GAP8 IoT processor designed for low-power AI applications
- Silicon Labs: Provides semiconductor products for IoT and edge computing
- SiTime: Specializes in MEMS-based timing solutions and competes in the broader semiconductor and IoT space These companies are involved in various aspects of AI, edge computing, and semiconductor technologies, competing with Syntiant in different market segments. The competition drives innovation and advancements in edge AI and neuromorphic computing, benefiting the overall industry.