EdgeCortix, a Japanese startup, is developing energy-efficient AI processors for edge computing. The company’s CEO, Sakyasingha Dasgupta, a 25-year veteran of the AI industry, founded EdgeCortix in July 2019. The startup’s unique approach combines hardware and software co-exploration to meet the rising demand for AI applications while improving energy efficiency and processing speed.
Dasgupta said, “By processing data at the edge, we can go where the problem is, as 75% of the world’s data is generated or processed at the edge.” He emphasizes a software-first approach to make computing more efficient and tackle the limitations imposed by the stalling of Moore’s Law. EdgeCortix’s mission is to bring near cloud-level performance to edge devices, including robotics, smart city devices, manufacturing systems, and satellite technology. The company’s software enables various processors to work together seamlessly in heterogeneous computing environments.
The startup began by creating a compiler software to translate machine-focused assembly code into higher-level languages, forming the core of its Compiler and Software framework and its Dynamic Neural Accelerator (DNA).
Developing energy-efficient AI processors
This platform provides robust support for deploying cutting-edge neural network models and integrates easily with existing systems.
Renesas, a prominent Japanese chip company, became one of the early adopters of EdgeCortix’s software. The startup has also licensed its software to other chip companies, generating revenues while advancing its mission to create efficient neural processors. EdgeCortix transitioned to a fabless semiconductor model and released the Sakura chip, followed by the improved Sakura-II AI Accelerator.
This coprocessor delivers 60 trillion operations per second with just eight watts of power, making it suitable for applications requiring fast, real-time AI inferencing. The World Economic Forum named EdgeCortix a Technology Pioneer in 2024, recognizing it as one of the world’s 100 most promising startups. The company collaborates with industry leaders to provide high-performance AI inference solutions across various domains, including defense, robotics, smart manufacturing, smart cities, and automotive sensing.
