Hailo is a startup founded in 2017 by Hadar Zeitlin, Avi Baum and Orr Danon, all three of whom served in the technology unit of IDF, the Israeli army. Hailo specializes in manufacturing chips for Edge peripheral devices. In particular, it has developed Hailo-8, a processor for deep learning. It has just raised $136 million and thus enters the very closed club of unicorns.
Aileen Lee, an American analyst and creator of an investment fund, was the first to call unicorns American companies in the field of new technologies, less than 10 years old, not listed on the stock exchange and worth more than a billion dollars. Since then, the term unicorn has become commonplace worldwide for start-ups that meet these conditions. The funds raised by these companies since their launch make it possible to determine the evolution of their valuation.
This latest Hailo investment was led by Poalim Equity (FKA Poalim Capital Markets), the Tel Aviv-based investment platform of Bank Hapoalim, founded in 1990, and Gil Amon, CEO of Delek Motors. Other investors in the fundraising include Zohar Zisapel, a leading entrepreneur and chairman of Hailo, ABB Technology Ventures, Latitude Ventures, Our Crowd and new investors Carasso Motors and Schlomo among others.
This Series C round of funding is the largest in the Edge AI chip space. Hailo plans to develop a new generation of products and expand worldwide. It already has offices in Munich, Tokyo, Taipei, Silicon Valley and has an agreement with Macnica in Japan. It will of course continue to develop Hailo-8.
This AI chip has been designed for equipment located at the edge of networks (autonomous vehicles, industrial robots, smart home, personal assistants, smart cameras and TVs, drones…).
The processor is a “fundamental element that performs data processing for deep learning, and is used to process data such as video, to perform real-time video analytics that can be used in the context of cars or driver assistance systems, or to monitor product quality on a production line, or to secure perimeters for access control, but the common thread is that all of these applications need a powerful engine that can traverse the data and get insights. We provide the infrastructure for that, the processor, which is the heart of the system.
“Our processor does all of these tasks… with very low power consumption,”
said Orr Danon.
Its innovative architecture is based on the fundamental properties of neural networks and is expected to enable Edge devices to run Deep Learning applications without going through the cloud. It accelerates embedded AI applications to run 26 tetraoperations per second (Tops) and an eco-efficiency of 3 tops/W.