Advanced Computing in the Age of AI | Saturday, November 27, 2021

With $136M in Fresh Series C Funding, AI Chipmaker Hailo Aims to Grow Its Edge AI Products, Services 

Four years after launching its AI chip business for edge uses such as autonomous vehicles and IoT devices, Israel-based Hailo just secured another $136 million in Series C funding to help continue to grow the company.

The latest funding round, which brings the company’s total funding to $224 million, was led by Poalim Equity and Gil Agmon. The round was joined by existing investors, including Hailo chairman Zohar Zisapel, Swiss-based ABB Technology Ventures (ATV), London’s Latitude Ventures and Israel’s OurCrowd. Also involved are new investors including Carasso Motors, Comasco, Shlomo Group, Talcar Corporation Ltd., and Automotive Equipment (AEV).

The latest $136 million cash infusion will be used to bring in more sales, support and development team workers to help grow Hailo’s core technology – its processor architecture – which it calls a structure-defined data flow architecture, Orr Danon, the company’s CEO and co-founder, told EnterpriseAI.

“If you think of traditional compute architectures, all of the computations are being done on the same code, sequentially,” said Danon. “The idea is to build a computer architecture that looks at moving data from place to place, rather than concentrating all the computations in one place. That makes it better because you avoid running into bottlenecks.”

Ultimately, that means that computations are done in smaller units, rather than in one large system, making things faster and more responsive, he said. “The bottom line of our technology is about 10x or more improvement in power efficiency for running deep learning.”

The company’s flagship product, its Hailo-8 chips, are based on an architecture that mimics the neural networks of the human brain, delivering high performance for smart devices that’s comparable to compute provided by large data center computers, according to the company. In late September, Hailo launched its M.2 and Mini PCIe high-performance AI acceleration modules to power edge devices.

The company’s chips, which are manufactured by TSMC, can be purchased alone or in multiple form factors and in different sizes and connector types, depending on a customer’s requirements, said Danon.

“If they want them for other things, for instance robotics, we can customize it,” he said. “But when the deployment is limited customers will take the chips and build their own system around it.”

Hailo does not disclose the process technology it uses for its chips, but Danon said that one model measures 22mm by 30mm, making it compact and packed with 26 terabits per second of compute power and low power consumption.

For end customers, this means that they can use it as an accelerator device for PCs or they can use it to perform real-time video analytics and more, he said.

“We are partnering with box makers, camera makers, to put this into devices and … the usages are quite diverse,” said Danon. “It can range from systems like advanced driving assistance systems, which use cameras mounted in cars, and goes on to robotics and industrial applications such as production line monitoring and safety monitoring.”

The company has been selling its products and has some 100 customers working with them in development of their own products, said Danon.

Orr Danon, Hailo CEO

This is where the new funding will come into play, he said.

“This money is going to help with all that and with continued research, marketing, building out tech support, building out all those things for the company,” he said. “Now I am building my team. I have recently opened offices in Taipei, Tokyo, Munich, on the east coast of the U.S. and on the west coast, as well as in Israel, to help me take this to the market.”

Lian Jye Su, an analyst with ABI Research, told EnterpriseAI that Hailo has an interesting niche.

“Hailo is a machine learning ASIC company that focuses on providing high AI inference workload performance for edge devices,” said Su. “Originally targeting the autonomous vehicle market, the company has now expanded into edge computing boxes. These edge boxes can support high-performing ML inference engines for mission- and business-critical machine vision applications, including video analytics, traffic management, and access control in a power-efficient manner.”

The company’s high-performance Hailo-8 processors compete primarily with Nvidia Jetson, Intel Movidius and Google TPU in the edge AI market, said Su. ABI Research forecasts that the edge AI chipset market will grow to $28 billion in 2026, with a CAGR of 28.4 percent between 2021 and 2026, which provides validation to Hailo’s recent Series C funding, he said.

Lian Jye Su, analyst

“However, to remain competitive, having good hardware and partnership are not sufficient,” said Su. “Hailo has also boosted its ML developer toolkit. Developers working on Hailo-8 can leverage the Hailo Software Toolchain, consisting of a dataflow compiler, runtime software and application toolkit. In addition, Hailo offers pre-trained ML models for specific applications, such as classification, depth and pose estimation, lane detection, facial detection and recognition and semantic segmentation.”

Add a Comment

EnterpriseAI