Check out the on-demand sessions from the Low-Code/No-Code Summit to learn how to successfully innovate and achieve efficiency by upskilling and scaling citizen developers. Watch now.

Hailo, a startup developing AI accelerator chips for edge devices, today announced that it raised $136 million in a series C funding round led by Poalim Equity and entrepreneur Gil Agmon, with participation from ABB Technology Ventures, Latitude Ventures, OurCrowd, Carasso Motors, Comasco, Shlomo Group, Talcar Corporation, and Automotive Equipment. The capital brings Hailo’s total raised to $224 million, and CEO Orr Danon says the company will put the funds toward mass manufacturing of its current chips and the development of next-generation hardware in new and existing markets.

AI edge chips are coming into vogue as enterprises begin to realize the benefits of localized computing. Deloitte estimates that in 2020, over 750 million edge AI chips — chips or parts of chips that perform or accelerate machine learning tasks on-device — were sold globally. This number, representing $2.6 billion in revenue, is more than twice the 300 million AI chips the firm anticipated would sell in 2017, a three-year compound annual growth rate of 36%.

Hailo’s Hailo-8 processor features an architecture that ostensibly consumes less power than rival AI chips while incorporating RAM, software control, and a heat-dissipating design. Under the hood of the Hailo-8, resources including memory, control, and compute blocks are distributed throughout the whole of the chip, and Hailo’s software analyzes the requirements of each AI algorithm and allocates the appropriate modules.

“I founded Hailo in 2017 along with colleagues from the Israeli Defense Forces’ elite technology unit. While working with my cofounder Rami Feig on cybersecurity solutions for internet of things, a lesser-known term — ‘deep learning’ — kept popping up throughout our research,” Danon told VentureBeat. “Eventually, we brought together Avi Baum, an expert in machine learning, and Hadar Zeitlin, a smart mobility guru, to develop a new deep learning solution that aimed to solve the shortcomings of aging computer architecture in order to enable smart devices to operate more effectively and efficiently at the edge. After Rami’s unfortunate passing [in 2018], Avi, Hadar, and I carried through his vision of creating Hailo’s groundbreaking processor.”


Intelligent Security Summit

Learn the critical role of AI & ML in cybersecurity and industry specific case studies on December 8. Register for your free pass today.

Register Now

AI at the edge

Edge AI” describes architectures in which AI models are processed locally, on hardware at the edge of a network. As edge AI setups typically only require a microprocessor and sensors, not an internet connection, they can process data and make predictions in real time — or close to it.

The business value of edge AI could be substantial. According to Markets and Markets, the edge AI market is anticipated to grow from $590 million in 2020 to $1.83 billion by 2026. Digital health care, manufacturing, and retail businesses are particularly likely to expand their use of edge computing because of the technology’s ability to improve response times and save bandwidth while enabling less constrained data analysis.

“Smart retail, for instance, requires hundreds of cameras in-store, where data must be processed locally, quickly, and efficiently with minimal latency. Smart cities require similar output from a multitude of cameras and sensors — otherwise, critical city management tasks such as improving transportation flow will be much harder to achieve,” Danon continued. “For all of these industries and more, Hailo’s AI chip can process full resolution data streams in real time, enabling cameras to detect objects from greater distances. The processor is also capable of processing multiple camera streams on a single device, making it more cost-effective.”

Hailo-8 chip

The Hailo-8 is capable of 26 tera-operations per second (TOPs), which works out to 2.8 TOPs per watt. In a test conducted by Hailo, the Hailo-8 outperformed hardware like Nvidia’s Xavier AGX, Intel’s Myriad-X, and Google’s Edge TPU modules on several AI semantic segmentation and object detection benchmarks. At an image resolution of 224 x 224 pixels per inch, the chip processed 672 frames per second compared with the Xavier AGX’s 656 frames and used 1.67 watts (equating to 2.8 TOPs per watt) versus the Nvidia chip’s 32 watts (0.14 TOPs per watt).

Hailo recently announced the launch of its M.2 and Mini PCIe high-AI acceleration modules, which are based around the Hailo-8 chip. Foxconn and Socionext are two of the first publicly disclosed customers after NEC, Macnica, and ABB Technology — Foxconn’s BOXIedge edge computing platform features Hailo’s M.2 module. Previously, Hailo said it’s working to build its chips into products from OEMs and tier-1 automotive companies in fields such as advanced driver-assistance systems and industries like robotics, medicine, and smart cities and homes.

According to Danon, Hailo, which has more than 160 employees, doubled its customer base to upwards of 100 clients throughout the first two quarters of 2021 and expanded its presence globally, opening offices in Tokyo, Taipei, Munich, and Silicon Valley in California.

“AI, more specifically AI at the edge, will continue to drive innovation across sectors in the coming years … As businesses seek solutions that ensure their devices are more powerful, versatile, responsive and secure, the cloud will continue giving way to the edge in the coming years. Though a hybrid model will certainly take center stage, those who succeed in implementing AI at the edge will gain an edge across the board,” Danon said.

Hailo has rivals in Deep Vision, Alexera, Sima.aiAIStorm, Quadric, and Flex Logix, which are similarly developing chips customized for AI workloads. Mobileye, the Tel Aviv company Intel acquired for $15.3 billion in March 2017, also offers a computer vision processing solution for AVs in its EyeQ product line. Baidu last July unveiled Kunlun, a chip for edge computing on devices and in the cloud via datacenters. And Arm debuted two new AI edge computing chips in February.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Discover our Briefings.