Hailo was an Edge AI company founded in 2017 and based in Tel Aviv, Israel. It was acquired by Microchip Technology in 2026.
| Date | Stage | Amount | Valuation | Lead investors |
|---|---|---|---|---|
| Apr 2, 2024 | Series C | $120M | $1.2B | — |
| Oct 1, 2021 | Series C | $136M | $1B |
| — |
| Mar 5, 2020 | Series B | $60M | — | ABB Technology Ventures, NEC Corporation |
| Jun 5, 2018 | Series A | $12.5M | — | OurCrowd |
An edge AI inference processor delivering up to 26 TOPS at very low power (typically ~2.5W), built on a dataflow architecture that maps neural networks directly onto the chip's compute and memory. Available as M.2/mPCIe modules, the Hailo-8 powers real-time computer vision in smart cameras, retail, industrial, and automotive aftermarket devices. It is Hailo's most widely deployed product, shipping in large volumes.
A family of AI vision system-on-chips that integrate Hailo's neural engine with an image signal processor and CPU, enabling cameras to run advanced deep-learning models directly on-device. Targeted at smart-city, security, and enterprise cameras, Hailo-15 delivers up to ~20 TOPS and on-camera analytics without cloud dependence, extending Hailo from a discrete accelerator into a complete edge-vision platform.
An edge accelerator designed to run generative-AI and large-language/vision models locally on PCs and edge devices, bringing LLM inference to power- and cost-constrained hardware without sending data to the cloud. Hailo-10 targets the AI PC and on-device GenAI wave, offering high performance-per-watt for assistants and multimodal models, and broadens Hailo's portfolio beyond classic computer-vision inference.
Automotive-grade variants of Hailo's accelerators qualified to AEC-Q100 reliability standards for ADAS and in-cabin perception, enabling object detection, driver monitoring, and sensor processing within a vehicle's power and thermal budget. With design wins among automotive OEMs and tier-1 suppliers, this line positions Hailo in the high-volume automotive edge-AI market alongside its industrial and smart-camera business.
3 patents on file, but none with both an extractable figure and an abstract on Google Patents yet.