Ambarella unveils GenAI-powered edge AI SoCs targeting real-time video intelligence

Demonstrations included DeepSeek GenAI models, multi-stream video analysis, and on-device generative AI, highlighting real-time, multimodal video intelligence for security and analytics. 

Ambarella’s AI SoCs integrate image processing, encoding, and system-level functions, boasting superior AI performance per watt compared to competitors. 

“In addition to providing the industry’s best AI performance per watt, Ambarella stands apart from most competitors, which offer impractical stand-alone AI accelerators that act as add-ons to a main processor,” says Fermi Wang, President and CEO of Ambarella. “While we are continuing to enable advancements in GenAI processing at the edge, we also remain steadfast in our hallmark differentiators of not merely providing the proprietary CVflow AI accelerator, now in its third generation. Instead, we also integrate image processing, encoding and system-level functions into all of our AI SoCs, which is why our customers have been able to so rapidly deploy 30 million-plus Ambarella edge AI SoCs in the market, to date.”

The company introduced its AI Model Garden as part of the Cooper Developer Platform to simplify development for edge AI applications. Applications span security, retail analytics, public safety, and industrial monitoring, enabling real-time insights, anomaly detection, and predictive maintenance. 

Ambarella’s technology supports leading GenAI models (0.5 to 34 billion parameters) across various product tiers, ensuring flexibility and scalability. 

Last year Ambarella teamed up with LG to drive AI-powered in-cabin vehicle safety innovation.

Article Topics

AI SoCs  |  Ambarella  |  edge AI  |  edge computing  |  generative AI