Ambarella Unveils N1 SoC Series to Power Generative AI on Edge Devices

Ambarella, Inc., a company specializing in edge AI semiconductors, has unveiled its latest development: the N1 System-on-Chip (SoC) series. Announced at CES, the N1 series is engineered to run multi-modal large language models (LLMs), a feature previously limited to servers with significant processing capabilities. The new SoC series represents a breakthrough in bringing generative AI technology to edge devices such as video security systems, robots, and a variety of industrial tools.

Single SoC for Multi-Modal LLMs

The N1 series stands out by supporting LLMs ranging from one to 34 billion parameters with remarkably low power consumption. The SoC can handle complex tasks in multi-modal LLMs while ensuring power efficiency for edge endpoint devices. The versatility of the N1 SoC series is expected to revolutionize the implementation of generative AI in on-premise applications where power and cost factors are critical.

Enhanced Efficiency and Integration

Ambarella's new offering is not just about advanced AI capabilities; it's also focused on efficiency. In comparison to leading GPUs, the N1 SoC series provides power-efficient performance, up to three times more so per generated token. Moreover, Ambarella aims to simplify and quicken the deployment process for manufacturers, accentuating the appeal of its complete SoC solutions over standalone AI accelerators.

Redefining Edge AI Technology

The potential applications for the N1 series are vast, with Ambarella's Chief Technology Officer touting the transformative functions that LLMs can bring. In addition to the technological advancement, industry analysts highlight the significance of performance per watt and edge ecosystem integration - two areas where the N1 series shows promise. Combating the need for raw throughput with efficiency and functionality continues to be a priority for Ambarella.

Propelling Forward with the Cooper™ Developer Platform

To assist customers in expediting their products to market, Ambarella offers the Cooper™ Developer Platform, supporting their AI SoCs with pre-ported and optimized models accessible from the Cooper Model Garden. This library enables partners to fast-track the development and application specific tuning necessary for the edge devices.

Revolutionizing On-Device Processing

The Ambarella SoC architecture is uniquely designed to handle simultaneous video and AI processing with minimal power draw. This optimal configuration for multi-modal LLMs opens doors to smarter on-device applications. Edge processing benefits include speed, privacy, and reduced costs, aligning with the push towards localized, application-specific solutions over the traditional server-based approach.

Power Meets Practicality

The N1 SoC series, built on Ambarella's CV3-HD architecture, focuses on low-power consumption while maintaining high performance for multi-modal LLM operations. It enables a wide array of AI-intensive tasks to be performed on edge devices, making it a game-changer for industries looking to integrate sophisticated generative AI features while managing power and cost efficiencies.

About Ambarella

Ambarella's contributions to the field of human vision and edge AI applications are widely recognized. Its innovative SoCs facilitate high-resolution video compression, advanced processing, and powerful deep neural network processing for intelligent perception and planning. The introduction of the N1 SoC series reinforces Ambarella's commitment to advancing edge AI technology.

Ambarella, GenerativeAI, EdgeDevices