Kneron Secures $49 Million to Propel AI Integration in Self-Driving Cars

Kneron In today's announcement, Kneron disclosed a $49 million raise in an extension to its Series B funding, culminating in a total of $97 million for this funding phase, with Horizons Ventures at the helm and notable contributions from giants like Foxconn, Alltek, Liteon Technology Corp, Adata, and Palpilot.

The fresh capital will be directed towards the advancement of edge AI chips specifically designed for self-driving vehicles. Notably, Kneron claims it introduced its first edge AI chip in 2021, engineered to facilitate transformer neural networks, which are at the core of all GPT-based models.

According to Albert Liu, Kneron's CEO, current GPT models are heavily reliant on cloud data centers, which leads to various challenges like increased latency, elevated data transfer costs, and concerns regarding user privacy and security. The objective is to address these issues within the industry by employing edge AI chips that operate directly on devices rather than through cloud servers.

Founded in 2015 by Liu and Frank Chang, Kneron emerged as a visionary project. Before starting this venture, Liu garnered significant experience in research and leadership roles at major companies, including Qualcomm and Samsung.

Since its inception, the company has successfully raised a total of $190 million.

As competition heats up, Kneron is poised to rival Nvidia, particularly as the appetite for premium AI chips continues to surpass supply. Unlike the currently dominant graphic processing unit (GPU) chips, Kneron is innovating with neural processing unit (NPU) chips, which the company believes are better suited for AI computations.

Kneron recently unveiled The KL370, a cutting-edge NPU chip designed for automotive applications, is touted to support the latest lightweight GPT large language models (LLMs). Kneron aims to produce these chips with a clear focus on enhancing autonomous driving functions.

By leveraging transformers, technology typically associated with language processing, Kneron applies it to visual data applications, aiming to improve the understanding and analysis of both temporal data and images.

Thanks to improvements in contextual interpretation and processing capabilities, Kneron asserts that it can deliver at least a 30% increase in accuracy for image-oriented applications.

Alongside the recent financial infusion, Kneron is intensifying its collaboration with Foxconn to accelerate the integration of AI technologies within the automotive sector. This partnership will involve creating an ultra-lightweight AI chip capable of operating GPT models independently of cloud infrastructures.