Kneron, a San Diego-based semiconductor company, has secured an additional $49 million in Series B funding, bringing its total funding for autonomous vehicle AI chips to $97 million. The funding round was led by Horizons Ventures, with participation from Foxconn, Alltek, Liteon Technology Corp, Adata, and Palpilot.
The primary focus of Kneron is the development of AI chips specifically designed for self-driving vehicles. The company claims to have introduced the first edge AI chip in 2021, designed to support transformer neural networks. These networks are the foundation of various models, including the popular GPT (Generative Pre-trained Transformer) models.
According to Albert Liu, the CEO of Kneron, current GPT models heavily rely on cloud data centers, resulting in issues such as high latency, expensive data transfer costs, and concerns regarding user privacy and security. Kneron aims to address these bottlenecks by utilizing its edge AI chips, which run directly on the device itself, eliminating the need for cloud dependencies.
Founded in 2015 by Albert Liu and Frank Chang, both with extensive experience in research and management roles at companies like Qualcomm and Samsung, Kneron has raised a total of $190 million since its inception.
Kneron is positioning itself as a potential competitor to Nvidia in the AI chip market, as the demand for high-quality AI chips continues to outweigh supply. Instead of GPU (Graphic Processing Unit) chips that currently dominate the market, Kneron specializes in developing NPU (Neural Processing Unit) chips. The company claims that its NPU chips are more suitable for AI computing.
Recently, Kneron unveiled the KL370, an automotive-grade NPU chip specifically designed to support lightweight GPT LLMs (Large Language Models). The company’s focus is on enabling autonomous driving capabilities through the development of these chips.
By leveraging transformer technology, typically used in language processing, Kneron applies it to image-based applications. This approach enhances the handling of time-series data and facilitates holistic analysis of image data, leading to a minimum 30% improvement in accuracy for image-based applications.
In addition to securing new funding, Kneron is expanding its partnership with Foxconn to accelerate the implementation of AI technologies in the automotive industry. The collaboration includes the development of an ultra-lightweight AI chip capable of independently running GPT models without relying on cloud resources.
With its innovative edge AI chips and strategic partnerships, Kneron aims to revolutionize the autonomous driving industry, addressing crucial challenges associated with latency, data transfer costs, privacy, and security. The company’s efforts to develop NPU chips and its application of transformer technology showcase its commitment to advancing AI computing capabilities for self-driving vehicles.
