Kneron has raised an additional $49 million in its Series B funding round, bringing the total raised to $97 million. The San Diego-based AI startup says it will leverage the new capital to expand its full-stack hardware and software solutions that enable on-device AI inferencing for autonomous vehicles.
In an industry voraciously hungry for top-tier AI computing, the spotlight is on the availability of superior AI chips. Though GPU chips are presently the market's darling, Kneron's trailblazing work with NPU (neural processing unit) chips offers a bespoke solution tailored for AI computing. The company is addressing the high latency, security risks, and costs associated with cloud-based AI by creating specialized AI chips for efficient on-device processing. Last month, the company announced their KL730 chip that delivers up to 4x better energy efficiency over previous models, providing a optimized option for organizations seeking more affordable silicon for AI workloads.
Crucially, the KL730 architecture supports transformer-based models like nano GPT. As large language models and generative AI continue gaining traction, Kneron's ability to run models like GPT-4 directly on devices unlocks new possibilities while maintaining privacy and lowering power usage.
With a formidable clientele, including industry giants Toyota, Foxconn, Quanta, and Hanwha, Kneron is zeroing in on elevating image processing standards, a paramount necessity for the evolution of autonomous vehicles.
The extended round was led by Horizons Ventures with new participation from strategic partners like Foxconn. Kneron will leverage the funding to advance AI innovations for autonomous driving, including using transformers to improve image recognition accuracy by over 30% - a major boon for perception in self-driving cars.
Albert Liu, the visionary Founder and CEO of Kneron, reflected on the milestone, noting the prevailing trend of potent GPT models largely operating from cloud data centers. Such a setup, he remarked, gives rise to pressing challenges such as heightened latency, soaring data transfer expenses, and glaring shortcomings in safeguarding user privacy and security. "Our objective," Liu stated, "is to address these industry snags by designing hyper-efficient AI chips. With this fresh capital from our Series B funding, we're positioned to amplify our efforts in rendering AI more secure, universally accessible, and energy-conservative."
With the added capital, Kneron's sights are set on harnessing AI to actualize autonomous driving. The firm's cutting-edge ultra-lightweight AI chips utilize transformers, a technique typically employed for language processing, for image-centric applications. This innovative approach, leveraging transformers' prowess in streamlining time-series and holistic image data assessment, propels Kneron's image-based application accuracy by a minimum of 30%, marking a transformative stride for self-driving vehicles.
With over $190 million raised to date, Kneron aims to accelerate the deployment of performant yet accessible edge AI to enable the next generation of intelligent devices and vehicles.