Beijing Chaoxing Future Technology Co., Ltd., a provider specializing in AI chips for edge computing, recently announced the completion of a Pre-B round of financing worth hundreds of millions of yuan. The company, known as “Chaoxing Future” for short, has received support from investors such as Zhong An Capital, Liangxi KeChuang, Longding Investment, Tianzhi Investment, Shaanxi Auto Intelligent Vehicle Fund, and iFlytek Venture.
The funds raised will mainly be used to promote the development of the next-generation large model inference chips, expand the scale of existing revenue businesses, and further deepen industrial cooperation. In the AI industry chain, computing chips are revered as the core component and driving force of the field.
In terms of AI chip classification, they can be divided into four major categories based on the cloud/edge side and training/inference. Edge AI inference chips are closely integrated with the actual scene requirements. As large models move towards a key application stage, the importance of inference chips becomes increasingly significant. Research finds that smaller “small models” can unleash the potential of edge AI scenarios, also bringing a highly potential development space for the market.
At the beginning of 2023, at the CES in the United States, AI-on-device scenarios emerged in abundance, covering AI smartphones, AI PCs, AI cars, wearable AI devices, and embodied intelligence applications. Qualcomm’s white paper points out that localized AI deployment needs to meet requirements of low cost, low power consumption, low latency, high performance, as well as a focus on privacy and personalization, posing new challenges for chip design.
Founded in April 2019, Chaoxing Future has been focusing on providing efficient AI computing chips and collaborative software and hardware solutions for various edge intelligence scenarios from the outset, striving to lead the development of edge AGI computing. Its products have been applied in fields such as smart driving, smart electricity, and smart mining. The company’s core team has a strong Tsinghua University gene, with members having rich R&D experience in the field of high-performance digital chips.
For example, Chaoxing Future’s CEO Liang Shuang is one of the early researchers in the AI accelerator processor field, having participated in the development of China’s first AI chip “Tingtao”. CTO Chen Zhongmin has more than 20 years of chip technology R&D experience, having worked at AMD and was in charge of AI accelerator architecture solutions.
A report by the well-known investment firm Coatue shows that the Edge AI field has become the focus of the second phase of AI’s S-curve deployment. The Chaoxing Future team relies on their abilities in inference architecture design, system-level computation architecture design, large SoC chip integration, algorithm embedded deployment, network model compression optimization, and industry solution development, to deeply explore and develop edge AI applications.
Leveraging its leading technological platform, Chaoxing Future Company has developed its product line and launched several core computing products for edge AI. This batch of products is based on the advanced “Pinghu” and “Gaoxia” generations of AI inference engines. These include the edge AI computing chip “Jingzhe R1”, the intelligent computing development kit “NE100”, and the edge computing module “NM10”.
“JingzheR1” chip utilizes TSMC’s 12nm manufacturing process, providing up to 16TOPS of INT8 computing power and 30K DMIPS of general computing capabilities. It supports multi-sensor fusion and has flexible connectivity options, reducing the overall application development costs by 30%-50%, which greatly promotes the application of edge-side artificial intelligence.
In addition to strong hardware support, Chaoxing Future has also launched a set of AI deployment development toolchain named “Luban”. This toolkit includes pruning tools, quantization tools, compilers, runtimes, Middleware, and BSP, etc., comprehensively supporting a variety of operators, assisting users in efficiently optimizing and deploying AI algorithms, and creating a cost-effective edge-side AI application environment.
In terms of practical application scenarios, Chaoxing Future has ventured into areas such as intelligent driving, smart electricity, smart energy, unmanned systems, and embodied intelligence. Particularly in automotive, electric power, and coal mining industries, the company provides AI reference design solutions based on its self-developed chips, which have conspicuous ToB (to business) commercial attributes.
During its development, Chaoxing Future was focused on the field of intelligent driving, launching the “NOVA-ADCU” intelligent driving reference solution, covering integrated driving and parking needs from low to high levels. By mid-2023, the company further expanded its business scope, signing a strategic cooperation agreement with Zhengzhou Coal Mining Machinery Group’s subsidiary, Hengda Intelligent Control, to jointly expand the market for wellhead and underground AI analysis equipment and edge management platform software in the mining industry.
Regarding the direction and strategy of Chaoxing Future, its COO Zhu Yuqi once explained that local deployment of AI applications requires attention to both technology and scenarios. With the enrichment of Edge AI scenarios, the company has more opportunities to seize upon. Zhu Yuqi emphasized that the company has always been focused on high energy-efficient computing carriers, considering automobiles as typical scenarios for edge-side AI applications, and believes that fields like electric power and coal mining have great potential.
The “hematopoietic ability” of Chaoxing Future is also remarkable. It is reported that in 2023, the company’s annual revenue has reached tens of millions of yuan. Insider information revealed that Chaoxing Future has achieved large-scale model deployment at the edge side, with performance several times higher than existing products on the market. Looking ahead to the future, the team plans to combine chips with large models in 2024, further explore new application scenarios such as AI PCs and embodied intelligence, and soon expand into the new field of large model inference computing.
The company’s CEO, Liang Shuang, holds a clear business vision, aiming to discover and build solutions with true commercial value. According to Liang Shuang, the company has already made initial progress in the commercial vehicle sector and expects to achieve application deployment in the passenger car sector by 2024. The edge-side artificial intelligence field has vast potential, and the company expects to introduce new R&D achievements and announce relevant cooperation information in the middle of the year. Team Liang expressed full confidence in this round of artificial intelligence development opportunities.
[ad_2]