Qujing Technology completes Series B financing, achieves more than tenfold growth within half a year, and accelerates the large-scale delivery of high-quality AI Tokens | Xinglian Portfolio Updates
Recently, Approaching.AI, a high-quality AI Token production service provider, has closed its Series B financing with a total fundraising amount exceeding RMB 1.3 billion; the total capital raised in 2026 has reached RMB 2.4 billion. Multiple core business indicators have achieved more than tenfold growth within half a year, continuously gaining high recognition from the industry and capital markets.
This round of financing is jointly led by Zhecai Innovation, Xiaoshan Capital, ICBC Investment, Zhongguancun Science City, StarLink Capital and Qianjiang Century City. Follow-on investors include Yunchuang Capital, Hunan Gaoxin Zongheng, Jiantou Huake, Shenwan Innovation Investment, CITIC Juxin, Zhuhai Science and Technology Industry Group, SenseTime, Wangsu Science & Technology, Gansu Shuzhi, Changzhou Financial Control, Yuanzhi Future, Qingzheng Xingzhu, Xiaoka Capital, etc., while existing shareholder Shanshi Capital continued to increase its stake.
Following the completion of its Series A financing in July this year, Approaching.AI has secured another large-scale financing in a short period. As of October 2026, Approaching.AI has successively completed multiple rounds of financing in 2026. The intensive follow-up of capital not only reflects the market's consensus on the long-term value of the AI infrastructure track, but also stems from investors' recognition of Approaching.AI's comprehensive strengths including strategic focus, technological leadership and commercial delivery performance. Over the past half year, Approaching.AI has not only fully run through the full link from technical architecture to large-scale commercial delivery, but also continued to lead the technological evolution of the inference industry, driving a substantial performance leap of domestic computing power through technological innovation.
The funds raised in this round will be mainly invested in the expansion of high-quality AI Token production capacity, the iteration and upgrade of the ATaaS (Approaching.AI Token as a Service) platform, the R&D of underlying inference technologies and new products, as well as the improvement of commercial cooperation and delivery systems.
Investor's Voice: Chen Dong, Chairman of Zhecai Innovation
"As the AI industry moves towards large-scale application, inference efficiency, cost and system optimization capabilities are becoming important foundations for commercialization. Approaching.AI has accumulated continuous technological experience and engineering delivery capabilities in the field of large model inference. We look forward to long-term cooperation with the company to jointly promote the construction and large-scale application of efficient and inclusive AI infrastructure."
Investor's Voice: Zhu Lifeng, Chairman of Xiaoshan Capital Group
"Approaching.AI is a benchmark project for Xiaoshan's layout in AI Infra. Back when the industry was focusing on model training, the company proactively delved into the inference track. Relying on the more than ten years of technical accumulation of its Tsinghua-origin team and its self-developed ATaaS platform, it efficiently converts computing power into high-quality AI Tokens that can be produced at scale, promoting domestic computing power to enter the core production links. The two sides are working together to promote the construction of a ten-thousand-card level AI Token factory, accelerating the formation of an industrial closed loop of "chip-model linkage". Xiaoshan Capital will provide in-depth empowerment as long-term and patient capital, support it to quickly replicate the capabilities of high-quality AI Token factories, and inject lasting impetus into the high-quality development of the AI industry."
Investor's Voice: Zhang Mingchen, Founding Managing Partner of StarLink Capital
"Since StarLink Capital became a shareholder in the 2024 angel round, we have witnessed Approaching.AI continuously create far-expected achievements in technology, products, services, strategy and commercialization; and it has rapidly grown into a top-tier outstanding enterprise in the AI inference field in less than three years. This is the fundamental reason why StarLink Capital has continued to make large additional investments in every previous round. We are optimistic about the broad growth space of Approaching.AI as it moves from technology and commercial verification to large-scale construction and continuous operation, and we look forward to accompanying the team to continue leading the development of the AI inference industry."
From "Extensive Supply" to "Quality Service" | High-quality AI Token Becomes the New Trend of the Industry
In the early stage of the Token market development, the supply side of the industry mostly "prioritizes Token quantity over service quality", and rapidly expands the supply scale through extensive volume expansion. At that time, large models were still in the pilot verification stage and not deeply embedded in the core production links of enterprises, and the business had a large fault tolerance space for performance, stability and model quality, so this kind of supply model was accepted by the market. Since its establishment, Approaching.AI has insisted on targeting production-level inference standards, continuously invested in original innovation of inference systems, consolidated the engineering foundation, and taken the lead in defining the service concept of high-quality AI Token in the industry.
The "high-quality AI Token" defined by Approaching.AI targets production-level scenarios, providing AI Token services with high intelligence, low latency, high speed, strong stability, as well as large-scale delivery capabilities and controllable overall costs.
As the industrial focus shifts to the implementation of Agent-driven business closed loops, large models are gradually carrying core production-level processes, and the AI inference industry is ushering in a paradigm shift: enterprises have put forward strict production-level rigid requirements for the intelligence level, response latency, output speed, context integrity and service stability of AI Tokens. This change coincides with the high-quality AI Token service concept that Approaching.AI has always adhered to and advocated.
The industry inflection point has emerged, and the Token market has quickly formed a clear quality stratification: The extensive supply of "usable is enough" can hardly meet the demands of production scenarios, and the supply of high-quality AI Tokens that balance multi-dimensional indicators is scarce with surging demand, making it a rigid market demand. As the earliest layout practitioner and deep cultivator of high-quality AI Token services, Approaching.AI relies on its long-accumulated underlying technology of inference systems and engineering capabilities to continuously promote the large-scale delivery of high-quality AI Token services. Approaching.AI has now become a leader in the field of high-quality AI Token services.
Based on "Fewer Models, Deep Optimization, Higher Quality" | Accelerating Large-scale Delivery of High-quality AI Tokens
The continuous large-scale delivery of high-quality AI Tokens ultimately depends on two core propositions: Whether the unit computing power efficiency can be continuously improved, and whether the large-scale supply capacity can grow stably.
To achieve this goal and fulfill the production standards of high-quality AI Tokens, Approaching.AI adheres to the technical route of "Fewer Models, Deep Optimization, Higher Quality", focuses on mainstream large models with clear market demand and wide application, and continuously promotes the engineering implementation of a number of world-first technologies such as "Heterogeneous Inference", "Computing with Memory", "Virtual-Real Isomorphism", focusing on the high-quality AI Token demands of enterprise production scenarios.
"Fewer Models, Deep Optimization" solves the problem of "how many high-quality AI Tokens can be produced by the same batch of computing power", while stable production load and cross-cluster operation solve the problem of "how much time these computing powers are actually in effective working state". "Fewer Models" itself does not naturally bring higher utilization rate. Only when both the unit computing power output and effective utilization rate are improved at the same time, can the unit economy of the AI Token business be established.
The business logic behind this route is very clear: Enterprise-level customers ultimately pay for business results, not for the number of model compatibility.
At present, Approaching.AI has formed a replicable high-quality AI Token factory construction and operation mode: there are not only put-into-production projects with a daily output of trillions of Tokens, but also strategic cooperation and under-construction layouts for regional industrial ecology and leading customers. More importantly, the overall business has crossed the cost line. The large-scale delivery and continuous operation capability of the ATaaS platform has been verified by real performance.
Technological Innovation Reconstructs Domestic Computing Power System | Cost Performance Exceeds the World's Most Leading Computing Power
Enabling domestic computing power to support the production of high-quality AI Tokens has always been the goal of Approaching.AI.
Facing the performance gap between domestic and foreign chips, Approaching.AI has explored a new path to achieve computing power efficiency breakthrough through system-level technological innovation. It has successively proposed a number of innovative technical solutions in the directions of "domestic Prefill-Decode (PD) heterogeneous collaboration", "high-performance heterogeneous KV Cache conversion", "heterogeneous computing power integration", etc., reconstructed the domestic computing power computing mode and inference system, and has been officially put into production in high-standard production scenarios.
In the real production-level environment, domestic chips are enabled to carry trillion-level leading large models, achieve large-scale production of high-quality AI Tokens, and the overall cost performance exceeds the world's most leading computing power.
The heterogeneous collaborative inference technology pioneered by Approaching.AI, on the premise of ensuring model output quality, response speed and service stability, achieves higher effective Token output with lower hardware investment, and stably undertakes the real business demands of leading customers.
This breakthrough has dual pioneering values: at the commercial level, it builds a differentiated core technical moat, opens the growth ceiling for Approaching.AI to build high-quality AI Token factories at scale, and supports the continuous leap of performance and capital value; at the industrial strategic level, it helps to get rid of the technical competition that simply compares hardware parameters, redefines the value scale of domestic computing power, enables the domestic chip assets of different architectures scattered across the country to release sufficient inference production capacity, provides a practical core engineering solution for the country to build an AI inference base that is independent and controllable, with sufficient supply and commercial economy, and further consolidates the security of China's large model industrial chain and the foundation of computing power independence.
Expand Production with Technology | Achieve Tenfold Performance Leap in Half a Year
Continuous independent technological innovation is the core driving force driving the exponential growth of Approaching.AI's high-quality AI Token output.
In the past half year, with the continuous launch and engineering polishing of a number of pioneering technologies, the company's average AI Token production efficiency per unit of computing power has risen month by month, continuing to lead the industry; the daily output of high-quality AI Tokens has increased by more than 100 times, climbing to more than one trillion level, and the business revenue in 2026 has reached nearly ten times of that of the whole of last year.
In terms of industrial pattern, Approaching.AI is promoting the upgrading of AI infrastructure from traditional computing power resource supply to large-scale production and operation of high-quality AI Tokens. Relying on the capabilities of the ATaaS platform and Token factory construction and operation, the company has cooperated with industrial partners to promote more than ten AI Token factory projects with a daily output of one trillion Tokens within half a year, promote the large-scale application of domestic computing power, and jointly build a new AI Token supply network connecting computing power infrastructure, large models and industrial applications with local state-owned assets, industrial leaders and ecological partners.
Regionalized Infrastructure Layout
Focusing on Henan, Hangzhou and Hainan, the company promotes the construction of high-quality AI Token production and service network by combining regional resources and industrial demands. In Henan, it promotes joint operation with local state-owned assets, and collaboratively carries out Token production, sales and customer consumption; the East China regional headquarters is located in Qianjiang Century City, Hangzhou, planning to build a ten-thousand-card level Token factory to serve the enterprise-level inference demands in the Yangtze River Delta; in Hainan, it explores a cooperation mode where local state-owned assets are responsible for construction and Approaching.AI provides platform and operation services. Through the tailored construction and operation mode, it promotes the conversion of computing power investment into sustainable Token supply capacity, and drives the development of local operation capabilities and industrial ecology.
Industrial Ecology and Co-construction & Co-operation
The company has reached ecological cooperation with SenseTime, Parallel Technology, Hongxin Electronics, Wangsu Science & Technology, Moore Threads, MUSA, Iluvatar CoreX and other enterprises, to promote the construction and operation of high-quality AI Token factories, explore high-quality AI Token production technical solutions based on domestic chips, and jointly build an independent and controllable domestic AI computing power ecology. It cooperates with SenseTime to promote the construction of a 10-trillion-level daily output Token factory, and the co-built production line has the production and stable consumption capacity of one trillion Tokens per day; it cooperates with Parallel Technology to build a one-trillion-level daily output Token factory, opening up the full link of computing power infrastructure, inference optimization, standardized Token supply and industrial application; it has completed technical adaptation and pilot implementation for the two Token factories of its subsidiary Suihong Huachuang in Wuxi and Xiamen with Hongxin Electronics; it cooperates with Wangsu Science & Technology, combining its global distributed edge resources and network capabilities, to carry out inference optimization and intelligent scheduling collaboration, so as to improve the production efficiency and delivery stability of high-quality AI Tokens.
The continuously released technological efficiency dividend has promoted the business and capital to enter a high-speed growth cycle simultaneously. The financing rhythm is in sync with performance growth, with multiple rounds of closing implemented in an orderly manner within this year, maintaining a rhythm of one round every two months on average, and the scale of fundraising in each round continues to increase month on month; the total accumulated fundraising amount in 2026 has reached RMB 2.4 billion, receiving continuous heavy investment from state-owned assets, industrial capital and market-oriented institutions. After the series of financing is completed, the shareholder structure has become more diversified and mature, gathering local state-owned assets, financial capital, leading industrial groups and first-tier market-oriented institutions, providing solid support for subsequent regional computing power co-construction, domestic heterogeneous ecological collaboration and large-scale industrial delivery.
Top Scientific Research and Open Source Foundation | Build System-level Long-term Engineering Barrier
Behind the commercial progress is Approaching.AI's continuous accumulation in scientific research, open source engineering and enterprise-level production delivery.
From cutting-edge technological exploration to strict engineering implementation, the core team with both academic depth and industrial experience runs through the whole process: the company's management team consists of Dr. Ai Zhiyuan, Founder and CEO, Dr. Wu Wenjie, President and CFO, Chen Xianglin, CTO, etc., who jointly promote technology R&D, capital collaboration, industrial cooperation and project delivery; in terms of technical research, Academician Zheng Weimin serves as the Chief Scientific Advisor, and Professor Wu Yongwei serves as the Chief Scientist, providing academic and technical support for the evolution of the company's underlying technology; Associate Professor Zhang Mingxing serves as a director, guiding technical strategy and key R&D work, and promoting the connection between cutting-edge research and engineering practice.
In addition to academic research, the core team of Approaching.AI has long participated in the construction of the large model inference open source ecology. The team took the lead in launching KTransformers, and participated in the initiation and co-construction of Mooncake, providing engineering verification environments for relevant research for different models, hardware and load conditions, and also promoting the team to continue to explore in the directions of "full-system heterogeneous collaboration", "computing with memory", "virtual-real isomorphism" and other directions.
Starting from academic research and open source engineering, the relevant technical capabilities are further precipitated into the ATaaS platform, and continuously tested in the real business load of AI Token factories. Adhering to the concept of "more efficient computing power, more inclusive intelligence", Approaching.AI takes real demands to drive production investment, and tests technical capabilities through long-term delivery. It is outputting a standardized production system of high-quality AI Tokens through the ATaaS platform and AI Token factories, reshaping the AI inference supply paradigm, and empowering the production-level intelligent upgrading of thousands of industries.
Ai Zhiyuan, CEO of Approaching.AI, said: "In the race of converting computing power into high-quality productivity, the real scarce moat is not scale, but unit energy efficiency and commercial delivery capability. Over the past half year, Approaching.AI has proved the huge commercial potential of the "Fewer Models, Deep Optimization, Higher Quality" route with more than tenfold business growth. The continuous capital increase is a recognition of Approaching.AI's business delivery performance. Looking to the future, Approaching.AI will rely on this round of funds to accelerate capacity expansion, deepen the implementation of domestic PD heterogeneous solutions, support customers' large-scale applications with more stable, more efficient and more cost-effective high-quality AI Token services, promote domestic heterogeneous resources to fully enter real production scenarios, and jointly build a diversified, stable and sustainable AI inference supply system with industrial partners."