Zhitong Finance App learned that on October 15, Bairong Smart-W (06608) disclosed a discloseable transaction announcement. The company signed a partnership agreement with CICC Capital and two limited partners through a comprehensive affiliate, Beijing Digital Plus, to jointly initiate the establishment of an industrial fund with an initial size of 500 million yuan. Among them, Digital Science Premium contributed 195 million yuan, accounting for 39%; CICC Capital acted as general partner, executive partner and fund manager. The fund has an initial life of 7 years and an investment period of 4 years. The investment direction covers fields such as AI, underlying computing power, advanced manufacturing and intelligent hardware.
The strategic coordinates of this investment have been clearly stated in the announcement: AI-centered and AICC as the anchor, using managers' expertise and partner networks to expand opportunities to engage with AI startups and seek potential business collaboration and long-term returns.
In line with Bairong's current business progress, this arrangement is more worth understanding as an extension of the AI main business: AICC is verifying commercial value through real customers, real tasks, and continuous expansion, while industrial funds have added a capitalization path for the company to connect with related technologies, products, and application scenarios. Expanding from “doing AI” to “doing AI+ investing in AI,” Bairong is exploring growth methods supported by AICC's endogenous growth and AI industry chain investment.
I. The growth of AI strategies as an extension
Judging the nature of an industrial investment is never about how much money it is, but on how closely it fits into the main business. The strategic nature of this deal is hidden in three core details.
First, the investment is aimed at completely anchoring the main industry chain. The fund's investment direction revealed in the announcement is leading and high-quality growth stage enterprises in the digital economy and new productivity-related industries, covering AI, underlying computing power, advanced manufacturing and intelligent hardware. This completely coincides with the company's strategic positioning of “AI as the center and AICC business as the anchor”. It is not a blind diversification, but an active extension upstream and downstream along the industrial chain.
Second, the investment ratio binds the industry's right to speak. The 39% share of the investment is by no means a simple financial allocation. Corresponding to the governance structure, Digital Science Premium has the right to appoint at least 1 member to the Investment Decision Committee. The judgments and demands of the industry are included in the fund terms rather than at the level of verbal cooperation.
More importantly, investment boundaries are clear. Funds are clearly not allowed to invest in listed companies (including fixed increases, strategic placement, mergers, acquisitions and restructuring shares, except when companies hold untransferred shares after listing), and cannot participate in convertible instruments, deposit arrangements, or quasi-equity investments; in principle, they only invest in unlisted companies, which means that all of the capital is anchored in early equity and hard technology assets, and there is no room for flow to the secondary market or credit-like assets.
In other words, this fund is essentially the “industrial touch” of Bairong Intelligence — connecting upstream technology and assets with capital to feed back the competitiveness of downstream application scenarios. It's not a second-hand business outside of the main business; it's an epitome growth of AI strategies.
2. The growth engine has passed production verification
The motivation for two-wheel drive has never come from investment, but from the stability of the main business itself. The 2026 mid-term results are a typical answer during Bairong's transition period — the new growth curve has entered the phase of amplification.
The anchor point of volume falls on the AICC business. In the first half of the year, the Group's AICC business revenue was 166 million yuan, an increase of 51.7% over the previous year. At the same time, R&D investment is also trending at the same time. R&D expenses increased by 30.7% year-on-year to 394 million yuan in the first half of the year, and resources are clearly skewed towards pan-industry AI transformation. Revenue expansion goes hand in hand with increased investment, and the new growth curve is entering a phase of continuous expansion.
However, what has more weight than the growth rate is that the methodology replicated across industries has been verified at the production level — the logistics scenario is the first to run through the entire process. In July, the number of calls handled by the benchmark customer silicon-based customer service in a single day rose from less than 1,000 calls at the beginning of the launch to over 15,000, with a maximum call duration of over 20 minutes. The real significance of this case is that AI is beginning to enter a complex, open, and continuous real service process: the user expression, problem content, and emotional state of logistics calls vary greatly, and the conversation path is difficult to fully preset. Being able to handle tasks stably in this environment is a real test for silicon-based customer service agents on the production side.
The brokerage scenario quickly verifies the customer expansion logic. As of August, 13 brokerage customers have signed contracts and are in the contract process; a leading brokerage firm expanded its silicon-based customer service deployment seats from about 30 seats at the beginning of the year to about 210 seats. For enterprise-level AI companies, existing customers actively expand deployment after use is the hardest sign of product value — initial access is only POC verification, and continuous expansion means that customers are willing to hand over more core business processes to AI.
What further illustrates the potential for growth is the new scenario data: AICC's revenue from the new scenario was 11.26 million yuan during the period, an increase of 195% over the previous year, covering six major industries, including logistics, aviation, brokerage, banks, operators, and car insurance.
“Can AICC capabilities be replicated across industries”? This issue has been verified by two benchmark samples from logistics and brokerage firms. This is the actual basis for industrial chain investment to be implemented at the moment: only if the current downstream scenario passes through the actual production environment can upstream technology, computing power, and hardware investments have a real impact, and synergy effects are not empty words.
3. Why did Bairong run through AICC? There is a three-story moat here
Enterprise-level AI customer service is widely regarded as one of the top three high-deterministic AI implementation scenarios. It is juxtaposed with programming and enterprise search. The core features are documentation of work content, manual judgment of results, and quick investment results. There are quite a few racetrack players, but Bairong was able to run the benchmark case by relying on three layers of barriers that are difficult to replicate.
The first layer is a self-developed large-scale model system. The company has formed a complete matrix of the BR-Voice multi-modal speech base model, the BR-LLM-Proactive Intent Recognition and Active Guidance Basemode, and the BR-Vision-Doc document understanding base model. Among them, BR-Voice achieved a test accuracy rate of 94.84% in the “Multilingual Conversational Speech Understanding” task of the INTERSPEECH 2026 MLC SLM Challenge in July 2026, ranking third in the world and number one in the Chinese industry; all four tasks in the IEEE SLT 2026 related challenge entered the top five in the world, with two oral language understanding tasks ranked second in the world. Unlike GM's large-scale model route, the scale of Bairong's model parameters is relatively restrained, which directly leads to lower customer calling costs and higher commercial implementation efficiency.
The second tier is telephony engineering and delivery capabilities. AI customer service has never been a competition of single-point model capabilities, but rather system engineering for concurrent stability, interruption processing, emotion recognition, and closed loop work orders. Traditional telephone lines use 8kHz narrowband voice, retaining far less acoustic information than Internet broadband audio, which places extremely high requirements on recognition, noise reduction, interruption processing, and cycle judgment; the production environment also needs to meet the conditions of low latency, stable concurrency, reliable labor transfer, and complete quality inspection. Bairong built a complete engineering system around telecommunication lines, soft switching, real-time audio transmission and business interfaces, and went deep into the customer site through front-line deployment engineers to transform model capabilities into operational business processes. The highly reliable delivery system developed by serving financial institutions all year round is an entry threshold that is difficult for latecomers to replicate in the short term.
The third layer is compliant data assets. The company has continued to serve in the financial industry scene for more than ten years and has accumulated compliant private domain data; at the same time, the financial industry already has high requirements for data accuracy and compliance, and this kind of data is settled under strict standards. Combined with leading customers' industry interests and deep connections with business systems, they have combined to form extremely high replacement costs.
In terms of business model, the company uses a RaaS (pay-for-results) model: billing based on successfully completed call work orders that do not require manual transfers, billing based on number of deployed silicon-based employees, and profit sharing based on the size of the resulting business. Judging from the implemented scenario, the charge for a single successful call is significantly lower than the customer's corresponding labor cost — the customer can directly calculate the input and output account, which is the core reason that the new scenario can complete the POC to volume within half a year.
After visiting a number of local Chinese AI application companies, analysts at Guoyuan International determined in the industry research report “AI Themed Research Summary - The Era of Applied Meat-Eating” released on September 25 that enterprise-level AI applications have moved from “proving that AI can do it” to the stage of “proving that AI can do it fast and cheaply”. The delivery cycle, labor replacement rate, and proportion of mature products will determine who can cut the next round of customer budgets. Meanwhile, Bairong is standing at a critical turning point from demonstration to delivery.
4. Industrial chain investment: growth along the main business boundary, technology and capital collaboration
If AICC is a downstream application layer close to customers and data, then every layer of the upward industrial chain corresponds to actual bottlenecks, as well as to the fund's investment target. For example, in the AI model and application process, this layer is related to multiple rounds of dialogue and the upper limit of the ability to understand. If the fund is laid out here, it will iterate on the vertical model and supplement the intelligence capabilities; in addition, the underlying computing power and infrastructure, it determines inference costs and concurrent flexibility, affecting implementation cost performance and delivery gross profit. The corresponding investment backfeed falls on the reduction in the cost of computing power per unit and the improvement of the gross margin of delivery.
The core value of this structure is that investment is not parallel to another business other than the main business, but rather the natural growth of the main business capacity map. The investee companies obtained real business scenarios and verification channels, and Bairong obtained technology supply and scenario entry, forming a positive cycle of two-way empowerment.
It is worth noting that the investment direction of the fund was not chosen out of thin air, but rather highly synchronized with the company's internal technology research and development route. Bairong's interim results announcement clearly mentioned the development of full-duplex voice models and technical verification for personalised intelligence and wearable devices. This means that investment directions such as smart hardware and advanced manufacturing have a natural connection with the cutting-edge technical capabilities the company is exploring — if technology complements can be formed in suitable projects in the future, the company's voice interaction and smart body capabilities will have an opportunity to be applied through more terminals and scenarios.
The ideal collaboration path is clear: the application side puts forward real business requirements, industry chain partners provide corresponding technology and products, and both parties verify commercial value through the customer's site. For example, customized computing power solutions adapted to business requirements may help improve inference efficiency and reduce unit costs; mature terminals and system partners may greatly shorten the implementation cycle into new scenarios. The fund provides an organized channel for finding these partners, and actual results will gradually be reflected as projects are selected, invested and collaborated.
The fund's manager is CICC Capital, and its gold content is self-evident. As of December 31, 2025, CICC Capital's cumulative subscription management scale exceeded 680 billion yuan. It is the leading comprehensive private equity investment platform in China. It also acts as the management agency for the Beijing-Tianjin-Hebei Venture Capital Guidance Fund with a target size of 50 billion yuan. The investment direction clearly covers cutting-edge fields such as general artificial intelligence and physical intelligence. The background of the other two limited partners is also critical: one is a government investment fund wholly owned by the Chaoyang District State-owned Assets Administration Commission, and the other is a government-guided fund that ultimately holds more than 30% of the rights in the Beijing Municipal State-owned Assets Administration Commission system.
This brings dual core values to the fund. The first is professional guarantee — GP's mature due diligence, valuation, and post-investment management system, which is difficult for a single industry to build; second, project sources and policy collaboration — Beijing has set up a 10-billion municipal artificial intelligence industry investment fund, and Chaoyang District is also promoting the “Artificial Intelligence +” campaign, clearly proposing to explore CVC cooperation with chain owners to establish vertical industry segmentation sub-funds.
The seven-year initial survival period and four-year investment period are not arbitrarily set time windows, but rather closely match the growth patterns of the AI industry. Accordingly, the pace of investment was also smoothed out. In principle, fund capital recovery is divided into 5 installments, corresponding to a 4-year investment period. The capital is gradually invested rather than a one-time expense, and the impact on current cash flow has been greatly smoothed out.
5. Value Prospects: Looking at strategic signals in the short term, ecological collaboration in the long term
Back to the original question: Why is this fund being launched now? The answer lies in the resonance of three time windows.
First, the main business has reached a point of outward expansion. AICC has completed production-level verification for cross-industry replication. The logistics and brokerage samples prove that the methodology can be migrated; at this point, only by connecting computing power, hardware, and model capabilities externally can collaboration have a real impact.
Second, the industrial window is on the same frequency. Domestic companies' AI budgets are moving from pilot demonstration to large-scale implementation. Early asset valuations in the primary market, the Beijing-Tianjin-Hebei 50 billion guidance fund, the Beijing-Tianjin-Hebei 50 billion guidance fund, and the Chaoyang District CVC mechanism will centrally release policy and market dividends in the same time window. Using a limited capital of 195 million dollars to connect to this system, the purchase was not only a financial return, but also the source of the project and the ecological status of the industry.
Third, the pace of investment matches the industrial cycle. The fund is funded in 5 installments and lasts for 7 years. The capital is gradually invested according to the investment schedule, which matches the long growth cycle of the AI industry, and also leaves room for subsequent project screening and collaborative implementation.
From an industry perspective, Guoyuan International determined in AI application industry research that the entire circuit is moving from “proving that AI can do it” to the stage of “proving that AI can do it quickly and cheaply”. 2027-2028 will be a verification window for large-scale profits for leading companies. In this window, a single self-development ability is no longer enough; building a two-wheel ecosystem of “self-research+investment” is becoming a common choice for leading players. The essence of Bairong Smart's 500 million industrial fund is to replace upstream technology supply with scenario barriers in the main business and early growth dividends with accurate industry judgments — the former consolidates current delivery competitiveness, and the latter opens up a long-term growth ceiling.