昆仑芯为何需要一场「带货式IPO」？

This article centers on Kunlun Core, the AI chip subsidiary of Baidu, and its "sales-driven IPO" model as it prepares for a Hong Kong public offering, analyzing the model's underlying logic, industry context and potential risks.

1. At its core, the model requires interested investors to purchase a corresponding volume of chips along with their equity subscriptions. The arrangement is currently non-binding, and its core goal is to screen industrial investors with long-term procurement capabilities rather than bringing in purely financial investors.

2. Industry background and current status: AI chips are facing tight supply amid accelerating domestic substitution of computing power. Kunlun Core ranked third among domestic vendors in China's cloud AI accelerator shipments for 2025. It posted 2 billion yuan in revenue in 2024, and projects revenue of 3.5 billion yuan and break-even profitability in 2025, followed by 6.6 billion yuan in revenue in 2026. The company currently needs to expand its external customer base to reduce its reliance on Baidu and prove its independent standing.

3. This model helps Kunlun Core lock in orders to boost its valuation, and aligns with Baidu's transition to an AI infrastructure platform. However, the model will amplify existing risks stemming from the current AI industry bubble, and the company's long-term competitiveness will ultimately depend on its product performance across application scenarios.

This article summarizes the development status and innovative operation strategies of domestic AI chip brands, offering actionable insights for tech brand building, capital operations and ecosystem layout.

1. Current industry trends: Supply of AI chips cannot keep up with demand amid accelerating domestic substitution of computing power. Leading domestic chipmakers all face capacity constraints against strong market demand, with rigid demand for domestic chips across finance, energy, internet, telecom and multiple other sectors, putting domestic brands in a period of growth dividend.

2. Brand building insights: Brands spun off from large tech groups often face market skepticism over excessive reliance on their parent companies. Kunlun Core's "procurement + subscription" IPO model strengthens its positioning as an independent player with broad external customer support, dilutes its parent company association, and proves its independent market competitiveness, offering strong reference value for similar brands.

3. Ecosystem operation takeaways: The industry is already seeing a trend of ecosystem alliance through capital-order binding. Brands can build a positive growth cycle by partnering with upstream and downstream industrial capital to improve valuation and market positioning. That said, long-term brand competitiveness still depends on product performance, and interest binding amplifies industry risks, so the AI bubble should not be ignored.

This article sorts out the market environment, innovative models, opportunities and risks in the AI chip track, offering reference for industry practitioners.

1. Current market opportunities: Accelerating domestic substitution of AI computing power has created broad overall demand for AI chips, with leading domestic vendors all facing capacity constraints. The cloud AI chip market has large room for growth, and chipmakers have strong demand for external customer expansion, creating abundant growth opportunities for track practitioners.

2. A reference innovative operation model: The "capital + order" binding model introduces industrial investors with actual procurement needs during the IPO stage to lock in large orders in advance, creating a positive cycle of "equity subscription → procurement orders → revenue growth → higher valuation". The model has been tested on a small scale and proven feasible by Kunlun Core.

3. Risks to watch for: The AI industry is already showing signs of bubble, and deep interest binding between upstream and downstream players amplifies systemic risk, where a problem in any single link can trigger a chain reaction. For projects spun off from parent companies, it is critical to address market concerns over single-customer dependency early on, and long-term competitiveness ultimately depends on actual product performance across different scenarios.

This article analyzes the demand environment and development trends of China's domestic AI chip industry, offering practical insights for AI chip manufacturing players.

1. Current business opportunities: The AI chip industry as a whole is facing supply shortages amid accelerating domestic substitution and extremely strong market demand. Leading domestic AI chip companies all struggle with insufficient capacity, bringing abundant, stable order opportunities for chip production and manufacturing links.

2. Product development and manufacturing direction: Current AI chips are primarily targeted at cloud computing scenarios, serving finance, energy, automotive, internet, telecom and other industries. Customer procurement volumes range from dozens of cards to tens of thousands of cards, so product design and manufacturing need to accommodate differentiated computing power requirements across different scenarios.

3. Insights for digital transformation: The industry is already moving toward ecosystem development that binds industrial capital and orders together, allowing chip companies to lock in long-term stable orders through this model. Manufacturing players can advance digital production to match this model, plan capacity in advance to meet steadily growing market demand, and remain alert to the risk of overcapacity caused by the AI bubble.

This article outlines the latest developments in the AI chip industry, summarizes core industry pain points and actionable solutions, offering high reference value for AI industry-related service providers.

1. Latest industry trends: A new trend of "circular investing" has emerged in the AI industry, integrating capital, orders and capacity into a single growth flywheel. AI chip companies have started to bring this binding model forward to the IPO stage, and this approach is likely to become the new normal for China's domestic AI chip industry, creating new service demand for providers.

2. Core pain points for key clients: AI chip companies spun off from large Chinese internet groups universally face two core pain points: they need to dilute their parent company association to prove independent market competitiveness to investors, and they need to lock in large orders in advance to support their IPO valuation, while also needing to screen long-term stable partners amid ongoing supply shortages.

3. New service directions to explore: The "procurement + capital subscription" binding model solves multiple pain points at once. Service providers can develop targeted capital matchmaking, order matching and IPO roadshow services around this industrial binding demand to capture new industry opportunities.

This article sorts out new trends in capital operation in the AI chip industry, offering practical insights for industrial and capital platforms focused on hard tech and AI sectors.

1. New demand from industry players: When preparing for an IPO, AI chip companies are no longer satisfied with bringing in purely financial investors. They increasingly prioritize industrial investors with actual procurement demand, to secure stable orders through capital binding while boosting their independent market recognition, which creates new service directions for platforms.

2. New directions for platform operation: In investment sourcing and matchmaking business, platforms can specifically match AI chip companies with industrial clients that have procurement demand, to build an integrated "capital + order" matchmaking model. This model has already been market-tested: Kunlun Core has verified its feasibility, and it enjoys high market acceptance.

3. Risk mitigation guidance: Deep interest binding between upstream and downstream players amplifies industry risks, and the AI sector already has a certain degree of bubble. When operating related business, platforms should proactively disclose potential risks, remind counterparties that long-term competitiveness ultimately depends on product performance, and avoid systemic risk caused by excessive binding.

This paper sorts out the brand-new capital operation model emerging in China's domestic AI chip industry, provides a fresh industry case, and offers high value for research on AI industry development.

1. New industry development: A new trend of deep "capital + order" binding has emerged in China's domestic AI chip industry. Kunlun Core has brought this binding model forward to the IPO stage with its special "sales-driven IPO" approach. Aligned with the current industry backdrop of domestic substitution, this approach is likely to set a precedent for other domestic AI chip companies preparing for public listing and become a new industry norm.

2. Summary of the new business model: This model qualifies as a "circular investing" business model for the AI sector. By binding upstream and downstream capital, orders and capacity together, it forms a positive growth cycle. Similar models are already used by leading global players including NVIDIA and AMD, and the trend of ecosystem alliance is gradually taking shape in China's domestic AI industry.

3. New questions for further research: This model solves the problem of independent market recognition for AI companies spun off from large tech groups, but it also introduces new risks, as interest binding amplifies the impact of industry bubbles. This case also reflects the capital path and transformation logic of spinning off AI businesses from large Chinese internet groups, providing a typical sample for research on the development of China's domestic AI industry.