梁国勇 以国际合作弥合人工智能投资鸿沟

Artificial intelligence investment is growing rapidly worldwide, but most developing countries still face weak infrastructure and underinvestment. China has shifted its computing power development from scale expansion to system-level coordination, promoting coordinated layout of computing, energy, and networks through initiatives such as the East-to-West Computing Resource Transfer. Competition in foundation models has shifted from merely pursuing parameter scale to capability, efficiency, and governability, and China's leading models are quickly closing the gap with the U.S. frontier.

Open-weight models represent an important path for Chinese foundation models to expand their influence, lowering barriers to model access and deployment and enabling more people to use AI. The report recommends policies that balance scale and efficiency to guide investment, coordinate computing and electricity infrastructure, and expand the beneficiary base through capacity building so more actors can afford and effectively use AI. International cooperation is key to bridging the investment gap, through multilateral mechanisms for technical assistance and innovative financing mechanisms for digital infrastructure.

AI is reshaping the global economy and industry, and brands need to pay attention to changes in user behavior and consumption trends. China, leveraging its massive user market and rich application scenarios, has created conditions for rapid iteration, providing a data foundation for brands to understand consumer demand. Open-weight models lower the barrier to AI adoption, enabling brands to use AI capabilities more efficiently in product development and marketing innovation.

At the infrastructure level, coordinated planning of computing, energy, and networks will affect the cost and stability of AI services; brands should keep an eye on how data center resource constraints affect AI application deployment. International investment is shifting from single technologies to AI chips, foundation models, and security, so brands should value the overall innovation capability of partners when selecting technologies. The Chinese experience of industrial alignment emphasizes validating and obtaining data feedback in real industry scenarios; brands should actively apply AI in actual business operations to guide capital toward more effective solutions.

International AI investment is growing rapidly. From 2016 to 2025, global cross-border greenfield investment totaled 4,420 projects, with capital expenditure of approximately $430 billion, flowing to more than 120 economies. This creates new growth market opportunities for sellers, especially the demand brought by digital transformation in developing countries. The Chinese experience of industrial alignment suggests that forming validation and data feedback in real industry scenarios can guide capital flows; sellers can use actual operating data to attract AI-related investment or partnerships.

On the policy front, the report recommends guiding investment through policies that balance scale and efficiency, and promoting investment facilitation. Sellers should watch for supportive policies in various countries for AI infrastructure and digital trade. Meanwhile, AI is increasingly becoming a focus of foreign investment security reviews, so sellers should be mindful of compliance risks in cross-border investment or technology cooperation. Capacity building is key to broadening the base of beneficiaries; the report emphasizes skills upgrading so that more actors can afford and effectively use AI. Sellers can strengthen digital skills training to capture the benefits of AI applications. International cooperation brings technical assistance and financing mechanisms to lagging regions, and sellers can explore new business models under multilateral mechanisms.

Competition in AI infrastructure is no longer about single computing devices but about system-level alignment of computing, energy, networks, and data center layout. When advancing digitalization, factories should pay attention to computing synergy and electricity resource constraints, avoiding blind infrastructure expansion. The Chinese experience of factor alignment emphasizes coordinated planning of computing, energy, and networks; factories can rationally plan infrastructure investment for AI applications based on their own energy consumption and digitalization needs.

Industrial alignment is an important aspect of the Chinese experience. Factories can validate AI technology in real production scenarios and use data feedback to guide equipment upgrades and process optimization. The report mentions that open-weight models can lower the threshold for AI adoption, and factories can use open-source models to develop quality inspection, production scheduling, and other scenarios, reducing R&D costs. Meanwhile, international AI investment is shifting toward chips, foundation models, and security, so factories should consider suppliers' innovation capability and security compliance when selecting technology vendors. On the policy side, national coordination of computing and electricity construction can provide factories with more stable digital infrastructure. Factories can actively participate in equipment renewal and intelligent transformation projects and enjoy investment promotion and facilitation policies.

The AI industry is witnessing a surge in investment and accelerating international investment flows. Service providers should focus on clients' pain points arising from constraints on computing power, electricity, materials, and water resources. In some concentrated areas, data center electricity consumption can account for 20%–30% of local electricity use, making energy-efficiency optimization and sustainability solutions a key opportunity.

China's computing power development is moving from scale expansion to system-level coordination. Practices such as the East-to-West Computing Resource Transfer and computing–electricity synergy are promoting coordinated layout of computing, energy, networks, and data centers. Service providers can offer planning advisory and integration services centered on system-level synergy. The benchmark for foundation models is shifting toward capability, efficiency, and governability; when selecting models for clients, service providers should emphasize real-world effectiveness and governance-friendliness rather than parameter scale. Open-weight models lower barriers, so service providers can customize industry solutions based on open-source models to reduce deployment costs. Industrial alignment emphasizes validation and data feedback in real industry scenarios; service providers should develop vertically by staying close to clients' business scenarios. As international investment focus shifts from single technologies to chips, foundation models, and security, service providers can expand into AI security assessment and compliance consulting.

AI investment is highly unevenly distributed, with most developing countries facing weak infrastructure and underinvestment. This creates opportunities for platform providers to expand into emerging markets. The Chinese experience of inclusive diffusion emphasizes lowering AI adoption barriers through open weights, architecture optimization, and platform services; platform providers can build open platforms to attract small and medium-sized enterprises (SMEs) and developers.

Policies such as the East-to-West Computing Resource Transfer and the national integrated computing network are promoting coordinated layout of computing infrastructure. Platform providers operating data centers or offering computing services need to pay attention to electricity supply and long-term sustainability. The report notes that AI international investment is becoming more capital-intensive in nature; platform providers can focus on attracting high-quality projects in AI chips, foundation models, and AI security when promoting investment. Meanwhile, AI is increasingly a target of foreign investment security reviews, and platform providers should manage compliance and risk control in cross-border business or when introducing foreign capital. Policies that balance scale and efficiency guide investment; platform providers should follow investment promotion and facilitation policies and provide better infrastructure and financing connections for companies on their platforms. Capacity building and skills upgrading are key to expanding the beneficiary base; platform providers can organize AI training and technical services to lower usage barriers and increase platform stickiness. In terms of international cooperation, innovative development financing mechanisms and multilateral technical assistance will offer potential opportunities for platform providers to collaborate across borders.

AI, as a general-purpose technology, is reshaping the global economy and industry, but the highly uneven distribution of investment and infrastructure risks widening the digital divide. This emerging issue deserves in-depth research, especially the global distribution of capital, computing power, and infrastructure. The report shows that in 2024, the United States, China, and Europe accounted for approximately 45%, 25%, and 15% of global data center electricity consumption, respectively. At the same time, emerging economies are beginning to become destinations for large data center projects, but electricity constraints are becoming more prominent, providing an empirical basis for research on infrastructure investment and sustainability.

The Chinese experience is summarized under four frameworks: factor alignment, inclusive diffusion, industrial alignment, and institutional adaptation. Researchers can use these as theoretical tools for analyzing AI development paths in developing countries. Competition in foundation models is shifting from parameter scale to capability, efficiency, and governability. Open weights have become an important mechanism affecting technology diffusion, and this evolution has reference value for research on industrial policy and innovation ecosystems. On international investment, cross-border greenfield investment comprised 4,420 projects and $430 billion in capital expenditure, with concentrated sources and growing capital intensity. At the same time, AI has become a focus of foreign investment security reviews and is extending to outward investment management. Researchers can explore the balance between security reviews and investment liberalization. The policy recommendations propose guiding investment with a balance of scale and efficiency, expanding the beneficiary base through capacity building, and bridging the investment gap through international cooperation, providing clear directional propositions for related research.