Beyond the U.S.-China Binary: The New Landscape of AI Partnerships in the Global South（来源原文，在新窗口打开）

Introduction

The global AI boom is colliding with a persistent infrastructure and development gap. Although nearly three-quarters of the world’s population is now online, 2.2 billion people remain unconnected, and 96 percent of them live in low- and middle-income countries. Internet use reaches 94 percent in high-income countries but only 23 percent in low-income economies. These disparities include a lack of basic connectivity, affordable cloud services, reliable power, advanced computing capacity, local-language datasets, technical talent, and the institutions needed to deploy AI safely and effectively.

Governments across Global South countries are looking for public and private sector partners that can provide not only fiber and mobile networks, but also data centers, computing power, digital public infrastructure, AI models, skills, and governance support. In the lead-up to the UN-led Global Dialogue on AI Governance in July 2026, the Group of 77, a collection of Global South countries and China, issued a statement claiming, “The next artificial intelligence divide will not simply be about access to technology. It will increasingly be about access to the capacity to shape technology - to develop it, govern it, adapt it to national context and ensure that it serves sustainable development.” In other words, it will be about the AI ecosystem each country cultivates.

Building an AI ecosystem requires investment across several interdependent layers, including hard infrastructure and compute; software, models, and data that can be adapted to local languages and context; workforce development; and governance capacity to regulate, procure, and oversee the use of AI. China remains distinct in its ability to operate across the entire ecosystem, while other countries are offering comparative advantages across elements of the stack: Gulf states in cloud and compute, India in open models and digital public infrastructure, and middle powers in workforce and governance capacity. The increasingly multipolar landscape for technology partnerships in the Global South is giving countries greater choice on how they build their AI ecosystems while creating a more fragmented landscape for standards and governance.

China Moves Beyond Infrastructure to Governance

China has long led digital infrastructure investment in the Global South through its Digital Silk Road initiative. Through Chinese firms, policy banks, and government-to-government relationships, Beijing has spent years supporting telecommunications networks, cloud infrastructure, smart-city systems, data centers, and digital government projects across the Global South. Its value proposition is derived from its ability to connect infrastructure with financing, technical training, commercial vendors, and diplomatic engagement in ways few other countries can match.

That institutional push is reinforced by organic, bottom-up adoption. African developers building tools for local languages and markets are increasingly turning to open-weight Chinese models, including Alibaba’s Qwen, DeepSeek, and Moonshot AI, largely on cost: Developers in Kenya and Uganda report building AI systems at a fraction of the cost of comparable U.S. models. According to Hugging Face, by early 2026, Chinese models accounted for 41 percent of downloads on its open-source repository, surpassing U.S.-developed models.

China is now attempting to exert greater influence over global AI governance with the launch of the World AI Cooperation Organization, an intergovernmental body headquartered in Shanghai and intended to promote cooperation and develop approaches to AI regulation. Its 29 founding members include Belarus, Brazil, Cuba, Kazakhstan, Pakistan, Russia, and Serbia, along with a larger bloc of African and Asian states such as Ethiopia, Indonesia, and South Africa. At a summit in mid-July 2026, President Xi Jinping pledged 5,000 AI training slots for Global South countries over five years and announced new cooperation centers with the Association of Southeast Asian Nations (ASEAN), the African Union, and BRICS. Notably, the summit came less than two weeks after the UN-led Global Dialogue on AI Governance in Geneva, where China also positioned itself as a champion in supporting Global South countries in AI development and cooperation. As AI diffusion across the Global South increases, China is looking beyond infrastructure and instead making moves to influence the rules under which those systems will operate globally.

India Prioritizes Open-Source and Locally Relevant AI

Like China, India has positioned itself as a champion of open digital ecosystems designed to lower the barriers to technology adoption across the Global South. This approach builds on India’s experience with digital public infrastructure (DPI), including its digital ID platform Aadhaar and its Unified Payments Interface. India promoted DPI internationally during its 2023 G20 presidency as a model for building foundational architecture for interoperable digital services. India is now extending a similar emphasis on affordability, openness, and local AI adaptation.

The government-backed IndiaAI Mission is investing in shared compute infrastructure, open datasets, and sovereign AI capabilities, while research institutions are developing the foundational building blocks needed for multilingual AI. Initiatives such as BharatGen, led by a consortium of leading Indian academic institutions, and AI4Bharat at the Indian Institute of Technology Madras are producing open datasets, language technologies, translation systems, and foundation models designed for India’s linguistic diversity. Building on this research, private sector companies such as Sarvam AI have released open-weight multilingual foundation models.

The India AI Impact Summit in New Delhi in February 2026 reinforced this vision on the international stage. As the host of the first summit in the series to be held in the Global South, India explicitly framed the meeting around broadening participation in the AI economy. The AI Impact Summit Declaration emphasized equitable access to compute, open innovation, multilingual AI, capacity building, and international cooperation to ensure that developing countries are active participants in the AI revolution.

The Gulf Bets on Compute

Saudi Arabia and the United Arab Emirates are advancing technology partnerships centered on one of AI’s scarcest resources: compute. Leveraging abundant energy, sovereign wealth, and partnerships with leading technology companies, both countries are investing heavily in hyperscale data centers, AI cloud infrastructure, and advanced computing capacity.

Regional data center capacity is projected to grow from roughly 1 gigawatt in 2025 to more than 3.3 gigawatts by 2030, positioning the Gulf as an emerging hub for AI infrastructure. As AI increasingly relies on cloud-based computing, countries do not necessarily need to build their own hyperscale infrastructure to participate in the AI economy, and Gulf states have made clear that they intend to fill that role for Africa and South Asia. Saudi Arabia’s Public Investment Fund has described the kingdom’s strategy as becoming “the prime backend provider of compute-as-a-service for emerging markets across Africa and Asia,” while the United Arab Emirates has positioned itself as an “AI bridge” linking Western technology and Gulf capital to the wider Global South, combining sovereign investment with partnerships involving firms such as Microsoft, NVIDIA, Oracle, and OpenAI. However, the most ambitious attempt to date, a $1 billion Kenya data center project announced by G42 and Microsoft in 2024, stalled in 2026 after Kenya’s national grid proved unable to support it and the two sides failed to agree on payment terms, underscoring that converting the Gulf’s domestic buildout into an actual compute-export business for lower-income markets has proven harder than the stated ambition suggests.

Middle Powers Work Through Multilateral and Regional Institutions

A growing number of middle powers are amplifying their influence by working through multilateral and regional institutions to shape how digital technologies and AI are governed and deployed across the Global South. By leveraging organizations such as the United Nations, the World Bank, and regional development banks, countries including South Korea, Italy, and Singapore are extending their technical expertise and policy priorities.

South Korea has become one of the most active middle powers in scaling its digital expertise through multilateral platforms rather than individual bilateral projects. The Korea–World Bank Global Digital Knowledge Center, launched in December 2025, functions as the World Bank’s sole digital think tank, channeling Korean expertise into research, capacity building, and technical guidance for developing countries. In May 2026, Seoul went further, declaring a “Global AI Hub” vision jointly with nine international organizations, including the International Labour Organization, UN Development Programme, UN Children’s Fund (UNICEF), World Health Organization, and International Telecommunication Union, and five multilateral development banks, including the World Bank, the Asian Development Bank, and the Inter-American Development Bank.

Italy has channeled its Global South AI engagement through the UN Development Programme, codesigning the G7-endorsed AI Hub for Sustainable Development with the UN agency and launching it in Rome in 2025. The hub connects African governments and startups with compute resources, financing, and technical expertise. Spain has taken a similar approach with the AI Governance for Humanity Lab, launched in Valencia in May 2026 under the UN Office for Digital and Emerging Technologies with Spanish government support. This model allows countries to leverage the expertise and global footprint of multilateral institutions to expand their reach and exert influence over regulatory frameworks, procurement practices, and the design of digital systems.

Regional institutions provide another pathway for middle powers to scale their influence. Singapore has used ASEAN to advance regional approaches to AI governance, chairing the ASEAN Working Group on AI Governance and helping develop the ASEAN Guide on AI Governance and Ethics. It has also promoted its open-source AI Verify testing framework and AI assurance approach internationally. In Latin America, Chile’s National Center for Artificial Intelligence leads Latam-GPT, a collaborative effort involving governments, universities, research institutions, international organizations, and private sector partners to develop an open language model reflecting Latin American languages and contexts. The Chilean government has helped expand the initiative through partnerships with governments including Brazil.

Middle powers have also recently called for greater cooperation on safety and global governance. At the UN General Assembly in September 2026, Finland and Norway led a “Call for Control of Frontier AI Models,” initially backed by leaders from countries including Canada, Kenya, Singapore, South Africa, Spain, Turkey, and the UAE. The statement called for common safety standards, shared reporting of serious AI incidents, greater access to independent evaluation and scientific expertise, and exploration of an international institution capable of setting standards and verifying compliance. The initiative illustrates how countries outside the United States and China are increasingly seeking to collectively shape the rules governing advanced AI systems.

U.S. Focuses on Commercial Interests

Today, the United States is largely absent from the AI competition as a development actor, even as U.S. firms remain dominant commercial participants. U.S. hyperscalers have made significant investments in emerging markets, particularly where demand for cloud and AI services is growing. Microsoft has committed $1.7 billion to cloud and infrastructure in Indonesia and $2.2 billion in Malaysia. In Mexico, Amazon Web Services announced plans to invest more than $5 billion in its cloud infrastructure over 15 years.

The U.S. government is also seeking to expand adoption of U.S. technology through the American AI Exports Program, which packages U.S. chips, models, cloud services, and other components of the AI stack for overseas markets, backed by financing and other government support. At the same time, strategic competition with Beijing has led Washington to restrict access to advanced chips and advanced AI models, with those restrictions shifting as policymakers balance national security concerns with the desire for global market share. Despite the priority placed on competing with China, these efforts remain primarily focused on exporting U.S. technology, rather than building the infrastructure, local-language data, or institutional capacity that would allow lower-income countries to absorb the U.S. AI stack.

The consequences are visible on the ground: Developers across Kenya, Nigeria, and Uganda are increasingly building on cheap, capable Chinese open-weight models rather than U.S. ones, citing cost and language performance. While many smaller markets lack sufficient demand to attract significant private sector investment, the United States could respond by better coordinating existing development finance and technical assistance, along with bilateral and multilateral partnerships to invest in infrastructure and capacity development. Regional investments could also help aggregate demand in smaller markets and make investments more commercially viable.

A Multipolar World

Much of the AI debate is framed as a binary contest between the United States and China, but that framing is evolving. This is evidenced in Kenya, which has worked with Chinese firms on digital infrastructure, U.S. hyperscalers on cloud services, Gulf investors on compute investments, India on locally relevant applications, and the multilateral system for capability building and support. Rather than choosing a single technology partner, countries can draw on different partners to fulfill a wide range of needs.

This plurality gives countries more leverage, but it also means AI governance norms and technical standards will be set through a patchwork of competing bilateral, regional, and multilateral partnerships. For U.S. policymakers, the challenge is not simply to compete with China or other emerging technology partners, but to offer a compelling model of partnership that responds to countries’ own priorities. Pairing commercial offerings with investments in connectivity, power, skills, local-language data, and governance capacity could make U.S. technology more accessible and relevant in these emerging markets.