Status Quo

The United States and China lead global AI development and deployment, competing for technological advantage. Despite playing a critical role in the global AI supply chain, much of the Global South primarily acts as an AI consumer. Leaders across the Global South want to change that — for example, key players like India and Brazil are working on establishing technological sovereignty and building digital public infrastructure (DPI). In another case, workers in Kenya are working to rectify human rights violations and exploitation early in the AI lifecycle in the dangerous, low-paid jobs that are typically outsourced to the Global South. While the U.S. and China battle over frontier AI capacity, the Global South and middle powers want to make sure that AI benefits them and doesn’t just increase dependencies on foreign market leaders.

Inciting Event(s)

A U.S. government decision to restrict foreign-national access to leading frontier AI models illuminates a perilous reality for middle powers and the Global South: Reliance on a foreign AI stack may expose these countries to the risk of sudden access restrictions for frontier AI, creating new points of leverage and vulnerability. In an effort to avoid reliance on powerful foreign AI suppliers, the EU and Global South come together to build an independent multinational coalition for AI development and deployment that prioritizes sovereignty, access, and growth, adopting the DPI model that is widely utilized for technology adoption across low- and middle-income countries.

Mechanism(s)

To build this coalition, the Global South aligns with likeminded countries from the Global North on baseline environmental, social, and governance (ESG) standards that jointly address dependency risks and boost growth and development in the Global South. The North-South coalition for an alternative AI stack establishes legal frameworks for fair compensation and working conditions, particularly in data sourcing and labeling, mineral mining and extraction, and AI upskilling initiatives. Rather than replacing national authority, the coalition coordinates interoperable standards, pooled investment, procurement, knowledge-sharing, and mutual-recognition arrangements among participating countries.

Participating countries contribute to different parts of the AI stack in line with their capabilities, priorities, and comparative advantages, with the coalition focusing on safety and trustworthiness guardrails for child safety; stronger data ownership and privacy rights; protections from advanced AI-enabled cyberattacks; regulation of critical mineral mining centers; and redistributing economic gains among workers and local populations. Private sector investments from the Global North extend beyond physical supply chain assets to education and workforce development, boosting AI literacy and enabling local communities to own and operate context-specific AI tools, and establishing trust between Global South consumers and workers and big tech companies.

Outcomes

In this scenario, the Global South deepens in-region networks and partners with aligned Global North countries to establish AI as digital public infrastructure (DPI) and reduce foreign dependencies at every stage of the AI lifecycle.

This alternative stack spans critical mineral extraction and mining, compute infrastructure, models, applications, and independent evaluators and auditors. Coalition members co-develop native risk-based standards for transparency, traceability, and environmental and social responsibility. These standards are backed by verifiable systems, such as cross-border mechanisms to track AI-generated content provenance and certify the origin and conditions of minerals and hardware components.

The coalition, focused on decentralization of AI assets, is built on two pillars: industry and governance.

The industrial pillar brings together private sector organizations to co-design specialized AI systems with local communities across healthcare, agriculture, and education. Collaborators from the Global North recognize that AI adoption in local markets depends on solving real-world problems with long-term impact, building on their support for Global South countries. The pillar plays into the comparative advantages of participating countries and addresses a critical bottleneck for Global South developers: access to advanced hardware (e.g. semiconductors).

The governance pillar is led by public institutions, academia, and civil society through effective transparency, risk evaluation, and mitigation tools, complementing ongoing processes like the UN’s Global AI Dialogue. Together, they approach AI as a public utility that advances collective prosperity over external geopolitical interests. Public investment, development finance institutions (DFIs), and public-private partnerships fund infrastructure and workforce development.

The establishment of this coalition could provide leverage and market access for the Global South. The Airbus case may provide an example. In the 1960s, in response to American dominance in the aeronautical industry, leaders from Germany and France successfully led a consortium to develop their own regional champion, Airbus. Success for Airbus came largely from state-sponsored R&D, decentralization of manufacturing and assembly, and shared government ownership. A similar framework, applied across Europe and into the Global South, can support the development of greater control and agency over AI for smaller nations. While the political economy of AI differs substantially from aerospace, Airbus illustrates how coordinated public investment, distributed production, and shared governance can help countries build strategic technological capacity.

Another useful model could be India’s approach to developing DPI via Aadhar, its biometric identity system. Aadhar has been massively successful, issuing an estimated 1.4 billion unique user IDs as of August 2026, since its establishment in 2009. The program is widely credited for increasing financial inclusion and improving access to government services.

Next Steps: Achieving This Scenario by 2031

Alternative scenarios are, first and foremost, thought exercises that illustrate different versions of potential futures. The next phase of the Responsible AI Fellowship will build policy and technical playbooks to make these desired futures a reality by 2031. Key to this process is the solicitation of suggestions from partners, community members, and regional experts to identify practical tools and pathways; if you have ideas for how particular parts of this scenario can become reality, please fill out the survey here:

Stimson Responsible AI Fellowship Scenario Feedback – Fill out form

In the meantime, we’ve drafted several potential mechanisms to spark discussion, which are included below.

Policy Options

Setting Baseline Environmental Standards

One potential shared environmental standard would cover measurement of AI’s impact on the environment, in order to set the stage for mitigating said impact. A key challenge for addressing environmental risks and opportunities of AI is the difficulty of measuring and quantifying impact across diverse settings; such a standard would need to provide a flexible, easily translated, yet universal system of measurement as a baseline for policy action. For example, the North-South coalition could develop a holistic metric for environmental impact measuring relative stress on water and energy infrastructure, along with relative pollution to baseline levels in a given setting.

Building a DPI Foundation for AI

Another useful, shared initiative would be a replicable framework for implementing AI as DPI.  This framework could include a set of elements needed to successfully support public AI, such as shared compute infrastructure, public datasets, and repositories of transparent AI models and harnesses tailored to specific applications. The framework could also build on or leverage existing DPI efforts, such as digital identities and digital banking, to authenticate users and build out public AI infrastructure. It would also need to address existing critiques of DPI, such as its potential for systemic exclusion and surveillance of citizens.

Relevant Facts

U.S. and China Dominate AI Supply Chains

High-Income Dominance

High-income countries (HIC) account for 87% of notable AI models, 86% of AI start-ups, and 91% of cumulative venture capital funding in AI start-ups as of July 2025.

Exports and Supply Chains

In 2023, 87% of global cloud computing exports came from the United States with China becoming a regional alternative, hosting 47 high-performance computing (HPC) systems (9% of the global total).

• The $8.3 billion (USD) in exports was allocated as such:

Income Group	Percentage
High-income countries (HIC)	84
Upper-middle-income countries (UMIC)	14
Lower-middle-income countries (LMIC)	2
Low-income countries (LIC)	0

Digital Public Infrastructure (DPI)

India Case Study

In 2009, 400 million Indians were without proof of identity. Through digitization, 1.3 billion Indians (95% of the population) could prove identity by 2022. India Stack accounts for 67 billion digital identity verifications and $14.05 trillion (INR) in total value of monthly real-time mobile payments, with 8.6 billion total monthly real-time payments.

Workers’ Rights

Data Annotators and Content Moderators

Workers in the Global South are subject to sub-minimum wages and traumatic content that can lead to PTSD, depression, and anxiety diagnoses.

Kenya Case Study

In 2023, a Kenyan judge ruled that Meta was the “true employer” of Kenyan workers, despite working with third party hirers. In 2024, the Kenyan Court of Appeals ruled that Meta could be sued for human rights violations in-region. The Data Labelers Association (DLA) is an advocacy group focusing on mental health awareness and skills development. The Global Trade Union Alliance of Content Moderators (GTUACM) is an alliance that opposes what they consider to be big tech companies’ exploitation of workers.

AI Capacity

Government AI Readiness Index 2025

The United States ranks first in AI readiness; China is ranked eighth globally.

AI Preparedness Index 2023

Singapore ranks first in AI preparedness; the United States is ranked third globally; and China is ranked 30^th.