The real challenge of AI for the Global South is not a race to the frontier model, but a lack of collaboration. Only through partnerships between app developers, compute providers, and countries are AI solutions that can change lives at scale.
Author: Nidhi Singh
The AI environment around the world has taken two different paths. On one hand, America and China are competing with each other to remain at the forefront, while on the other hand, the countries of the Global South face a completely different challenge. This challenge will not be solved by joining the same race, but by working together to find ways to use AI that can change people’s lives.
The GTS Innovation Dialogue, organized by Carnegie India in collaboration with the Ministry of External Affairs in December 2025, revealed a startling fact: The biggest hurdle for AI creators to implement AI solutions at scale is not computing power, but lack of mutual collaboration. This information should completely change the way we think about AI development in emerging economies.
lack of cooperation
Today the entire system of AI is divided into many levels. App developers work alone, often repeating the same tasks and re-creating solutions that already exist elsewhere. A maternal health AI tool built in Kenya may solve the same challenges as a tool built in rural India, yet the two teams never connect. This is not just a waste of resources, but a waste of limited resources that innovators in the Global South cannot afford.
Real success comes when AI app developers learn from each other, synergize their approaches, and build on each other’s work. When agricultural AI developers collaborate across borders, they create solutions that are more robust, better suited to local needs, and easier to implement in different environments. On the contrary, working alone will limit us to scattered pilot projects that will never reach the larger population.
The gap between creators and computing power
An equally large gap exists between those creating AI applications and those providing computing infrastructure. Hyperscalers and compute providers often have little understanding of what truly profitable AI applications require. At the same time, app developers often don’t know what infrastructure is available and how they can best use it for their specific AI use.
This lack of coordination has serious consequences. India’s most impactful AI applications rely mostly on small language models and niche models, rather than large LLMs with enormous computing power. Hybrid architectures combining edge computing and cloud infrastructure better match actual deployment needs than one-size-fits-all approaches. Yet, this information is rarely included in infrastructure planning.
What’s needed is true partnership: Compute providers work closely with app developers to understand real-world needs, optimize architectures, and jointly develop deployment strategies. No single party alone can provide what is needed for sustainable deployment of AI (e.g. data infrastructure, multilingual capabilities, security measures, financing models). This is possible only when the entire ecosystem works together.
Strategic need (middle powers)
For developing countries, the choice is clear: they cannot and should not try to compete in the race to the frontier model dominated by the US and China. The capital, technical talent and computing resources required to compete are beyond the reach of most countries.
But this is not a disadvantage; This is a chance to forge a different path. The real strength of the Global South lies in building applications: that is, tailoring AI solutions to local languages, contexts and challenges that frontier models often ignore. The focus here is not just on building foundational AI models, but on applying technology and AI solutions to solve problems for our larger population. The way forward is for the countries of the Global South to work together with each other.
Imagine a network where AI innovations related to agriculture in Sub-Saharan Africa inform deployment in Southeast Asia. Where health solutions developed in Latin America accelerate progress in South Asia. Where education applications tried in one region serve as a model for others. This is not just a fantasy; What is needed for this is present in platforms like India’s Bhashini, which shows how digital public infrastructure can facilitate access.
From credit to contract
Current AI deployment models, which rely heavily on compute credits from hyperscalers, cannot sustain beneficial uses at scale. The Global South needs alternative methods of financing: such as service contracts that pay for results, public-private partnerships that share risks and rewards, and output-based models that link incentives with impact.
These options work at scale only when there is cooperation: such as joint purchasing, shared infrastructure, and coordinated policy frameworks that can create markets large enough to attract sustainable investment.
The India AI Impact Summit
The AI Impact Summit is an important opportunity to strengthen this collaborative approach. The Global South needs to move beyond competing in a race it can’t win and instead focus on winning the race that matters: that is, deploying AI that actually improves people’s lives at scale.
This requires bridging the divide between app developers, building true partnerships between developers and compute providers, and establishing a network of deep collaboration between countries in the Global South. There are technical challenges, but lack of collaboration is the real obstacle standing between pilot projects and impact to the larger population.
The choice is ours: either stay divided or move forward together. For the Global South, cooperation is not just a good habit; This is the only effective strategy to ensure that AI serves the billions of people who need it most.
About the author
Nidhi Singh is a Senior Research Analyst and Program Manager at Carnegie India. His current research interests include data governance, artificial intelligence and emerging technologies. His work focuses on the implications of information technology law and policy from a Global Majority and Asian perspective. She has previously contributed to The Indian Express, The Secretariat, MediaNama and The Hindu Business Line.