Bridging the AI gap: What the Global South AI capacity-building debate at WSIS 2026 means for technology equity

During the 2026 WSIS Forum in Geneva, the China Association for Science and Technology's UN Consultative Committee on ICT hosted a side event titled "Bridging the Intelligence Gap: AI Capacity Building and Knowledge Sharing for the Global South." The panel brought together experts from the UK, Mexico, South Africa, and other countries to discuss a question that is rapidly becoming more urgent than model performance benchmarks: who controls AI, and who gets left behind?

Knowledge point: The three layers of the AI divide

The AI capacity gap is not one problem but three stacked inequalities. The first is infrastructure: training a modern large language model requires thousands of high-end GPUs, massive electricity, and cooling systems — resources concentrated in a handful of countries and companies. The second is data and talent: AI systems trained primarily on English and Chinese text perform poorly on languages spoken by hundreds of millions of people in Africa, South Asia, and Latin America. The third is governance participation: countries without domestic AI industries have little voice in setting the rules that govern AI safety, data rights, and liability — yet they are among the most affected by those rules.

What "capacity building" means in practice

The discussion at WSIS moved beyond the common framing of AI as a development aid tool. Instead, speakers emphasized productive capacity — giving developing countries the ability to train, fine-tune, and deploy their own models, not just consume AI services built elsewhere. Concrete dimensions include: open-source model availability (which reduces the cost of entry), compute-sharing agreements (where countries pool GPU resources), localized training data (building datasets in underrepresented languages), and regulatory knowledge transfer (helping governments write AI laws that are compatible with international standards without ceding sovereignty).

Why this matters now

The timing is significant. In 2024, the UN General Assembly adopted a resolution on AI capacity building co-sponsored by 143 countries. In 2025, China launched the AI Capacity Building Inclusive Plan and the World AI Cooperation Organization. The WSIS side event is part of a broader institutional shift from discussing AI as a technical race to discussing it as a governance and equity challenge. The lesson for anyone following AI: the technology's next phase will be shaped less by model architecture breakthroughs and more by decisions about who gets access, at what cost, and under what rules.