South Africa has AI leverage its draft policy still fails to spend

South Africa’s AI debate is not just another case of a developing country trying to catch up with regulators in Brussels, Washington or Beijing. The argument carried by IEEE Spectrum is sharper: South Africa has leverage that most countries do not, but its draft AI policy still treats that leverage as background rather than as a strategic instrument.

The first lever is physical. South Africa holds about 88 percent of global platinum-group metal reserves. Those materials are not GPUs, models or data centers, but they sit inside the wider material economy that makes AI infrastructure possible. A country with that kind of resource base weakens its own hand if it frames AI policy only around ethics, innovation and generic risk management. The broader importance of these minerals is visible in USGS platinum-group metals data, which makes clear that this is not a decorative footnote to the AI economy.

The second lever is infrastructure. South Africa has the continent’s largest data center market, and cloud providers have already made the country a regional anchor. The AWS Africa region in Cape Town and Microsoft Azure’s South African geography are not just pins on a technical map. They represent contracts, local capacity, public-sector dependencies and long-lived architecture choices that become difficult to unwind once they are embedded.

That is why the critique of the draft policy matters. If AI policy is reduced to adopting models, encouraging startups and writing responsible-use principles, it misses the harder question: who controls the compute layer, what terms govern public-sector procurement, how much local industry enters the value chain, and what South Africa demands in return when global platforms want its market, data and institutional reach.

The geopolitical layer is also central. Competition between American and Chinese technology companies over AI infrastructure is not playing out only in high-level strategy papers. It is being built into systems that may support public administration, education, health care and state services across the African continent. South Africa is not merely a technology consumer in that contest. It can be a negotiator. But leverage only matters when the negotiator knows what terms it wants.

On that reading, the draft AI policy should connect governance with industrial policy more directly. That means stronger rules for public procurement, clearer transparency around suppliers, local obligations tied to major infrastructure investments, domestic skills development and protection against locking public institutions into external platforms with limited room to maneuver. Ethical principles are necessary, but they are thin if infrastructure, data and minerals are treated as separate policy lanes.

The real danger is not that South Africa publishes one more AI document too late. It is that the country misses the moment when it actually has bargaining power. Once contracts, data flows and public systems settle around someone else’s architecture, later negotiations become weaker, more expensive and politically harder. This is why the story is not just about AI regulation. It is a test of whether a country with rare physical and infrastructure advantages can turn them into digital sovereignty.