Amodei wants to slow the AI frontier. Africa is still trying to reach it

Amid a chorus of alarmist calls from Silicon Valley, Dario Amodei writes that the pace of AI progress must be throttled. He isn’t demanding a halt; he wants a deliberate slowdown so safety research and governance can catch up. Meanwhile, in a continent that is still hunting for the hardware and capital to build its own AI factories, the same narrative feels like a distant echo. Africa’s struggle is not to outpace safety but to catch up to the very frontier it is meant to inhabit.

The continent’s potential is not in algorithmic genius but in the raw materials that feed the silicon beast. The International Energy Agency reports that Africa supplies 75 % of global manganese, 70 % of cobalt and nearly 20 % of copper. The Democratic Republic of Congo alone produces 75 % of the world’s cobalt and 13 % of mined copper. Yet these minerals are extracted, shipped, and processed far from where the next generation of AI models will be trained. The supply chain is a one‑way street: African resources feed the global tech machine, but the machine’s gates remain locked.

This is the paradox that Jide Awe of Jidaw.com points out: Africa has the physical ingredients, yet it sits at the bottom of the value chain. The world still treats African nations as consumer markets and mineral sources rather than strategic partners. The continent’s infrastructuredata centres, fibre networks, cooling systems, batterieslags behind the needs of a hyper‑connected, data‑driven AI ecosystem. Without the electricity grid to power massive GPU farms, the continent cannot host the compute that fuels frontier models.

Amodei’s plea for a slower, safer AI frontier is framed in terms of recursive self‑improvement and autonomous agents that could outpace human control. His audience is a global tech elite that already controls the majority of computing resources. For them, “pacing” is a technical adjustment. For Africans, the issue is access. The continent does not yet have the capacity to even field a single large‑scale model; it is still trying to secure the very hardware that would make that possible.

When Amodei calls for continuous third‑party evaluation and democratic coordination, he assumes that the actors involved have the means to participate. But African innovators are still fighting for basic connectivity. The mismatch is stark: one side is drafting safety guidelines; the other is drafting survival strategies for electricity and bandwidth.

If the global AI race continues as it has, the divide will deepen. The continent will become a silent supplier of minerals and a passive consumer of finished AI products. The risk is not only economic but existential: without a voice in the design of safety protocols, African systems may inherit biases or vulnerabilities that reflect the priorities of distant labs. Amodei’s suggestion that “democratic governments” coordinate on AI risks is a nice idea, but it presumes that African governments are already part of the conversationa presumption that the current reality disproves.

Moreover, the call for international cooperation is hampered by geopolitical realities. The very countries that control the supply chainCongo, Zambia, Zimbabweare governed by regimes that may prioritize export revenues over domestic technological development. Without a concerted push for local capacity building, the continent will remain a passive node in a network that it does not control.

If nothing changes, by 2030 Africa will still be a raw‑material supplier while the rest of the world has already moved beyond the first wave of generative AI. The continent will have missed the first opportunities to embed its own cultural narratives into mainstream AI models, a gap that will widen the cultural bias already present in global AI outputs. That is the real danger: a world where AI reflects only the perspectives of those who control the hardware.

For Nigerian filmmakers, Canadian podcasters, and UK-based content creators, the stakes are concrete. Without access to affordable, high‑performance computing, their creative outputs will be limited by the tools they can afford. They will have to rely on foreign platforms that may not understand local contexts. The result is a cultural homogenization that erases the very diversity that makes African stories compelling.

The call to slow AI progress is, in a sense, a call to democratize it. But until Africa secures its own supply chains and infrastructure, the slowdown will be a luxury for a few, not a safety net for all.

The next decade will decide whether Africa can turn its mineral wealth into a technological foundation. If it does, the continent could become the next hub for ethical AI development. If it does not, it will remain a silent supplier to a world that may one day outpace it in ways it cannot even comprehend.