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

In his September 2026 essay, “We Must Pace the Frontier,” Anthropic CEO Dario Amodei argues that AI capabilities are advancing faster than society’s ability to understand, evaluate and control increasingly powerful systems. His proposal is not to halt AI development, but to slow the pace of capability gains long enough for safety research and governance to catch up.

Amodei points to the potential for recursive self-improvement and increasingly autonomous AI agents, warning that unchecked progress could eventually outrun humanity’s ability to control these systems. He calls for continuous third-party evaluation of frontier AI companies, greater coordination among AI labs and democratic governments, and broader international cooperation on AI risks.

“We must slow the pace at which we improve the capabilities of AI models,” he writes.

It is a consequential argument for countries and companies operating at the technological frontier. But across much of Africa, the AI debate begins from a very different place.

The continent is not racing to build frontier models so quickly that safety systems cannot keep up. It is still struggling to secure the computing power, infrastructure, capital and industrial capacity needed to participate meaningfully in the AI economy.

That creates a contradiction at the heart of the global AI race: Africa possesses many of the physical resources needed to build the AI economy, yet has limited influence over the technology, supply chains and geopolitical arrangements that determine who captures its value.

“It is both,” says Jide Awe, an innovation and technology policy adviser and founder of Jidaw.com, referring to Africa’s infrastructure and funding gaps and its exclusion from major technology alliances. “The world still treats African nations as consumer markets and mineral sources rather than as strategic partners.”

Africa has the ingredients for AI

The importance of Africa to AI begins well before a model is trained.

Artificial intelligence depends on enormous physical infrastructure: semiconductor manufacturing, data centres, electricity networks, fibre, cooling systems and batteries. Those systems require minerals including cobalt, copper, lithium, manganese, graphite and rare earth elements.

Africa is a major source of several of them. The International Energy Agency estimates that the continent supplies around 75% of global manganese, 70% of cobalt and nearly 20% of copper.

The Democratic Republic of Congo is particularly important, producing an estimated 75% of the world’s cobalt in 2024 and about 13% of global mined copper. Zimbabwe, meanwhile, was the world’s fourth-largest lithium producer, accounting for about 9% of global production.

But the critical question is not simply what Africa produces. It is what happens after those resources leave the ground.

China, for example, produced only about 2,000 tonnes of mined cobalt in 2024 but around 48,000 tonnes of refined cobalt, relying on imports for almost 98% of its supply. Chinese companies operated or funded 15 of the DRC’s 19 cobalt mines.

The same pattern is emerging in lithium. Zimbabwe exported about 1.014 million tonnes of spodumene in 2024, rising to 1.128 million tonnes in 2025, with most going to China for further processing.

The structure is consistent: Africa often occupies the extraction end of the chain while other countries control more of the processing, manufacturing, trading and technology layers.

The IEA estimates that Africa currently captures less than 1% of the value generated by manufacturing clean-energy technologies and their components, despite supplying large shares of the minerals required for these technologies.

That is also an AI problem.

“Meaningful participation means going up the global AI value chain,” Awe says. That means refining minerals locally while building domestic cloud services, data centres, research and innovation capabilities.

“African nations cannot afford to remain mere suppliers of raw materials and consumers of imported technology in the AI era,” he says.

AI power goes beyond minerals

The emerging geopolitical contest around AI is increasingly about control of the entire stack.

The Pax Silica framework, a U.S.-led initiative launched in December 2025 to build a trusted technology and industrial ecosystem among allies, aims to secure the infrastructure and inputs needed for advanced computing. Its scope stretches from frontier AI models and computing capacity to semiconductors, advanced manufacturing, critical minerals processing and energy.

Yet none of the African countries that supply major quantities of critical minerals is among its signatories.

For Africa, the exclusion exposes a deeper imbalance. The continent supplies critical inputs such as cobalt and lithium but remains largely outside the systems that turn those resources into higher-value technology.

China offers an alternative. Through the World Artificial Intelligence Cooperation Organisation (WAICO), formally established in Shanghai in July 2026, Beijing is positioning open-weight models, infrastructure and technology cooperation as an alternative to Western restrictions.

For African governments, that could create more choices. But access is not the same as ownership.

If foreign companies build data centres, supply telecommunications equipment, provide cloud infrastructure, process African minerals, and deploy their AI systems across African markets, African countries may gain access to technology without gaining control over the underlying stack.

The geopolitical competition could therefore give Africa more suppliers without fundamentally changing its position in the value chain.

Africa’s AI problem is access

This is where Amodei’s argument exposes an important disconnect.

His concern is that AI capabilities are advancing so rapidly that safety research could fall behind. “Left unchecked,” he writes, recursive self-improvement “could outrun our ability to understand and control these systems.”

For African economies, the constraint is almost the inverse. Africa’s immediate AI problem is not an excess of frontier models operating without adequate safeguards. It is a shortage of the infrastructure required to build, train and deploy advanced AI at scale: domestic computing capacity, large data centres, reliable electricity and access to advanced chips.

The continent’s data-centre industry accounts for less than 1% of the global market. More than 580 million people in Sub-Saharan Africa — about 48% of the region’s population — lack access to electricity.

The result is a peculiar position in the global AI hierarchy. Africa can have millions of people using AI applications developed elsewhere while having relatively little capacity to train, host or control those systems domestically.

A country that lacks meaningful computing capacity is dependent on someone else’s cloud. A country that does not process strategic inputs is dependent on someone else’s supply chain. A country that does not own significant AI companies or foundational models is dependent on someone else’s technology roadmap.

And a country lacking all three has little leverage when the rules governing AI are written.

AI’s owners set the rules

Amodei’s proposed “Democratic Coordination” would see frontier AI companies in democratic countries coordinate on common safety standards and limits on unchecked AI progress.

The idea is understandable. But it also reveals who gets to define the frontier.

The institutions participating in that conversation are concentrated in countries and companies that already possess the capital, chips, models, and computing infrastructure necessary to operate at the frontier. Africa is largely absent.

“The main blind spot in his argument is the absence of the very countries that will be most affected by AI,” Awe says.

“Creating a system in which only the greatest technological powers define the rules for everyone else is both unrepresentative and unsustainable.”

The consequences extend beyond governance to algorithmic bias, job displacement, local languages and data sovereignty.

Africa needs an industrial AI strategy

African countries do not need to immediately build their own versions of the world’s largest frontier models. The more strategic question is where Africa can build leverage.

Mineral-producing countries can invest in processing and refining. Countries with abundant energy resources can combine power generation with data-centre development. Governments can use public procurement to build domestic AI capabilities rather than simply importing finished systems. Regional integration can give African AI companies markets large enough to compete beyond individual countries.

The greater opportunity lies in connecting critical minerals to industrial policy.

“African countries can create greater leverage by linking mineral, infrastructure, and investment agreements to local value creation, including skills development, technology transfer, research capacity, and opportunities for African firms,” says Adeola Bojuwoye, manager and Nigeria project lead at the Digital Impact Alliance.

“Investment in reliable energy, connectivity, regional compute capacity, and interoperable data-sharing systems will also be important.”

The IEA estimates that moving beyond extraction to beneficiation, processing, and manufacturing could significantly increase the value Africa captures from its mineral wealth, with mineral production value potentially approaching $120 billion by 2040 under its high-potential scenario.

Awe argues that Africa’s AI strategy must rest on more than its mineral endowment. “Governments must prioritise the foundations: reliable power generation, digital connectivity, strong data governance, technical education and proactive policies that create an enabling environment for innovation and drive value addition,” he says.

African countries will also need to negotiate collectively. Through institutions such as the African Union and the African Continental Free Trade Area, governments could leverage the continent’s combined market to secure better terms for investment, infrastructure, data access, and technology transfer.

A continental critical-minerals strategy could strengthen that bargaining position by linking mineral exports to commitments to finance domestic computing capacity, transfer technology, and develop local skills. Implementing the AfCFTA Digital Trade Protocol could likewise help create a more integrated African data market and strengthen the continent’s negotiating position over AI services and data governance.

The stakes are high. Intensifying U.S.-China competition in AI, semiconductors and critical minerals could give African countries new negotiating leverage — or further entrench their role as suppliers of raw materials and consumers of foreign technology.

“It could go either way,” Awe says. Competition could give African countries more choices and allow them to negotiate better deals, but only if governments are clear about their strategic interests and coordinate their positions.

“Otherwise, Africa could simply become a place where both sides battle to sell their technologies and compete for access to African minerals and markets,” he says.

The continent does not need to control every layer of the AI stack. But it needs enough of it to negotiate from a position of strength.

True scale demands moving beyond surface-level integrations to robust execution. We’ve filtered the noise out of Moonshot 2026, optimising the conference strictly for high-calibre connections between startup founders, global financial operators, enterprise leaders and individuals rewiring Africa’s technical frameworks. Get 20% off Early Bird tickets for a limited time.