NextFin News - Africa is being pulled into the artificial-intelligence economy from two directions at once: as the source of the minerals that build data centers, and as a candidate to host them. The continent holds 30% of global critical mineral reserves, yet accounts for about 0.6% of the world's data center capacity — a gap that is now attracting billions in investment, but only where reliable power exists.
The Two-Way Pull: Minerals Up, Compute Down
The minerals side of the story is already moving at speed. The Democratic Republic of Congo, which produces more than 70% of the world's cobalt, replaced a months-long export ban in September 2025 with a quota system capping cobalt shipments at 96,600 tonnes a year for 2026 and 2027 — roughly half the volume exported in 2024. The effect on prices was immediate: cobalt hydroxide jumped 263%, from $4,012 a tonne in the fourth quarter of 2024 to $14,560 a tonne in the fourth quarter of 2025.
That is not an isolated policy move. Zimbabwe has restricted exports of unprocessed lithium to force local value addition, and the DRC's first-quarter 2026 copper exports fell 14.6% to 955,000 tonnes. Producers are responding to the new rules rather than simply to prices: Glencore's DRC cobalt output dropped 39% in the first quarter of 2026 while its copper production rose 19%, a deliberate reordering under the quota regime.
The compute side is moving more slowly, but the intent is visible. Africa's operational data center IT load was about 360 megawatts in early 2026, with 238 MW under construction and 656 MW in planning — still only 0.6% of global capacity, and heavily concentrated: South Africa alone holds roughly 70% of installed capacity. The market is projected to grow from about $2.22 billion in 2026 to $4.36 billion by 2031, a compound annual growth rate of 14.46%, according to the Africa Data Centres Association.
The tension between the two sides defines the investment thesis. The same AI buildout that is tightening demand for African cobalt, copper, tin and tantalum is also generating interest in siting data centers on the continent — but every megawatt of AI-grade capacity requires several more megawatts of power infrastructure, and that is where the continent's ambition meets its constraint.
Why the Minerals Link Is Structural, Not Cyclical
The first question investors should ask is whether this is a commodity cycle or something deeper. The answer matters because cyclical moves revert and structural ones do not — and the evidence here points to structure.
On the demand side, the driver is not a single product cycle but a change in what the global economy is being built from. Data centers are becoming one of the fastest-growing new sources of demand for critical minerals: lithium for battery energy storage, tin for the solder in every circuit board and chip package, tantalum for capacitors, and copper for power distribution. Solder alone accounts for roughly half of global tin demand, and the London Metal Exchange's non-ferrous index rose 34% year-on-year in early 2026, with tin the standout at roughly 70% higher, near $50,000 a tonne. The International Energy Agency's Global Critical Minerals Outlook 2026 frames this as a persistent demand shift, not a one-off spike.
On the supply side, the concentration is the point. China controls the majority of processing for lithium (71%), cobalt (80%), rare earth elements (92%) and graphite (96%). For Western governments and their corporations, African minerals are not just cheaper tonnage — they are a diversification channel. That strategic motive does not switch off when prices dip, which is what separates this from a normal upcycle.
The capital is following. In January 2026, the U.S. International Development Finance Corporation's maximum exposure rose from $60 billion to $205 billion, and its mandate was expanded to upper-middle-income African countries including Botswana, South Africa and Namibia. The Orion Critical Mineral Consortium — a public-private vehicle backed by the DFC, Orion Resource Partners and ADQ — has $1.8 billion in capital and is targeting $5 billion, and is in talks to acquire a 40% stake in Glencore's DRC mines, a package valued at more than $9 billion. Qatar's sovereign wealth fund has made a $500 million strategic investment in Ivanhoe Mines, with a memorandum of understanding explicitly aimed at supplying the metals that "power global electrification and the rise of AI and large-scale datacentres," as executive co-chair Robert Friedland put it.
"The signing of the MoU, together with the strategic investment by the Qatar Investment Authority, is a strong vote of confidence in Ivanhoe Mines' and our mission to supply the strategic metals that power global electrification and the rise of AI and large-scale datacentres."
Infrastructure is being built to move the ore. On January 2, 2026, the Africa Finance Corporation signed a $753 million financing package for the Lobito Atlantic Railway, a 1,300-kilometre corridor linking Angola's Port of Lobito to the DRC border — a route designed to give copper and cobalt a faster, Western-aligned path to market.
The mechanism, then, is not "AI is hot, so metals are up." It is that the compute buildout has converted African minerals from a marginal supply source into a strategic chokepoint at the same time that African governments have learned they can weaponize access. That combination — inelastic demand plus assertive supply management — is a structural shift in pricing power, not a cyclical squeeze.
The Data Center Bet Is Real — But It Is a Power Trade
If the minerals thesis is about scarcity, the data center thesis is about a different scarce input: electricity. The African Energy Chamber projects African data center demand will reach 2 gigawatts by 2030, and that this growth will catalyze $10 billion to $20 billion in new energy investment. For every 1 MW of data center capacity, research suggests a multiplier of up to 3 MW in broader infrastructure development.
The economics are capital-intensive. Constructing a standard Tier III data center now costs an average of $11.3 million per megawatt, according to JLL's 2026 Global Data Center Outlook, and facilities equipped with specialized AI GPUs can cost more than double that. At those prices, a developer is not just buying servers — it is underwriting a power system.
This is why the map of African data center investment looks less like a resource map and more like a power map. South Africa dominates because it has the most advanced grid on the continent. Namibia is attracting attention not because of its minerals but because of its green-hydrogen ambitions: the Hyphen project alone targets $9.4 billion of investment, with production costs estimated at $1.73 to $2.30 per kilogram, and an estimated $190 billion would be needed by 2040 to build out the sector. Cheap renewable power is the pitch; data centers are the offtake story layered on top.
Kenya illustrates both the promise and the execution risk. Microsoft and UAE-based G42 announced a $1 billion geothermal-powered data center there, but the project faltered in 2026 over payment demands and infrastructure disagreements, according to reporting at the time. NVIDIA has separately announced plans for an AI factory in Kenya with 12,000 GPUs. The two announcements, read together, capture the state of play: the intent to build AI infrastructure in Africa is genuine, but the gap between announcement and commissioning is still wide.
The second-order implication is the one the market is not fully pricing. If African data centers require dedicated renewable generation, then the investment is not really a real-estate trade — it is an energy infrastructure trade with a computing tenant. That reframes who wins: the beneficiaries are not just colocation operators but the developers of solar, wind, geothermal and transmission assets, plus the mineral producers supplying both the panels and the servers.
The Counter-Thesis: Extraction Without Transformation
The strongest case against the optimistic reading is that Africa ends up supplying the raw materials for someone else's AI boom while hosting little of the value-added activity. History is not encouraging: the continent has exported commodities for centuries while importing finished goods. The new version of that pattern would see African cobalt and copper flow into battery and chip supply chains elsewhere, while the high-margin compute capacity stays in Northern Virginia, Amsterdam and Singapore.
There is evidence this risk is live. Export restrictions like the DRC's cobalt quotas and Zimbabwe's lithium rules are designed to force local beneficiation, but analysis of the region's industrial base finds that fully integrated battery manufacturing remains unlikely even under improved conditions; the more realistic outcome is expanded mineral refining and intermediate processing. The constraints are concrete: Zambia's debt-to-GDP ratio limits fiscal space, Zimbabwe faces foreign-exchange shortages, and the DRC remains exposed to global price shocks and insecurity in its eastern provinces.
The counter-thesis also has a timing problem on the demand side. If AI capital expenditure slows, or if chip efficiency gains reduce the mineral intensity per unit of compute, the price premia built into cobalt and tin could unwind quickly. Tin inventories on the London Metal Exchange and the Shanghai Futures Exchange have been low, amplifying volatility; a demand disappointment would hit first and hardest there.
The answer to the counter-thesis is not that it is wrong, but that it describes a different investment. The "extraction without transformation" scenario is a risk to African industrial-policy hopes, not necessarily to mineral investors — and it is precisely why Western capital is structurally motivated to stay engaged rather than trade the cycle. The falsifying signal is specific: if the DRC's 96,600-tonne cobalt quota is lifted before 2027 without a replacement mechanism, or if China's share of DRC cobalt output — currently around 80% — is not meaningfully diluted by Western-backed projects like Orion within three years, then the diversification thesis has failed and the premium was cyclical after all.
What to Watch: Three Horizons
Short term (now to three months): Export licensing and quota enforcement dominate. The DRC's unused first-quarter allocations had to ship by June 30 or be forfeited to a national strategic reserve — rules tight enough to shape operating decisions. Any further tightening, or a repeat of the 2025 export suspension, would push cobalt and copper higher regardless of demand.
Medium term (three to twelve months): The Lobito Corridor is the swing factor. If it attracts refining investment as expected, it could begin rerouting Congolese and Zambian copper and cobalt toward Western smelters, validating the diversification trade. On the compute side, watch whether the stalled Kenya projects are restructured or abandoned — a concrete test of whether AI data centers can be financed and built in Africa's harder markets.
Long term (beyond one year): This is where the structural call is won or lost. If improved power supply and logistics do lead to expanded refining and intermediate processing — as forecast, even if not full battery manufacturing — then Africa captures more of the value chain and the minerals-compute linkage becomes self-reinforcing. If not, the continent remains a price-taker in minerals and a spectator in compute.
The base case is a split outcome: Africa's role as a critical-mineral supplier strengthens structurally, supported by Western strategic capital and African export discipline, while data center growth is real but concentrated in the few markets that can guarantee power — South Africa, Namibia, Kenya — and grows more slowly than the minerals side. The upside case is that cheap renewable power, particularly green hydrogen in southern Africa, makes the continent a genuine low-cost compute hub. The downside case is that execution risk and policy reversals leave Africa supplying ore while the margins accrue elsewhere.
Data as of October 2, 2026. LME prices: copper 3-month $14,622.50 a tonne; cobalt 3-month $50,991.92 a tonne; lithium hydroxide 3-month $17,640.91 a tonne.
The investment story here is not that Africa will host the AI revolution. It is that the AI revolution cannot be built without Africa — and that the continent's governments now know it. Whether that bargaining power converts into domestic value addition, or simply into higher royalties and tighter quotas, is the decision that will separate the winners from the suppliers.
更多独家洞察尽在 nextfin.ai.
