AI hedging for naira risk: saving Nigerian manufacturers billions

The cleanest way to discuss AI for naira currency risk in supply chains is not to start with the model. Start with the loss ledger. Veriv Africa’s aggregation of company reports puts combined FY2023 foreign-exchange losses for major Nigerian consumer-goods and telecom-linked names — Nestle, MTN, Nigerian Breweries, International Breweries, and Cadbury — at ₦839.24 billion. It also cites Nestle Nigeria’s ₦184.2 billion forex loss in the first nine months of 2024, finance costs up 1,345% in Q1 2024, and operating costs up 67% from H1 2023 to H1 2024.[1]

Those figures are not abstract treasury marks. They sit inside procurement files, supplier-payment calendars, import financing, and production plans. A buyer agrees input prices. Operations builds a plan around imported raw materials or packaging. Treasury tries to source dollars or hedge what can be hedged. Then the naira moves, and a gross-margin assumption becomes a finance-cost problem before the next committee meeting.

The exchange-rate path explains why the damage spread so quickly. Veriv Africa reports a 40.9% annual naira loss against the U.S. dollar in 2024, using Central Bank of Nigeria-linked movement from N997 to N1,535 per dollar, and roughly 73% cumulative depreciation from May 2023, when the naira was around N460 per dollar.[1] F.O. Akinrele & Co. similarly frames 2024–2025 as a period of exchange volatility after the shift away from the old subsidized official-rate structure.[2]

That distinction matters. Official CBN-linked rates are more defensible in a finance review than parallel-market quotations, which are harder to verify consistently. The old official-rate cushion also mattered operationally: when it disappeared, the gap between what a manufacturer assumed it could pay for foreign inputs and what it actually needed to settle became harder to hide inside procurement timing.

The loss was created before it reached the income statement

A forex loss line can make currency risk look like an accounting event. In manufacturing, it is usually the residue of earlier operating decisions. Import dependency fixes a dollar need before sales cash arrives. Supplier credit creates a timing gap. Inventory decisions lock in exposure before the naira cost is finally known. A delayed forward contract, an incomplete exposure file, or a stale ERP rate table can turn a normal purchasing cycle into a balance-sheet hit.

The companies in Veriv Africa’s aggregation are not identical businesses, and the ₦839.24 billion figure should not be treated as one neat hedging failure. Veriv is compiling from interim and audited reports rather than serving as the originating filing for each company. Even so, the aggregation is useful because it shows materiality. The losses were large enough to affect how boards read procurement performance, not just how treasury explains market conditions.[1]

The temptation is to ask why every exposed manufacturer did not hedge more aggressively. That is too easy. Nigerian importers have not been operating in a deep, frictionless hedging market. Coronation Merchant Bank describes businesses using tools such as forwards to lock rates, while the broader market still relies heavily on manual forward contracts and non-deliverable forwards; F.O. Akinrele & Co. cites FX derivatives turnover of $36.14 million, which is shallow relative to the operating exposures being discussed.[2][4]

A forward can protect a known payment. An NDF can manage offshore currency exposure where delivery is constrained. Neither solves the upstream problem if the exposure record is late, fragmented, or disputed between procurement, operations, and treasury. The instrument may exist; the decision system around it may still be too slow.

What the AI evidence actually proves

The strongest published evidence in the available material is not Nigerian and not manufacturing. Global Finance reports that a Citi and Ant International AI-powered FX hedging pilot with an airline customer reduced hedging costs by about 30% and achieved forecasting accuracy above 90%.[3] Those are serious numbers, but they are not a license to multiply 30% by ₦839.24 billion and call the difference savings.

There are at least three gaps between that pilot and a Nigerian manufacturer’s naira exposure. The customer was an airline, not a consumer-goods producer. The market context was different. The reported saving was on hedging cost, not on total reported forex losses. A corporate forex-loss line can include revaluation effects, dollar debt, payables, timing mismatches, and accounting treatments that no forecasting tool can simply erase.

Still, the pilot is relevant because it addresses the part of the problem that procurement and treasury can actually control: when exposure is identified, how much is hedged, at what tenor, and with what feedback from realized cash flows. FX exposure is partly a forecasting and timing problem. Better data will not create dollar liquidity, but it can reduce blind spots before a payment date turns into a variance explanation.

Okoora and Palm AI make broader claims that AI tools can improve FX forecasting accuracy by 20% to 30% for treasury teams, but those are vendor or industry-facing materials rather than independent Nigerian manufacturing evaluations.[5][6] They help define what the product category promises. They do not prove what a Lagos-based manufacturer would have saved on imported milk powder, malt, packaging, spare parts, or dollar-linked obligations.

The useful counterfactual is narrower than the sales pitch

The proper counterfactual is not: AI would have saved Nigerian manufacturers 30% of ₦839.24 billion. The better question is: which portions of those losses were tied to forecastable exposures that could have been hedged earlier, sized more accurately, or rolled with less cost if procurement, treasury, and ERP data had been connected in time?

That narrows the claim but makes it more useful. AI-assisted hedging could plausibly reduce avoidable costs where four conditions hold: the exposure is visible before settlement, hedge instruments are available at usable tenors, treasury policy permits action, and the model’s recommendations are auditable enough for finance control. Miss any one of those, and the tool becomes a dashboard over a loss that is already in motion.

For example, a hypothetical manufacturer with dollar-denominated input purchases due across a quarter would not need AI to know depreciation is painful. The model’s value would be in reconciling purchase orders, expected production consumption, supplier-payment dates, inventory buffers, receivables timing, and current hedge coverage before treasury chooses the tenor and hedge ratio. That is different from generic currency forecasting. It is supply-chain exposure management.

This is also where FX technology starts to resemble other planning systems built for external volatility. Tariffs, freight shocks, and currency depreciation all punish planning processes that update after the commercial decision has already been made. The same logic behind AI planning tools for tariff volatility applies here: the model is only useful if it changes the decision while procurement still has options.

Why deployment still lags the size of the problem

AI decision support is moving deeper into supply-chain work globally. RELEX’s 2026 survey of 514 global supply-chain leaders found that 67% were more confident in AI decision-making year over year, 71% planned generative-AI investment, 47% were investing in AI inventory optimization, and 57% identified raw-material procurement disruption as the most impacted area.[7] That does not prove Nigerian adoption. It does show that AI is no longer confined to experimental demand-planning decks.

The Nigerian FX case is harder than inventory optimization because it touches treasury authority, bank relationships, derivatives availability, accounting treatment, and board risk appetite. A planner can recommend a different inventory position. A hedging engine can recommend a financial commitment that must survive audit, policy limits, counterparty review, and liquidity constraints.

The platform market is not empty. Pangea presents itself as a CTA-regulated, AI-powered FX hedging platform and publishes a case study involving KSW Global, described as a West African cocoa producer.[8] Attara has announced AI-enabled financial tools, including an Explore Zone and an AI assistant called Tara for SME hedging support.[9] These are signs that AI-enabled hedging is moving toward the kinds of firms that face commodity, currency, and cross-border payment risk.

They need a label attached: vendor-published. The available sources do not provide independently validated evidence that these tools have reduced naira-related forex losses for Nigerian manufacturers. The difference is not pedantic. A finance committee can use vendor material to build a shortlist. It should not use it as proof that a Nigerian deployment will deliver the same saving as a Citi/Ant airline pilot or the same forecasting uplift described in broader industry materials.

Fintech infrastructure is changing around the edges as well. Risk Advisory describes Nigeria’s FX landscape being reshaped by digital providers, including multi-currency wallets and cross-border payment infrastructure from names such as Flutterwave, Waza, and the Moniepoint/Visa partnership.[10] Those services can reduce payment friction and improve currency access workflows. They are not, by themselves, evidence of AI hedging effectiveness for manufacturing exposures.

The missing evidence is now part of the risk

The uncomfortable gap is visible. On one side, Nigerian manufacturers and adjacent large firms have reported forex losses large enough to overwhelm ordinary procurement savings. On the other, published AI hedging evidence shows meaningful cost reduction and high forecast accuracy in at least one corporate setting, with additional vendor material pointing to a growing tool market. Between them sits the part no one should smooth over: there is no independently validated, public Nigerian manufacturing deployment in the available sources.

That absence cuts both ways. It protects against overclaiming. It also makes non-deployment harder to defend as a default position. If a manufacturer has recurring dollar exposure, imported inputs, naira revenue, and a history of large FX variances, the question is no longer whether AI sounds fashionable. The question is whether the existing manual process can show, with evidence, that it identifies exposure early enough and hedges it at a cost that a better decision system could not improve.

The implementation standard should be practical. Connect purchase orders, supplier-payment schedules, inventory plans, open payables, forecast sales cash flows, existing hedge positions, and ERP rate assumptions. Test model recommendations against historical exposure windows. Separate savings from avoided loss, reduced hedge cost, improved forecast accuracy, and better policy compliance. Keep the treasury override visible. If the model cannot be audited, it will not survive the first serious variance review.

That framing also avoids the common mistake of treating standalone FX tools and platform-embedded AI as the same buying decision. Some manufacturers may need a treasury-led hedging layer; others may want FX exposure surfaced inside planning suites alongside procurement and inventory decisions. The broader question of AI-native versus legacy supply-chain platforms matters because currency risk is not created in one system. It is assembled across planning, procurement, finance, and settlement.

A conditional verdict for Nigerian manufacturers

AI hedging is not proven as a plug-in cure for naira losses. The best published pilot is specific, quantified, and useful, but it is still an airline case in a different market. The Nigerian loss data is large and disturbing, but it is aggregated from company reports and contains components that no model could fully prevent. The responsible conclusion is narrower: AI-assisted FX hedging could have reduced some avoidable cost and timing error where exposures were visible, instruments were usable, and treasury policy allowed earlier action.

That narrower conclusion is enough to change the committee conversation. After ₦839.24 billion in reported FY2023 forex losses among major firms and a single manufacturer reporting ₦184.2 billion in nine-month forex losses, Nigerian supply-chain and treasury teams should not have to prove that AI is magic. They should have to prove that the current exposure-management process is good enough to leave untouched.

References

1. Navigating the Storm: The Impact of Currency Devaluation on Nigeria's Consumer Goods Industry, Veriv Africa
2. Nigeria's Exchange Volatility, F.O. Akinrele & Co.
3. AI-Powered FX Hedging: How Corporate Treasuries Cut Costs, Global Finance Magazine
4. Lock in Your Profits, Coronation Merchant Bank
5. The Role of AI in Real-Time Currency Hedging, Okoora
6. Manage foreign exchange risk in your supply chain, Palm AI
7. RELEX Report: AI Moves Into Core Supply Chain Decisions as Volatility Persists, RELEX Solutions
8. Pangea, Pangea
9. Hedging Fintech Attara Launches AI-Enabled Enterprise Grade Financial Tools, Attara
10. The digital pivot: How fintech is redefining Nigeria's FX landscape, Risk Advisory