Close
Climate Finance

Value leaks from African farms where quality and volume go unproven

(20 min read)

A fortnight of African agritech newsletters reads like a list of unrelated deals: a Tunisian insect protein plant, solar dryers in Kenya, a Nigerian tractor scheme, a Safaricom earnings jump. Read together with Kenya’s dairy data and a few World Bank events on women and jobs, they tell one story. Africa grows plenty. The value leaks out where nobody can prove quality, aggregate volume, move money safely or show who benefits. Those are the gaps Kilimora builds AgriKonnekt to close for smallholder cooperatives. This article asks three things: how much is lost, where the losses sit, and which of this month’s stories point to a fix that cooperatives can actually use. Later sections add three themes the dairy story does not cover: why good products still fail to be adopted, why carbon money needs price discipline before it can be trusted, and what residues and rural youth jobs add to the case.

Ten findings

  • Africa buys its milk abroad. Tech Safari reports US$4 billion of milk powder imported a year, from a continent with about 350 million cattle.
  • Kenya is the counter example, but only on paper. Production is about 5.2 billion litres, yet only 1.01 billion litres (about 19%) reached formal buyers in 2025.
  • Formal intake is growing fast. It rose 11.6% in 2025 and is up about 25% in two years. Dairy imports fell from US$86.3 million to US$57 million in 2024.
  • Losses are the biggest hidden crop. Estimates run from 15% of grain value to 50% of fruit and vegetables, and about four in ten Kenyan avocados.
  • Satellites now see a smallholder plot. Sentinel-2 sees in 10 m pixels, so a hectare is 100 pixels, against about a sixth of one pixel for MODIS at 250 m.
  • Women hold the gap. 49% of women took part in the global labour force in 2023 against 73% of men, and capital in mixed households flows disproportionately to men.
  • Money rails exist; verified evidence does not. M-PESA carries over 40% of Kenya’s digital payments, yet less than 3% of agricultural investment reaches smallholders, by Kilimora’s figures.
  • Adoption is the bottleneck, not access. In clean cooking surveys 29% of customers reported problems and 74% of those said they stayed unresolved, about 21% of all customers on our arithmetic.
  • Carbon prices need design. The EU carbon market ran 14 years before adding a stability reserve, with prices of €3 to €7, and Alberta’s credits traded 40% below the official price.
  • Youth jobs can be cheap to start. Kenya’s first JOY phase created 5,000 jobs against a target of 3,000 from a €1.2 million allocation, about €240 a job on our arithmetic.

The milk that Africa does not make

Tech Safari’s “Dear Dairy…” issue of 12 November 2025 starts from a paradox. Africa has more cattle than any other continent, yet imports about 15 billion litres’ worth of milk powder. Its explanation is practical. Powder keeps for years and fresh milk spoils in days. Trucks reach remote shops over bad roads, and three generations have grown up on reconstituted powder. In Nigeria, it reports, 98% of consumed milk is imported, much of it fat filled powder blended with palm oil. Five companies, it says, control 99% of the market. We have not verified those Nigerian figures, but the direction is clear.

Kenya took a different road, and the data show both the gain and the gap. The Kenya Dairy Board puts national production at about 5.2 billion litres, from roughly 1.8 million smallholders who supply about 80% of it, and the sector supports some 1.2 million jobs. Tech Safari says smallholders produce 95% of locally sourced milk, so even basic shares differ by source. We use the Dairy Board’s 80%.

Only one litre in five meets a formal buyer

Share of Kenya’s roughly 5.2 billion litres a year sold through licensed processors in 2025. The informal share is our subtraction

Formal: 1.01bn L, 19%Informal and on-farm: about 4.2bn L, 81%Formal 1.01bn L · 19%Informal and on-farm, about 4.2bn L · 81%Total about 5.2 billion litres a year (Kenya Dairy Board; Nation, 2026)

The formal slice is the one that moves. Processors bought 810.8 million litres in 2023, 908.4 million in 2024 and 1.01 billion in 2025, which is the highest since the Board began recording in 2001. Cheaper imports matter less here than supply chains that can be audited: chilled, tested and paid for.

Formal milk intake keeps climbing

Litres bought by Kenyan processors, million. Imports of dairy products fell from US$86.3M (2023) to US$57M (2024)

2023: 810.8M L2024: 908.4M L2025: about 1,010M L810.8908.41,010202320242025Two-year growthabout 25%Kenya Dairy Board, as reportedDairy roadmap target: 10bn L by 2033

Why the cooperative keeps showing up

Tech Safari’s sharpest line is that better cows alone will not fix a broken system. Rwanda, it reports, lost 80% of its cows about three decades ago and is now close to self sufficient in fresh milk. It did so with cooling centres, collection networks, subsidised breeds, farmer training, strategic tariffs and a US$54 million powder plant. In Kenya, Mukurweini Wakulima shows the cooperative version: it turns members’ milk into yoghurt and UHT milk that competes with imports. In both cases the unit of success is an organised group with shared infrastructure and a record of what it delivers.

That is the premise of AgriKonnekt. Farmers join through a cooperative they already trust, learn by USSD on any phone, and the cooperative becomes the accountability layer that buyers and funders can see. Kilimora reports 1,500 farmers across six Kenyan counties, 120 plus cooperative networks taking part, and 96% documented order fulfilment. Rwanda’s lesson fits: dependable aggregation is what lets a small producer bargain with, or displace, a global brand.

This month’s storyThe numberWhat it means for a cooperative
Dairy imports (Tech Safari)US$4bn powder a yearLocal milk wins only with cooling, testing and shared bargaining
Farm to Feed, Kenya2.1M kg from 6,500 farmers“Rescue grade” produce becomes income when it is aggregated and graded
Synnefa solar dryersUS$300,000 grantDrying cuts spoilage at group level, where one machine serves many
Sentinel-2 over MODIS10 m pixelsPlot level checks are now affordable evidence
Standard Bank and Sawa EnergyR750M; US$2.9MSolar and batteries are financed once someone can verify the load
World Bank women and jobs events49% vs 73%Programmes must be built around who controls the money

The harvest that never reaches the market

Losses are the second leak. The FAO and World Bank put post harvest grain losses in sub Saharan Africa at about US$4 billion a year, roughly 15% of a US$27 billion crop on 2005 to 07 prices, and said current values would be higher. UNEP says losses of fruit, vegetables and roots can reach 50%, and FAO has put losses of some tropical fruit above 40%. Our own earlier reading of the Kenyan avocado chain, using SAFIC’s webinar figures, found about four in ten fruits lost between tree and truck.

The more perishable the crop, the bigger the leak

Estimated losses, % of value or volume. Measures differ by source, so compare direction, not decimals

SSA grain, share of value (FAO/WB)Cereals, many African countriesKenyan avocado, tree to truckFruit, veg and roots, up to (UNEP)15%25%about 40%up to 50%15%25%about 40%50%

Two newsletter stories attack this leak at its source. Farm to Feed has sold 2.1 million kg of rejected but edible produce from 6,500 farmers, about 323 kg each on our arithmetic, and says as much as 40% of African food goes to waste. Synnefa’s US$300,000 grant funds smart solar dryers. Both work because produce is bought, dried or graded as a lot. A lone farmer rarely has the volume to justify a dryer or a grading buyer.

Proof is the product

The third leak is evidence. The Agritech Digest reports that Sentinel-2 has overtaken MODIS for crop yield forecasting because its finer resolution gives clearer field level data. The size of that difference is easy to underrate. MODIS’s finest products are 250 m pixels, each covering 6.25 ha. Sentinel-2 sees 10 m pixels, so one hectare is 100 pixels and a 250 m MODIS pixel holds 625 of them.

One coarse pixel, 625 fine ones

The same 250 m by 250 m patch of land, drawn to scale as MODIS and Sentinel-2 see it

MODIS: 1 pixelSentinel-2: 625 pixelsA one-hectare plot is100 Sentinel-2 pixels0.16 of a MODIS pixelOur arithmetic

Better pictures are not proof, though. A funder wants to know that the soil, the practice and the farmer are real. AgriKonnekt answers with two layers that check each other. Solar powered field sensors read soil carbon, moisture, pH and organic matter, with AI running on the device so it works offline. Sentinel-2 imagery at 10 m cross checks vegetation and soil trends for every plot. Kilimora’s team reviews the result and stores a time stamped record before project developers see it. Kilimora says this dual layer is designed to meet BioCarbon Standard and ISCC Carbon Farming requirements and cuts verification cost by 70%. Those are its own claims, and the cost saving in particular needs independent confirmation. The design is sound for a reason the Kenyan avocado story also shows: a record that is checked twice is something a buyer, lender or carbon registry can rely on.

The climate ledger in a glass of milk

Dairy carries a climate bill that cooperatives can start to measure. FAO’s global dairy study put the sector at 4% of human caused greenhouse gases and the average at 2.4 kg of CO2 equivalent per kg of milk, with methane about 52% of the total. Its wider livestock assessment put livestock at 14.5%, cattle accounting for 65% of that. Applying the global average to Kenya’s output gives roughly 13 million tonnes of CO2 equivalent a year (our arithmetic; Kenyan systems differ, so read it as scale only).

The same logic cuts the other way. GenePlus cows giving 15 litres a day instead of 5, as Tech Safari reports, mean fewer animals per litre and lower emissions per kilo of milk. Healthier herds from Cowtribe’s drone delivered vaccines mean fewer losses. On the land side, AgriKonnekt’s entry product is an organic fertiliser made from abattoir blood and bone meal and from crop residues. Kilimora reports 31.1% organic carbon and a C:N ratio of 18.1, in a batch tested by Cropnuts to ISO/IEC 17025. Waste that would rot and release methane becomes a measured input, and the same carbon that raises yield is the carbon that can later be counted.

Power, payments and the cost of proof

Two financing stories show where capital is moving. Standard Bank is putting R750 million (US$43.34 million) behind New GX Capital’s Airnergize platform for commercial and industrial solar and batteries, and ElectriFI has backed Uganda’s Sawa Energy with US$2.9 million to finance solar for businesses switching from diesel. Kilimora’s energy work follows the same logic for households: efficient stoves and local power, with the emissions avoided measured so partners can verify and fund the result. The common point is that money follows measured load.

The payment rails already exist. Safaricom’s half year profit rose 52% to KES 42.8 billion, helped by M-PESA, which carries over 40% of Kenya’s digital payments. MTN Rwanda reports 5.8 million Mobile Money users and 578,000 merchants. AgriKonnekt is built on them, paying farmers directly to their mobile accounts within 24 hours once funders release money, with no data plan needed for the lessons. The cooperative stays the accountable unit, but the money lands with the household.

Who gets the capital

The World Bank’s November 2025 programme of events keeps asking one question. Only 49% of women joined the global labour force in 2023, against 73% of men, and women hold more precarious, lower paid jobs. The Partnership for Economic Inclusion says 90% of economic inclusion programmes target women, yet only a third make women’s empowerment a core objective. Its warning is that targeting without design can reinforce inequality. A WEESA webinar adds a finding that matters for any cooperative lending model: when a woman is the sole owner, capital turns into business gains, but when a household runs separate male and female enterprises, investment flows disproportionately to men.

The labour force gap that programmes inherit

Share of people taking part in the global labour force, 2023, World Bank Partnership for Economic Inclusion

WomenMenWomen 49%Men 73%49%73%24-point gap

Kilimora says more than 60% of its funding recipients and roles are women by design, and its gender lens programming sits beside its agriculture and energy work. The WEESA finding gives that claim a test: measure who in each household controls the mobile account that receives the payment. The same World Bank list has a forestry event with a striking ratio, 73 extra jobs in the wider economy for every 100 forest sector jobs. That supports agroforestry as a source of rural work, an area Kilimora works in alongside KEFRI.

Behaviour before technology: what clean cooking teaches

NextBillion’s essay of 10 November 2025 by Jean Louis Racine of the Clean Cooking Alliance is about stoves, yet its argument fits every product a cooperative introduces. More than 2.1 billion people still cook with polluting fuels (2023), which the author links to over 3 million deaths a year and over 1 gigatonne of CO2 emissions annually, comparable to the airline industry. Subsidies and information campaigns have not delivered adoption. The Alliance’s User Insights Lab found that people favour familiar products, wait for several trusted voices before switching, and often keep the old method running alongside the new. In Kenya, a high share of LPG users go back to traditional fuel for long boiling foods such as beans and githeri, and LPG stoves tended to burn chapatis.

Two further findings matter. Smaller subsidies on expensive products worked better than large subsidies on cheap ones, and an efficiency label raised willingness to pay for charcoal stoves but nowhere near enough to cover the higher price. Then came the customer survey below.

From a sale to a satisfied user

Of every 100 clean cooking customers surveyed by the Alliance. The last bar is our multiplication of 29% by 74%

All customersReported a problem (29%)Problem never resolved (74% of them)10029about 2110029about 21

This is why AgriKonnekt starts with an input farmers already recognise: organic fertiliser, payable in instalments, delivered through the cooperative, with USSD lessons and climate fit seedlings chosen so the first season shows visible results. Social proof is built in, because members see the plot next door. For Kilimora’s energy work the lesson is to design for the pot of beans, not the brochure, and to measure continued use, not stoves sold. The emissions avoided that a funder verifies depend on the stove being used.

Carbon budgets, carbon prices and the case for patience

Our World in Data’s latest budget estimates frame the stakes. With global CO2 emissions at about 42 gigatonnes in 2025, the remaining budget with a 50% chance of staying under 1.5°C is 170 Gt, about four years of current emissions. The 1.7°C budget is 525 Gt (12.5 years) and the 2°C budget is 1,055 Gt (25 years). At current rates the 1.5°C budget runs out around 2030.

The budget is measured in years, not decades

Remaining CO2 budget from the start of 2026 at a 50% likelihood, Gt, and years at 2025 emission rates. Our World in Data, citing the Global Carbon Project and IPCC

1.5°C1.7°C2°C170 Gt525 Gt1,055 Gt170 Gt · about 4 years525 Gt · 12.5 years1,055 Gt · 25 years

Carbon pricing is one of the tools meant to stretch those budgets, and the World Economic Forum piece of 10 November 2025 by IEEFA analysts shows how easily it goes wrong. India’s new Carbon Credit Trading Scheme covers nine industrial sectors, with compliance starting in 2026 and initial targets averaging only a 3% reduction. Combined with indefinite banking of credits, modest targets can build surpluses that depress prices. Alberta’s TIER system had over 53 million surplus credits by 2023 and prices 40% below the official price. Australia’s Safeguard Mechanism needed reform after seven years. The EU’s market worked for 14 years before adding its Market Stability Reserve, with prices between €3 and €7. California, which launched in 2013 with a price floor of US$10 rising 5% a year plus inflation, kept stronger signals. Research cited by the authors suggests nearly two thirds of efficiency losses could have been avoided with earlier adjustments.

The cooperative reading is simple: a price that cannot be trusted cannot fund a household’s decisions. AgriKonnekt’s stated sequence reflects this. It routes mitigation and adaptation funding to cooperatives first and phases in carbon credits as pricing stabilises, with an Article 6 pathway under review. Kilimora’s 2030 target of 180,000 verified tonnes a year is about 0.0004% of 2025 global emissions, so its claim is not scale but credibility: each tonne checked twice. Its “above US$9 million a year” market value implies roughly US$50 a tonne (our arithmetic). That is the kind of price assumption this essay says needs a stability mechanism before anyone depends on it.

From residue to revenue

A Human Progress essay of 6 November 2025 by Aditya Goyal of North Dakota State University argues that waste is becoming a resource. It cites the World Bank’s What a Waste 2.0, which projects global waste rising 70%, from 2.01 billion to 3.4 billion tons by 2050, and a circular economy opportunity of US$4.5 trillion by 2030. That second figure is a projection and the lab results, such as nanoparticles that capture lake phosphates for fertiliser and flax stalks turned into packaging, are preliminary and American. The link to our work is the phosphate point: Morocco holds about 70% of the world’s phosphate rock, so nutrient recovery is a supply security issue, and the author notes agricultural cooperatives are among the partners seeking value from crop residues.

Kenya has its own example. In Njoro, Nakuru County, entrepreneur Robert Mwangi’s CAMLPO Limited uses black soldier fly farming to turn agricultural waste into protein rich animal feed and organic fertiliser. He says margins reach up to 60% and that the fertiliser proved even more profitable than the feed, though these are his own figures. Nakuru is one of the six counties where Kilimora works, and AgriKonnekt’s fertiliser follows the same logic with abattoir blood and bone meal and crop residues tested in a laboratory before delivery. Its Cropnuts report shows 1.65% phosphorus and 1.72% nitrogen, so the nutrients that Goyal wants to recover from lakes are already being returned to Kenyan soil from waste streams.

Jobs for the people who do the verifying

KBC’s report of 23 October 2025 on the IFAD funded Jobs Opening for Youths (JOY) project supplies a benchmark. Kenya’s first phase allocation of €1.2 million (about KSh180 million) created 5,000 jobs against a target of 3,000, and more than 43,000 rural youth, half of them young women, have been connected to opportunities. Across all countries the programme reports 60,000 opportunities, with a second phase planned for early 2026 that adds Zambia and Sri Lanka. It works through hubs at Egerton University, Pwani University and Shamberere Polytechnic. One partner, Inuka Solutions, trained 15 students by the end of 2023, seven of whom now run their own enterprises, and works with over 1,000 smallholders.

Jobs against target, Kenya, JOY phase one

Jobs created versus initial target. The 67% overshoot and about €240 per job are our arithmetic from the €1.2 million allocation

TargetCreated3,0005,0003,0005,000

The figures need care, since jobs differ in length and quality and the programme’s own counts are not independently audited. Even so, the model resembles Kilimora’s cascade approach, where each new cooperative network trains its own farmer educators. Kilimora aims for 5,000 direct roles by 2030, more than 60% held by women, spanning farmer educators, MRV data collectors, cooperative managers and clean energy technicians. JOY’s hub based training suggests where such roles could be recruited and who could certify them.

Opportunities worth watching, with caveats

The Agritech Digest also lists openings. NextProtein raised US$20.7 million in a Series B for a Tunisian insect protein plant of 12,000 tonnes a year, which could cheapen feed for the livestock and aquaculture farms that dairy and fish cooperatives depend on. Nigeria’s mechanisation programme offers tractors on flexible financing to youth and women, with applications that closed on 10 November 2025. Check the Seeding The Future challenge before sharing it: the digest gives both US$25,000 and US$250,000 and a January 2025 deadline, so the figures should be confirmed on the IFT page. Note also that Africa wide generalisations hide huge variation. Tech Safari’s Nigeria numbers come from a newsletter, not a statistical agency, and the Kenyan data are annual estimates.

The verdict, and what we would ask for

Kenya shows that the problem is not the cow or the crop. The leaks are in aggregation, loss, proof and who controls the money, and each has a solution that works at cooperative scale: shared cooling and drying, quality testing, dual layer verification, mobile payments and gender aware design. Three signals would change this view: published independent verification of the 70% cost claim, a full gender breakdown of who receives each payment, and national statistics that reconcile formal and informal milk.

  1. Publish one reconciled national dairy dataset covering formal and informal milk, with source and year.
  2. Finance shared cooling, drying and testing at cooperative level, not farm by farm.
  3. Treat plot level satellite and sensor records as eligible evidence for credit, insurance and climate finance.
  4. Report the farm gate share of the final price each season, by grade.
  5. Track the receiving account for every cooperative payment by gender of holder.
  6. Pair every input scheme, such as tractors or fertiliser, with training and a record that survives the project.
  7. Publish independent validation of verification cost claims, including Kilimora’s own.
  8. Measure sustained use, not just uptake, by tracking repeat purchases and unresolved complaints for every input and appliance a cooperative sells.
  9. Lead with mitigation and adaptation finance and add carbon credits only alongside price stability rules and records that are checked twice.
  10. Count residue based enterprises and jobs by type, duration and gender, so the circular economy is reported as livelihoods and not only tonnes.
Sources & notes

SOURCES: Tech Safari, “Dear Dairy…”, 12 November 2025 and “This Week in African Tech”, 9 November 2025; AgriTech Digest, “This Week in Agritech”; World Bank Live, Partnership for Economic Inclusion and WEESA event notices, 17 to 20 November 2025; Kenya Dairy Board data as reported by Nation, The Star and Business Daily; Invest Kenya Dairy Sector Pack, March 2026; FAO and World Bank, Missing Food: Postharvest Grain Losses in Sub Saharan Africa; UNEP and FAO loss estimates as cited by IPS; Jean Louis Racine, Clean Cooking Alliance, NextBillion, 10 November 2025; Saurabh Trivedi and Subham Shrivastava (IEEFA), World Economic Forum, 10 November 2025; Aditya Goyal, Human Progress, 6 November 2025, citing World Bank What a Waste 2.0; Margaret Kalekye, KBC Digital, 23 October 2025, on the IFAD JOY project; Hannah Ritchie, Our World in Data, carbon budget data insight, citing IPCC, Forster et al. (2025) and the Global Carbon Project; FAO dairy and livestock emissions studies (GLEAM) as reported by FWI and The Dairy Site; Kilimora AgriKonnekt, figures as shown on 5 November 2026. Newsletter figures are as published and have not been independently verified. The 19% and 81% milk shares, the 25% two year growth, 323 kg per farmer, 625 pixels, 6.25 ha, 0.16 pixel the 13 million tonne emissions estimate, the 21% unresolved share, the 0.0004% and US$50 a tonne comparisons, the 67% overshoot and the €240 per job are our own arithmetic. Kilimora performance figures (1,500 farmers, 96% fulfilment, 70% cost reduction, 60% women, soil test results) are the company’s own.