At 4 AM, Marikiti, Nairobi’s largest wholesale fresh produce market, is already trading. Lorries arrive overnight carrying potatoes, tomatoes, onions, and cabbages from farms across Kenya and neighbouring countries. Brokers match sellers with buyers, negotiate prices, and move perishable stock before it loses value.

This machinery feeds Nairobi, operating in one of the city centre’s busiest trading hubs without the warehouses, payrolls, and corporate overheads of a venture-backed distributor. Twiga Foods raised more than $185 million trying to formalise fresh produce distribution in East Africa. What did venture capital misunderstand about the market it wanted to replace? .

Twiga’s failure is now playing out in insolvency proceedings. GT Flow Limited, formerly Twiga Foods One Limited, entered statutory administration in August 2026. In September, its parent, Twiga Foods Limited, renamed Templar Field Limited, followed. Mohamed A. Mohamed of Maawiy Financial Advisory Services was appointed administrator of both companies, taking control of their assets and affairs as creditors submitted claims.

The corporate unwind preceding statutory administration reveals the stark contrast between venture capital expectations and operational realities:

• In December 2023, Twiga secured a $35 million convertible bond from Creadev and Juven, alongside $1 million in personal capital from co-founder Peter Njonjo, primarily to settle mounting debts with trade suppliers.
• Njonjo resigned from the board in 2024, handing management to former Jumia executive Charles Ballard, who cut over 600 roles between 2023 and 2025.
• In mid-2025, Twiga abandoned its custom-built distribution hub at Tatu City and acquired controlling stakes in three traditional fast-moving consumer goods (FMCG) distributors—Jumra, Sojpar, and Raisons—to pivot away from volatile fresh produce.
• Internal documents leaked in April 2025 outlined an alleged plan to transfer the Twiga brand and core assets into a clean corporate shell (“NewCo”) while leaving legacy liabilities behind in renamed entities.

Twiga’s financial unwind exposes the cost of building a distribution business around heavy fixed commitments and thin margins. Its logistics infrastructure, technology costs, and long-term leases created obligations that were difficult to sustain as the business struggled to fund day-to-day operations.

The group’s 2025 acquisitions of FMCG distributors Jumra, Sojpar, and Raisons added more businesses to a corporate structure already struggling to turn distribution volumes into enough cash to service its debts.

Marikiti’s edge

At Marikiti, time costs money. Tomatoes soften, cabbages wilt, and yesterday’s asking price can become today’s loss. Brokers manage this by adjusting prices throughout the day and matching produce with buyers before it loses more value.

A centralised distributor has less room to manoeuvre. It buys produce, moves it through warehouses, sorts it, and delivers it to retailers. Every step adds costs, from fuel and labour to storage and inventory management. Those costs remain even when prices fall, or a shipment arrives damaged.

The margins leave little room for mistakes: at a gross margin of 5% to 8%, a badly damaged shipment can wipe out profit from several successful deliveries. The distributor must absorb the loss, discount the stock, or risk losing the customer.

Marikiti’s brokers can respond faster: if tomatoes start softening, they cut prices and redirect the stock to food vendors who can use it that day. As the market thins, prices fall further. The goal is to turn produce into cash before it becomes waste. Brokers earn commissions without owning every crate they move.

Their advantage is the ability to reprice, reroute, and find another buyer before the clock runs out. A centralised platform has to recover the cost of its entire operation from each transaction. At Marikiti (or any other big market), prices can change in minutes, but overheads cannot.

Moonshot 2026 is almost here! Get 30% off all tickets during our FLASH SALE and join us in Lagos, Nigeria on October 28–29.

Get your tickets now!

Why informal credit outperforms fintech

The secondary flaw in the B2B tech model involves the mechanics of informal trade credit. Informality in retail is a rational risk-management framework designed around severe cash flow volatility.

A kiosk operating a roadside stall buys inventory on morning credit from a supplier, sells produce throughout the day, and settles the bill in the evening. When tech platforms try to digitise this credit relationship through formal financial technology rails, they run into a fundamental enforcement problem.

If the platform extends trade credit via a mobile app and the vendor experiences a poor sales day, the vendor simply defaults on the digital wallet (Copia tried this model, and it failed too). The platform cannot repossess two bags of greens, nor can it economically pursue a micro-vendor through formal legal channels.

The informal broker manages credit through relational capital and immediate market exclusion. The broker knows where the vendor lives, where her stall is located, and which family members operate adjacent shops. If a vendor defaults on a broker, she/he is instantly blacklisted across the entire physical wholesale market cluster. Because access to daily inventory is her sole livelihood, the vendor prioritises paying her informal broker above all other obligations. .

Furthermore, the informal broker executes an asymmetric risk transfer that corporate startups cannot replicate. When wholesale prices collapse, the broker pushes the loss backwards onto the producer. If a broker commits to collecting tomatoes from a farmer but wholesale prices fall overnight, the broker simply delays arrival or demands an immediate 40% price reduction on the spot. Facing total crop rot within 48 hours, the farmer has no leverage and accepts the reduced payout.

A venture-backed startup, bound by supply agreements, public relations, and corporate governance, cannot easily renege on promised farmgate prices without destroying its brand reputation or facing litigation.

This failure mode is systemic across Kenyan market tech:

• MarketForce raised $42 million to digitise informal retail ordering before shutting down its B2B e-commerce platform RejaReja due to unsustainable trade credit defaults.
• iProcure raised tens of millions to digitise agricultural input supply chains before entering insolvency proceedings after burning capital on inventory distribution hubs.
• Wasoko and MaxAB executed a defensive merger across North and East Africa to trim operational burn rates after recognising that scaling B2B retail distribution multiplied losses.

Attempts to cut out middlemen in informal produce markets have repeatedly underestimated what brokers do. At Marikiti, they set prices, extend credit, connect buyers and sellers, and clear stock before it spoils. These relationships are part of the market’s infrastructure.

Technology firms need to understand how they work before building products for the traders who depend on them. Payments, messaging, and route-planning tools can help brokers run their businesses without buying produce, holding inventory, or running an expensive distribution network. Twiga’s lesson is to understand the market’s economics before trying to rebuild it.

Kenn Abuya

Senior Reporter, TechCabal

Thank you for reading this far. Feel free to email kenn[at]bigcabal.com, with your thoughts about this edition of NextWave. Or just click reply to share your thoughts and feedback.

Microsoft has shut its Africa Development Centre (ADC) in Nairobi, Kenya, leaving more than 400 engineers without jobs, following the closure of its Lagos engineering operation in May 2024. It has also closed sales offices in Windhoek, Abidjan, Dakar, Accra and Port-Louis. Microsoft says it “remains committed to Kenya”, but its retreat from the places where it once promised to build African engineering talent tells a different story. As the company and its peers pour billions into artificial intelligence (AI), these closures also reinforce a message that benefits the industry selling it. AI is coming for your job, and the companies building it are worth investing in.

The Nairobi decision is difficult to explain through performance alone. Employees familiar with the centre say it had no quality or performance problems, and some received strong appraisal ratings before learning their jobs were at risk. Microsoft has not publicly said poor performance prompted the decision. People close to the matter say about five employees across customer and partner solutions, recruitment, and sales will remain in the affected operation.

Announced in 2019, the centre was part of Microsoft’s $100 million commitment to build engineering capacity in Africa. The company planned to hire 500 engineers across Nairobi and Lagos, Nigeria, giving it access to local talent while offering African engineers the chance to work on products used worldwide. The Lagos hub closed in 2024, and Nairobi is now following suit, exiting Microsoft’s engineering business on the continent.

Microsoft may continue to serve African customers through its sales teams, cloud services, and other operations. But the closure of both engineering centres raises questions about the company’s long-term commitment to developing local technical talent. The people losing their jobs are not the only ones affected. Governments and universities that treated Microsoft’s investment as evidence of Africa’s place in the global technology industry must now reckon with its withdrawal.

The cuts also feed a narrative that AI will soon become central to how businesses operate. AI companies have raised enormous sums on the promise that their technology will change work, and layoffs attributed to AI give investors another reason to believe that promise.

Layoffs are becoming part of the AI sales pitch

Microsoft is not alone in using AI to explain workforce reductions. Amazon eliminated 14,000 corporate jobs in October 2025 and another 16,000 in January 2026, while Salesforce has linked its AI agents to a reduced need for customer support staff after cutting about 4,000 roles. In February 2026, Salesforce also laid off fewer than 1,000 employees across several functions. Across software, customer service, and other knowledge-based industries, AI has prompted companies to reconsider how many people they employ.

Some of these changes are real, given that AI tools can write code, summarise documents, answer customer queries, and handle repetitive administrative tasks. Companies have good reason to test whether these tools can reduce their costs, and some jobs will disappear as businesses automate work. It would be wrong to dismiss every AI-related redundancy as a publicity exercise.

The problem is that companies rarely separate the work AI can already perform from the work they expect it to perform in the future. A business might cut jobs because demand has weakened, teams overlap, or management wants to improve margins. AI may contribute to the decision, but it can also provide a convenient explanation for restructuring that would have happened anyway.

Cutting staff can improve a company’s financial results immediately, while the productivity gains promised by AI may take years to materialise. A company can reduce its payroll and report better margins before proving that its AI investments will generate enough money to cover its costs. The promise of future productivity can therefore help justify decisions whose immediate financial benefits come from employing fewer people.

There is also a commercial benefit to making AI appear unavoidable. When Microsoft, Amazon, or Salesforce announces cuts linked to AI, other executives face pressure to show they are preparing for the same future. Businesses fear falling behind, workers worry about their jobs, and investors look for companies positioned to profit. The announcements help sell AI as a necessity rather than a technology whose value must be judged against its cost.

There is no evidence that companies are coordinating layoffs to drive up demand for AI stocks. But the language surrounding the cuts can still benefit an industry whose valuations depend on investors believing that AI will become indispensable. That is worth examining, especially as two of its biggest players prepare to approach public markets.

Moonshot 2026 is almost here! Get 30% off all tickets during our FLASH SALE from October 16–24 and join us in Lagos on October 28–29.

Get your tickets now!

The IPO is where the story gets expensive

OpenAI announced a funding round involving $122 billion in committed capital in March 2026, valuing the company at $852 billion after the investment. Microsoft has committed $13 billion to OpenAI through their partnership since 2019, of which $11.9 billion had been funded by June 2026, according to Microsoft’s annual filing. These figures cover different parts of OpenAI’s financing and should not be treated as separate investments.

Anthropic has attracted similarly large sums, raising $65 billion in May 2026 at a post-money valuation of $965 billion. That same month, the company reported an annualised revenue run rate above $47 billion.

A public listing would give the companies access to a broader pool of investors and allow some existing shareholders to sell their stakes. Investors who backed them privately could realise substantial returns if the market values them at levels above their previous funding rounds. The story they sell to public investors will need to justify those valuations.

OpenAI has reported rapid revenue growth, but it has yet to establish sustained profitability. Its spending on computing infrastructure and model development remains high, and projections suggest it may not generate positive cash flow until 2030. Revenue growth alone does not indicate whether the company can earn enough to cover the costs of developing and operating its systems.

Anthropic reported $16.8 billion in third-quarter revenue in October 2026 and adjusted profitability for a second consecutive quarter, although its growth has slowed from earlier in the year. Adjusted profitability does not tell investors everything about the company’s long-term costs or whether it can sustain its margins.

Both OpenAI and Anthropic have found strong demand for their products, and AI already has practical uses in software development, research, and customer support. They still need to prove they can turn that demand into profits and justify what investors are paying for them.

The blockchain boom of the 2010s offers a useful comparison, as investors were told the technology would change finance, agriculture, and supply chains, yet many promised applications never reached mainstream use. Blockchain had practical uses, but that did not make every company built around it a sound investment. AI could have a far greater commercial impact, but its potential does not guarantee that every company selling it will succeed.

Microsoft’s African closures do not prove a plan to inflate AI valuations, but they show how the company can abandon a long-term investment in local engineering while the industry it belongs to asks investors to believe AI will define the future of business. More than 400 engineers are losing their jobs, while investors are being offered a chance to profit from the technology being presented as the future of work.

The market should judge AI companies by the money they make, not just the money they raise or the jobs their customers cut. By the time OpenAI and Anthropic reach public markets, investors will have heard plenty about AI’s ability to replace workers. They will need to ask a harder question about whether the businesses promising that future can make enough money to justify their valuations.

Kenn Abuya

Senior Reporter, TechCabal

Thank you for reading this far. Feel free to email kenn[at]bigcabal.com, with your thoughts about this edition of NextWave. Or just click reply to share your thoughts and feedback.