When Uber ended operations in Nigeria in early September, it closed a business that had been operating for 12 years. Uganda followed after roughly a decade on the platform. The company said the decisions followed a review of its operations and stressed that they were specific to those markets rather than signalling a wider withdrawal from Africa, but it did not disclose the financial performance of either business or explain in detail what had made continued operation unattractive.

That leaves considerable room for interpretation, particularly because the exits have come at a moment when Uber is committing increasingly large amounts of capital and management attention to autonomous mobility.

The temptation is to connect the two developments directly: Uber is preparing for a driverless future in Europe and North America while pulling out of conventional ride-hailing markets in Africa, therefore one must somehow explain the other. Yet there is no inherent contradiction in the company doing both. An investment in autonomous mobility is a technology and capital-allocation strategy with the possibility of being deployed across multiple markets. Leaving Nigeria or Uganda is a decision about whether individual operations continue to justify the resources required to sustain them.

Indeed, a company prepared to invest billions of dollars in autonomous vehicles has every reason to become more discriminating about where the rest of its capital and management attention are deployed.

The more useful question raised by Uber's departure is therefore not whether Lagos has somehow lost out to London in a global race towards driverless taxis. It is what the exits reveal about the economics underneath ride-hailing in African cities, where substantial passenger demand can coexist with a transport system in which the productive asset—the vehicle itself—is expensive to acquire, expensive to finance and increasingly expensive to keep on the road.

For years, much of the argument about ride-hailing economics has centred on commission. Drivers complain that platforms take too much from each fare; platforms argue that commissions pay for technology, customer acquisition, payments, safety systems and the infrastructure required to create a functioning marketplace. New entrants frequently distinguish themselves by offering lower commissions or different charging models.

Whether the platform takes 25 per cent, 20 per cent, 15 per cent or considerably less certainly affects what remains in the driver's pocket. But the commission debate has also encouraged an incomplete understanding of what the driver is actually operating.

The driver does not operate a percentage. He operates a depreciating commercial asset.

The car has to be acquired before it earns its first fare, and for many drivers that means somebody has to finance it. It must then be insured, fuelled, serviced and repaired. It consumes tyres, brake pads, filters, lubricants, suspension components and spare parts. The driver requires a smartphone and a continuous data connection simply to participate in the platform. Licensing and compliance have to be maintained, accidents create repair bills and downtime, while every kilometre travelled depreciates the vehicle whether there is a paying passenger inside it or not.

Fuel moves faster than fares

Those costs have become particularly difficult to absorb in Nigeria. By April this year, ride-hailing drivers in Lagos were reporting petrol prices of approximately ₦1,300-₦1,330 a litre, equivalent to about US$0.98-US$1.00, compared with roughly ₦800, or about US$0.60, at the end of February. In a matter of weeks, one of the largest variable costs in the driver's business had therefore increased by more than 60 per cent.

The effect on the operating margin was immediate. One Lagos driver told a local publication that when petrol was selling at around ₦800 a litre, spending ₦20,000—approximately US$15—on fuel could support gross earnings of between ₦50,000 and ₦60,000, or roughly US$38-US$45. By April, with petrol selling at between ₦1,300 and ₦1,330 a litre, the same fuel expenditure was producing only about ₦30,000-₦35,000, approximately US$23-US$26, even on what he regarded as a good day.

Another driver described much the same deterioration, saying that ₦20,000, or about US$15, in fuel was struggling to generate ₦30,000, approximately US$23, in revenue, compared with at least ₦45,000, or around US$34, previously.

The experience was not universal. Another Lagos driver said that by concentrating on Uber, being selective about the journeys he accepted and positioning himself around stronger periods of demand, he could spend approximately ₦50,000, or US$38, on fuel and generate as much as ₦150,000, approximately US$113, in revenue on a favourable day.

That variation is important because it shows why there is no single ride-hailing-driver economy. Vehicle efficiency, location, hours worked, trip selection, demand density and the ability to avoid long unproductive journeys can produce materially different results for drivers using the same platform in the same city. It also illustrates why commission alone tells us relatively little about whether the underlying asset is profitable.

A driver can keep a larger share of every fare and still operate an economically weak vehicle.

The cost begins before the first passenger gets in

The acquisition price makes the equation considerably more difficult. By August, marketplace data reported in Nigeria put the average asking price of a selection of 10-year-old economy cars—including models such as the Toyota Corolla, Nissan Almera, Honda Civic and Hyundai Accent—at around ₦15 million, approximately US$11,300. These are not new premium vehicles but decade-old cars of the type that would ordinarily occupy the more accessible end of a commercial passenger-vehicle market.

For a driver who already owns such a car outright, there may be no monthly instalment, but that does not make the vehicle free. Capital remains tied up in a depreciating asset that accumulates unusually high mileage and will eventually require replacement.

For a driver who does not own the car, the cost becomes much more immediate. Somebody—the driver, a fleet owner, leasing company, informal financier or hire-purchase provider—has supplied capital and expects to be paid for it.

That financing occurs in an economy where the central bank's benchmark interest rate remained at 26.5 per cent in July. A ride-hailing driver does not necessarily borrow at the benchmark rate and may face substantially different terms depending on the lender and financing structure, but the figure provides an indication of the broader cost-of-capital environment in which commercial vehicles have to be funded.

It is here that the conventional description of ride-hailing as a transaction between a passenger, driver and technology platform becomes inadequate. There is often a fourth economic interest behind the journey: whoever owns or financed the vehicle. Sometimes that is the driver himself, sometimes it is not, but the asset still has to generate enough economic value to compensate labour, operating expenditure and capital before there is a sustainable return for the platform sitting above it.

This is also why looking at the economics of an individual trip can be misleading.

For the passenger, the transaction begins when the car arrives and ends when the passenger steps out. The economics of the vehicle began earlier. The driver may have travelled several kilometres to reach the pickup point. Those kilometres consumed fuel, tyres, time and vehicle life without generating a fare. The paid journey then ends somewhere else, where the driver may or may not find another passenger. A seemingly attractive fare to the outskirts of the city can become much less attractive if the driver has to travel a substantial distance empty before finding the next booking.

The app records the booked journey. The vehicle owner experiences the entire operating cycle.

The better unit of analysis is therefore not simply the fare or even the paid kilometre. It is the vehicle-day: the total time, mileage and operating expenditure required to produce a day's revenue.

The road eventually arrives in the maintenance bill

Once the business is viewed that way, congestion ceases to be merely an inconvenience and becomes a financial variable. A commercially operated car earns partly through utilisation. A vehicle capable of completing three paid journeys during a particular period has fundamentally different economics from one capable of completing six, even if the fare per kilometre is identical. When traffic leaves the car crawling or stationary, depreciation continues, fuel is still consumed, the driver's time remains occupied and finance payments do not stop. What disappears is productive throughput.

The fare can be adjusted to compensate for some of that loss, but only up to a point because the passenger has an economic limit too. This is particularly important in African cities where ride-hailing competes not simply with another private-car service but with buses, minibuses, matatus, danfos, boda-bodas, shared taxis and other informal or semi-formal transport modes. If fares rise sufficiently, passengers do not necessarily absorb the increase. They change modes.

Ride-hailing therefore operates between two boundaries. Below it sits the cost required to put the vehicle on the road and keep it there; above it sits the amount the passenger is willing and able to pay. When vehicle, fuel and financing costs rise faster than household purchasing power, the commercially viable space between those boundaries narrows.

A platform can alter commission, introduce incentives or use dynamic pricing to move money around within that space. It cannot manufacture purchasing power at one end or permanently eliminate operating costs at the other.

Road condition introduces another cost that rarely appears in debates about ride-hailing pricing because it is not deducted from a driver's account immediately after a journey. It accumulates.

A kilometre is not always economically equivalent to another kilometre. A vehicle travelling ten kilometres over a good urban road encounters a different maintenance environment from one travelling the same distance over badly deteriorated surfaces. Tyres, wheel alignment, shock absorbers, bushings, bearings and other suspension components absorb the difference, while slower speeds and repeated acceleration can affect fuel consumption.

The passenger may have paid for the trip weeks before the cost eventually appears as a repair bill.

This relationship is sufficiently established that road economists explicitly incorporate pavement condition into models of vehicle operating cost. For a commercial passenger vehicle, the implication is straightforward: poor infrastructure is not merely a public-sector inconvenience. Eventually it becomes private operating expenditure.

Ride-hailing can actually magnify this exposure because one of its great strengths is its ability to reach passengers beyond the routes and stops served efficiently by conventional public transport. An application can locate someone several kilometres inside a neighbourhood, calculate the route and dispatch a driver almost instantaneously. The technology has solved the information problem of finding the passenger.

It has not improved the road the vehicle must travel to reach them.

Kampala shows the same problem in a different vehicle

Uganda offers another illustration. Research into Kampala's platform-mediated boda-boda economy found that riders using digital platforms generated about 12 per cent more gross weekly income than riders operating outside them. At first sight, that is exactly what a digital mobility marketplace is supposed to achieve: connect riders to more customers, increase utilisation and generate more income.

The picture changed after costs. Once commission and the additional fuel and mobile-data expenditure associated with platform work were taken into account, the research found that platform riders' net income was approximately 7 per cent lower. Both platform and non-platform riders were working an average of roughly 72 hours a week.

The significance goes beyond motorcycles or Kampala. It demonstrates that increasing the volume of economic activity flowing through a platform does not necessarily increase the economic surplus retained by the person supplying the labour and vehicle. More trips are valuable only when the additional revenue generated by them exceeds the additional cost of producing them.

Platform participation itself can also require capital. The Kampala research recorded equipment expenditure associated with joining different services. SafeBoda riders reported average equipment expenditure of UGX427,000, approximately US$120 at the exchange-rate basis used in the study; Bolt riders reported UGX247,000, around US$70; and Uber riders UGX399,000, approximately US$112. A registration charge of UGX50,000, or around US$14, was also commonly reported.

The absolute numbers are obviously very different from the cost of acquiring a car, but they illustrate the same underlying principle. Digitalisation does not remove the physical cost of supplying transport. In some cases it introduces additional costs that greater utilisation must first recover before the driver is genuinely better off.

Uber may be asset-light. The transport system isn't.

This is where one of the most persistent descriptions of the ride-hailing business can become misleading. Uber is an asset-light company in the sense that it does not have to own the millions of conventional vehicles operating through its marketplace. That has been one of the great economic innovations of the model: an enormous global transport network can be assembled without putting every Toyota, Volkswagen or motorcycle required to operate it onto Uber's own balance sheet.

The mobility system itself, however, remains profoundly asset-heavy.

Somebody still bought the car. Somebody financed it. Somebody insures it. Somebody replaces the tyres, pays for accident repairs and absorbs the depreciation. Somebody carries the cost when the vehicle spends two days in a garage or four hours crawling through traffic.

Removing those liabilities from Uber's balance sheet does not remove them from the economics of an Uber trip. It changes who carries them.

That distinction is particularly important when comparing markets. Vehicle leasing works best when it sits inside an ecosystem capable of understanding and managing the asset over its entire life. Mature fleet markets have established credit systems, predictable insurance products, dense maintenance networks, roadside assistance, reliable spare-parts distribution, telematics, professional accident management and developed used-vehicle markets from which financiers can estimate residual values with reasonable confidence.

The vehicle plugs into an existing machine.

Across many African markets, individual components of that machine exist and some are developing quickly, but they remain fragmented or expensive. A driver or small fleet owner may therefore become the integrator of almost everything underneath the platform: finding the car, arranging finance, obtaining insurance, managing servicing, negotiating with mechanics, sourcing parts, maintaining documentation, buying data, absorbing downtime and dealing with roadside enforcement.

The technology platform can be extraordinarily sophisticated while the productive infrastructure beneath it remains improvised.

That is why simply reducing commission, while helpful, cannot solve every economic weakness in the model. Imagine that a platform charged only 3 per cent and left the driver with 97 per cent of the fare. The improvement would be substantial. Yet the vehicle would still have to be bought or financed, petrol would still have to be purchased, tyres would still wear, maintenance would still be required, congestion would still reduce utilisation and empty return kilometres would still consume fuel without generating revenue.

If the total cost of supplying a kilometre of transport is already approaching what passengers can sustainably pay for that kilometre, leaving the driver with 97 per cent of an inadequate fare does not repair the underlying business.

This is not an argument for ignoring commission. Platforms determine important parts of the economic relationship: fares, incentives, commissions, access to customers and, increasingly, the information through which drivers decide whether a journey is worth accepting. Driver complaints about pricing transparency and fares that fail to adjust quickly enough when fuel and maintenance costs rise deserve serious scrutiny.

It is instead an argument for recognising that the percentage taken by the platform is only one layer in a much larger cost structure.

What autonomous vehicles actually tell us

Seen this way, Uber's autonomous-vehicle strategy is less a competing explanation for its African withdrawals than an interesting comparison with them.

Autonomy does not make the vehicle disappear. It makes the infrastructure surrounding the vehicle even more important.

Uber has been assembling partnerships with autonomous-driving companies, vehicle manufacturers and fleet operators rather than attempting to build every component itself. Its emerging autonomous ecosystem involves companies supplying the vehicle, autonomous-driving technology, fleet operations and marketplace demand in different combinations.

A robotaxi may remove the conventional driver from the front seat, but it does not remove financing, insurance, maintenance, tyres, cleaning, charging or fuelling, depots, communications, accident management, roadside support or fleet management. It adds expensive sensors, computing hardware, autonomous-driving software, remote assistance and an unusually demanding regulatory framework.

In other words, autonomous mobility requires an ecosystem.

The cities in which these systems are first being deployed tend to have much of that surrounding architecture already available or capable of being assembled at scale. That is a fundamentally different strategic proposition from operating a conventional ride-hailing marketplace in Lagos or Kampala, and it is why the two decisions should not be lazily collapsed into a single story about Uber abandoning African drivers for driverless cars.

What the comparison does reveal is how much of the African mobility challenge remains below the application.

Over the past decade, enormous progress has been made in digitising the relationship between passenger and vehicle. A customer can locate a car, see its estimated arrival time, follow its route, communicate with the driver, pay electronically and rate the experience using technology that would have appeared extraordinary not very long ago.

Much less progress has been made in fundamentally changing what that car costs to own and operate.

The next African mobility opportunity sits below the app

That is where the next African mobility opportunity may lie.

Rather than another application competing principally through lower commission, the more consequential innovations may be vehicles financed specifically for high commercial utilisation; lease repayments linked to verified platform income; insurance priced using telematics and actual usage; maintenance, tyres and accident management incorporated into vehicle contracts; professional fleet operators capable of spreading fixed costs across hundreds or thousands of cars; locally assembled vehicles with reliable parts support; purpose-built electric vehicles; battery leasing; structured charging networks; and deeper used-vehicle markets capable of producing credible residual values.

The financing model itself may have to change. A conventional consumer loan assesses an individual borrower and asks whether that person can repay a vehicle over several years. A mobility-finance model can increasingly assess the productive asset: how many kilometres it travels, how much revenue it generates, where it operates, how it is driven, whether it is maintained and what its likely residual value will be.

The better the ecosystem becomes at measuring those things, the less risk has to be priced blindly into the cost of the vehicle.

And governments have a role that extends well beyond regulating the platforms. Road quality affects vehicle operating cost. Traffic management affects utilisation. Import duties and taxes influence acquisition cost. The predictability of licensing and roadside enforcement affects operating risk. Energy policy determines fuel and electricity costs. Financial regulation affects the availability and price of capital.

These are not separate issues once the objective is to produce an affordable commercial kilometre.

The market can exist even when Uber leaves

None of this establishes that vehicle economics caused Uber to leave Nigeria and Uganda. Uber has not released the information required to make that conclusion, and it would be wrong to present inference as corporate fact.

Nor does Uber's departure mean ride-hailing cannot work in either country.

Nigeria and Uganda still have substantial urban transport demand and competing platforms. Other companies may have different overheads, stronger local positions, different vehicle categories, lower customer-acquisition costs or simply a willingness to operate at margins that do not meet Uber's internal requirements. Local operators may understand particular markets differently or structure their businesses around motorcycles, cash payments or fleet partnerships that alter their economics.

A market can remain viable while becoming unattractive to one participant.

That distinction becomes especially relevant now that Uber itself is a very different company from the venture-funded business that once treated geographic expansion almost as an end in itself. In the second quarter of 2026, the company reported US$58 billion in gross bookings, including almost US$29 billion from Mobility, and generated approximately US$2.8 billion in free cash flow during the quarter. Trailing 12-month free cash flow exceeded US$10 billion for the first time.

A company producing that level of cash has the ability to make choices about where it believes future returns will be highest.

The question for Africa should therefore be larger than why Uber left two countries.

It should be what minimum operating conditions allow a mobility platform, the driver, the vehicle owner or financier and the passenger to coexist economically over the long term.

Can an appropriate vehicle be acquired at a price that its daily earnings can realistically amortise? Can it be financed at a rate compatible with transport margins? Can insurance be priced predictably? Can maintenance and tyres be incorporated into structured operating costs rather than arriving as financial emergencies? Can spare parts be supplied quickly enough to minimise downtime? How many kilometres does the vehicle travel without a passenger? How many paid journeys can it complete during a working day? How much revenue disappears into fuel? What does congestion do to utilisation? What does road condition do to maintenance? How much unpredictability does enforcement introduce? And, after all of those costs have been accounted for, is the resulting fare still affordable to the passenger?

That is the economic system underneath the app.

For much of the past decade, African ride-hailing has understandably concentrated on the extraordinary technological achievement at the top of that system. Platforms solved a real problem. They organised fragmented transport supply, gave drivers access to customers they could not previously reach efficiently and gave passengers a degree of convenience, transparency and safety that conventional taxi markets often struggled to provide.

The next challenge is more physical and considerably less glamorous.

It is making the productive asset underneath that technology cheaper to finance, more predictable to maintain and more productive for every hour that it is on the road.

Uber can pursue autonomous mobility in London, Los Angeles or elsewhere while deciding that Nigeria or Uganda no longer meets its requirements. There is nothing inconsistent about that. One is a bet on where mobility technology is going; the other is a judgement about the economics of operating in a particular market today.

The more important lesson from the exits is therefore not that Africa lacks a mobility opportunity. The demand is plainly there.

It is that demand alone does not make a transport business work.

For that, the fare has to support the driver, the platform and the capital tied up in the vehicle, while remaining affordable to the passenger. And beneath every debate about whether the platform should take 25 per cent, 15 per cent or 3 per cent sits a collection of costs that no commission policy can make disappear: financing, fuel, insurance, tyres, maintenance, parts, depreciation, data, congestion, downtime, empty kilometres and the condition of the road itself.

We have become very good at putting sophisticated technology on top of African transport.

The harder task is fixing the operating system underneath it.