Africa banks Microsoft outage: lenders had no real backup plan
The Africa banks Microsoft outage on Friday left lenders, hospitals and public bodies without email or insurer sign-off, and it showed that few had a working plan B. I was at Avenue Hospital in Nairobi when every screen went dark. The receptionist apologised, and then apologised again. Insured patients could not be attended to, because the hospital required sign-off from the insurer, and that sign-off depended on systems built on Microsoft. Meanwhile, at a location in the US, Microsoft was in the middle of a dreadful day.
We simply sat there, a waiting room of people stalled by a technical fault that not one of us had ever given a thought to. When lunchtime came, my phone was crowded with text messages from banks all saying the same thing: their email was down, and would I please phone them. Had anybody prepared for this?
What went wrong, and who was hit
Thursday's disruption started at about 4 PM GMT and reached users across 74 countries. Microsoft put the fault down to configuration changes pushed to its servers following an update. A component responsible for establishing secure connections was swamped, servers could no longer process incoming requests, and the company began reversing the change.
Downdetector ranks it as the biggest outage of any service it has tracked in 2026 (the writer should add the figure and a link). For African firms, the timing could hardly have been worse. Reports of problems rose between 8 AM and 11 AM in each affected country on Friday, just as workplaces across Africa were starting their working day.
Customer notifications soon followed. The affected institutions included:
- Kenyan customers also received messages from at least seven lenders, among them Equity, Standard Chartered, Stanbic, UBA, I&M, Loop and the digital-only arm of NCBA.
- In Nigeria, customers of Wema Bank and Stanbic IBTC were informed that email services were experiencing faults.
- In South Africa, Absa, Eskom, FlySafair and the SABC were impacted, while the South African Revenue Service cautioned taxpayers that its own services were also disrupted.
Why pay a price for doing business globally?
The Verto Card lets you spend straight from 11 wallets held in different currencies, with no FX charges when you pay in 10 major currencies. It works at over 150M Visa merchants around the world.
Banks swapped one American platform for another
Microsoft first admitted to the fault late on Thursday. I&M published its statement on Friday, and Loop sent texts to customers by midday. Over 15 hours therefore separated Microsoft's acknowledgement from some banks pointing customers to alternatives. Had a plan been practised in advance, the response could have come sooner.
Almost every notice banks sent to customers listed the same alternatives: call the bank, message it on WhatsApp, or get in touch with your account officer. Wema pointed customers with pressing needs towards its telephone numbers, WhatsApp and Instagram. Meta owns two of those three platforms.
Banks across Africa responded to the failure of one US platform by steering their customers towards another. Those WhatsApp numbers and chatbots are approved, well documented and competently run, and they did what was asked of them on Friday. Even so, Meta cannot simply stand in for Microsoft.
Both are American behemoths with records of outages. A lender that relies on one when the other fails have divided its exposure between two companies rather than removing it. Consider what would have happened had WhatsApp gone down that same morning.
Why the Africa banks Microsoft outage was tolerated
The statements issued by the lenders were keen to stress that payment rails and mobile banking apps had kept working. According to the report, no bank in Kenya saw those services go down. But that told only one side of the picture. Much of everyday banking now depends on email, whether it is customers asking for loans, requesting statements or lodging complaints.
Transfers and payments carried on flowing even as other services ground to a halt. If you have ever rushed to get a statement from your bank before a visa appointment on a Friday, you will know how quickly an empty inbox that stays quiet can become a genuine headache.
It may even be rational to have no true backup. Maintaining a spare email platform that is regularly drilled and manned is a monthly expense, and it might justify itself on perhaps one day in a year. If all the lenders go down together, each one's standing is shielded as well.
When banks across three countries fail at the same moment, depositors do not shift their money elsewhere, and no senior executive loses a post. A bank that sets up its own contingency and then sees it collapse shoulders the blame alone. Depending on one supplier remains inexpensive because every rival does likewise.
Directors may well be willing to tolerate that risk, yet they rarely admit as much in public. The continuity plan pledges resilience, the regulator tucks it away, the auditor approves it, and then, when a genuine crisis arrives, it collapses; in the end, everything hangs on a single telephone number.
Africa's top 10 founders. One platform. $1.5 million.
Kigali will host the grand finale of the 8th Africa's Business Heroes (ABH) competition!
A total of US$1.5 million in grant funding is up for grabs, and ten of Africa's foremost entrepreneurs, who come from 8 different countries, will go on stage to vie for a share of it.
The next outage
The disruption on Friday might easily have been far more serious. The bulk of the trouble affected email, and companies have long learned to tolerate sluggish email. Patients had to wait a little longer than normal and bank customers phoned rather than wrote, yet clinics did not close and branches carried on serving the public. The harm could be counted in hours.
Whichever outage comes next will hit workplaces that look quite unlike today's. Microsoft intends to place Copilot and its AI assistants inside those same company accounts that went offline this week. There they would carry out tasks that humans handle at present, such as signing off insurance claims and replying to customers. Copilot was also down on Friday.
Suppose that by 2028 an insurance firm has shifted its entire claims department to an agent operating in Microsoft 365. A comparable outage would not merely cause sluggish email. Claims would cease to be processed, and the employees who used to do that job manually might already have been dismissed.
People are what carried Africa through Friday. Bank staff, receptionists and call-centre workers replied to WhatsApp messages using their own handsets. Automation is designed precisely to do away with that sort of reserve human capacity.
Hosting data locally would not have made a difference
Some will argue that a data centre built at home would have spared us this. Consider the proposed Microsoft–G42 facility in Kenya, intended to anchor an East African Azure hub. It has stalled, because Nairobi declined the financial backing the two companies were seeking.
And had it been operating, Olkaria's servers would have received the identical faulty update and failed just as badly, with the only difference being a prettier Rift Valley backdrop. The physical location of the servers was never the weak point. Reliance on a single company is.
Directors and the monetary authorities have just been given a free emergency drill, and they ought to use it. After the Africa banks Microsoft outage, they should ask:
- When did their business-continuity planning last picture Microsoft failing everywhere simultaneously?
- How long did the changeover on Friday take, and who authorised it?
- How many customer exchanges landed on applications that the lender has no control over?
In most meeting rooms, the truthful replies will be never, the better part of the day's first half, and more than anyone can count.
Once AI systems have replaced the people at the help desk, and the following failure comes around, who will still be there to answer the phone?




