Hired in Nairobi, Decided Abroad: Microsoft's Engineering Cuts Expose Kenya's Tech Jobs BargainHired in Nairobi, Decided Abroad: Microsoft's Engineering Cuts Expose Kenya's Tech Jobs Bargain

The figures suggest a simple and uncomfortable reading. Kenya has become very good at hosting other people's technology work, from software engineering to content moderation to the promise of a giant cloud data centre, but it has very little say over when that work stays and when it goes. In 2026 all three pillars of that model have wobbled within six months.

Microsoft's Nairobi centre after seven years

Microsoft has told staff at its Africa Development Centre in Nairobi about "proposed changes to our engineering operations", and says the discussions with employees will stay private.

As Mwenendo reported, employees say the centre itself will close, while the company has confirmed only changes tied to a global restructuring and says it "remains committed to Kenya". It has not said how many jobs are affected.

The centre opened in 2019 as part of a combined investment of about KSh 12.98B ($100M) over five years across Nairobi and Lagos, with a target of 500 engineers. By Microsoft's own account in its May 2024 Kenya announcement the centre had 500 employees in Nairobi. Microsoft announced the closure of Lagos, its sister hub, in May 2024.

That timeline is the first warning. When a global company has two African engineering hubs and shuts one, the survivor is not safe; it is simply next in the queue for the same spreadsheet. Mwenendo's reading is that Nairobi's advantage was never permanent. It was a price advantage in a global cost review, and those reviews run every year.

Sama, Meta and the content moderation exit

An earlier blow landed in April. Sama, the outsourcing company that moderated content for Meta from Nairobi, issued a formal layoff notice covering 1,108 staff after Meta ended what Sama called a "major engagement at its Nairobi office". Sama has since moved its focus to services such as labelling data for artificial intelligence systems.

The lesson is the same as Microsoft's, from the other end of the pay scale. Content moderators and software engineers sit at opposite ends of the skills ladder, yet both groups learned that their jobs existed because a company abroad decided Nairobi was the cheapest good place to do the work. Neither group had a contract with Kenya; they had a contract with a budget line.

The Olkaria data centre that power could not carry

The remaining blow is the one Kenya came closest to controlling. In May 2024 Microsoft and the Abu Dhabi technology group G42 announced an initial investment of about KSh 129.79B ($1B), centred on a geothermal-powered data centre campus at Olkaria. Microsoft said a new East Africa cloud region would become operational within 24 months of definitive agreements being signed.

That deal stalled this year. President William Ruto said in May that "to switch on that one data centre, we would need to shut off power for half the country".

Later accounts of the Olkaria talks say they broke down after the companies asked the government to commit to buying a fixed amount of cloud computing capacity each year, a promise it could not make.

The first phase was meant to draw 100 megawatts, with an ambition of one gigawatt, set against a national grid of roughly three gigawatts. This is the key point. A data centre is the one kind of technology investment that cannot be packed up and moved by email, because it is built into the ground and wired into the grid. It was Kenya's best chance to anchor Microsoft physically. The country lost that bargaining chip over electricity, not talent.

Three exits, one budget logic

Put the three episodes together and a pattern appears. Each project was announced as proof that Kenya is Africa's technology capital. Each depended on a foreign board's view of costs. And in each case the Kenyan side had little to offer when that view changed: no ownership stake, no binding job commitment, and in Olkaria's case not enough spare power.

The development could mean a harder conversation about what Kenya counts as success. Headcount pledges make good headlines at launch events. They are worth far less if the intellectual property, the product decisions and the profits stay abroad, and the jobs leave with a single round of notices.

Kenya has seen this pattern among home-grown firms too. Twiga Foods' slide into administration showed that a well-funded local champion can also shed jobs quickly when outside money dries up. The common thread is dependence on capital and decisions that sit outside the country.

The case for Microsoft's Nairobi legacy

The strongest counterargument deserves a fair hearing. Microsoft says it will keep employees in Kenya and keep supporting customers across the region. Its cuts are part of a global restructuring rather than a verdict on Kenyan engineers. And seven years of hiring has produced hundreds of engineers trained to a global standard, people who can now build startups, join other firms or work remotely for employers abroad.

That spillover is real, and it is the best thing the centre leaves behind. But it is a by-product, not a plan. Mwenendo has found nothing in the public record showing that Kenya negotiated for it, measured it or tied incentives to it. A country that hopes skills will trickle out of foreign offices is relying on luck, and luck is not an industrial policy.

Kabogo's joint plan and the power test

ICT Cabinet Secretary William Kabogo has already pressed Microsoft to protect local jobs, and both sides have spoken of a joint plan covering AI adoption, digital government services and cybersecurity. The test of that plan is whether it contains anything binding: numbers of Kenyan roles, local research that stays local, or procurement tied to work done in the country.

The figures suggest the bigger test is electricity. As long as Kenya cannot promise reliable spare power, it cannot host the data centres that pin global technology companies to a place. Until it can, its technology jobs will remain what this year has shown them to be: rented, valuable and easy to cancel.