×
Tech

Microsoft to shut Nairobi engineering centre, putting hundreds of jobs at risk

Microsoft is shutting its Nairobi engineering centre

Microsoft is winding down its Africa Development Centre (ADC) in Nairobi, putting nearly 400 jobs at risk and reversing a seven-year effort to make Kenya a hub for building technology for global markets.

Two employees, who spoke on the condition of anonymity, told TechCabal that staff at the centre, housed in Dunhill Towers in Nairobi’s Westlands district, were called to a meeting on Monday morning and asked to meet with their managers by Friday, October 16, to discuss exit options, including severance terms.

The closure comes seven years after the company pledged more than $100 million to develop African tech talent, to make Nairobi one of its engineering hubs outside the US. It would unwind a project built on the promise that African engineers could build products for global markets without leaving the continent, raising questions about Microsoft’s long-term plans for engineering in Africa.

“We are communicating directly with employees in Kenya about proposed changes to our engineering operations,” a Microsoft spokesperson told TechCabal on Monday. “We recognise the uncertainty this brings and are keeping those discussions private. Microsoft remains committed to Kenya and will continue to have employees in the country while supporting the country’s ongoing digital transformation and innovation ambitions.” 

The statement did not confirm the centre’s closure or specify how many employees would remain or when the changes would take effect.

Two other employees, who asked not to be named as they aren’t authorised to discuss the matter, told TechCabal they expect their meetings with supervisors to focus on severance terms rather than on whether they will stay.

“Our case is sealed; we just didn’t know when,” one employee said.

Another person familiar with the matter said fewer than 10 employees are expected to remain, including staff in sales and recruiting, two in Microsoft Customer and Partner Solutions (MCAPS), which manages customer and partner relationships, and employees working on AI for Good, Microsoft’s initiative to apply AI to social and environmental challenges.

When Microsoft launched the Africa Development Centre (ADC) in 2019, Michael Fortin, then a corporate vice president, said it would allow engineers to do “meaningful work from their home countries” while working within a global engineering organisation. Microsoft planned to hire 100 engineers by the end of 2019 and grow the workforce across Nairobi and Lagos to 500 by 2023. 

L-R: Jack Ngare, managing director Microsoft Africa Development Centre – Nairobi, and former Kenya president Uhuru Kenyatta during the opening of the ADC in 2019. Image: ADC Kenya

By December 2022, Microsoft President Brad Smith said the Nairobi centre employed more than 450 engineers. The centre had met its original hiring ambition, making it harder to explain the closure as a failure to deliver on its initial promise.

The Nairobi closure follows Microsoft’s May 2024 decision to shut its sister centre in Lagos, a move the company said at the time would not affect its operations in Kenya. The Lagos closure caught staff by surprise, while Nairobi employees say they had seen signs of trouble for some time.

On Monday, Kenya’s ICT Minister William Kabogo said he met a Microsoft delegation led by Jeff Bullwinkel, the company’s vice president and deputy general counsel for Europe, the Middle East and Africa, with its East Africa team headed by Angela Ng’ang’a.

Microsoft briefed him on “its evolving operations in Kenya in the context of its global restructuring,” Kabogo said, while he pressed the company on “protecting jobs at Microsoft’s Nairobi engineering centre.”

The company reaffirmed its long-term commitment to Kenya, he said, and both sides agreed to draft a joint action plan. Neither has disclosed what the plan will contain or whether it could change a decision employees describe as final.

Microsoft has yet to disclose why it is closing the Nairobi centre, and people familiar with the matter point to several possible factors. The decision comes amid a wider round of cuts at the company, which announced about 4,800 job cuts in July 2026, or roughly 2.1% of its workforce, following about 15,000 cuts in 2025.

The reductions come as Microsoft pours record sums into AI infrastructure, raising questions about which parts of its business will continue to receive funding. A remote engineering centre could come under pressure, though no public evidence links the Nairobi closure directly to the global cuts.

Kenya has also struggled to advance another major Microsoft initiative. The $1 billion data centre that Microsoft planned with Abu Dhabi-based G42 in Olkaria has stalled amid concerns about electricity capacity and disagreements over the Kenyan government’s ability to guarantee payments for the facility’s capacity. President William Ruto has cited the country’s limited power supply as a problem. However, government officials have said the project has not been abandoned.

A third possible factor is the cost of maintaining a large local workforce, particularly its tax bill. One employee told TechCabal that Microsoft’s Kenyan employees generate at least KES 1.5 billion ($11.6 million) in annual Pay As You Earn (PAYE) taxes for the Kenya Revenue Authority. TechCabal could not independently verify the figures.

The closure would return hundreds of engineers to a job market Microsoft helped reshape when it opened the centre in 2019.  The company recruited talent from Kenyan banks, telcos and technology firms to build its local engineering team. 

Seven years later, those employers and other technology companies will be among the places its departing engineers may look to next, raising questions about where the specialised talent built at the centre will go and whether Kenya can retain the skills and experience it took Microsoft years to assemble.

True scale demands moving beyond surface-level integrations to robust execution. We’ve filtered the noise out of Moonshot 2026, optimising the conference strictly for high-calibre connections between startup founders, global financial operators, enterprise leaders and individuals rewiring Africa’s technical frameworks. Get 20% off Early Bird tickets for a limited time.

§

On Wednesday, US chipmaker Nvidia became the first company to reach a market valuation of $5 trillion. The moment didn’t just mark another record; it confirmed Nvidia’s role as the foundation of global AI development and one of the most valuable companies ever created.

What makes this milestone remarkable is how fast it happened. Nvidia added another $1 trillion in value in less than 4 months after crossing $4 trillion in July 2025. Since the launch of ChatGPT in 2022, the company’s stock price has grown more than twelve times, pushing the S&P 500 to new highs and making Nvidia the heartbeat of AI investment worldwide.

On the day it crossed the $5 trillion mark, Nvidia’s stock rose by 4.5% to $210.11, bringing its total value to about $5.1 trillion. That jump alone equals the combined value of many Fortune 500 companies.

The company’s valuation now exceeds the total global cryptocurrency market and equals nearly half the size of Europe’s Stoxx 600 index. From hitting $1 trillion in 2023 to $5 trillion in 2025, Nvidia has become one of the most significant forces in global market growth and a core member of the Magnificent Seven, alongside Apple, Microsoft, Alphabet, Amazon, Meta, and Tesla.

How Nvidia’s revenue and AI demand keep growing

Numbers, not hype, back Nvidia’s $5 trillion valuation. The company’s growth comes from a full pivot to its most profitable segment, the Data Centre business, which powers the world’s AI systems.

In fiscal year 2025, Nvidia recorded $130.5 billion in revenue, a 114% jump from 2024. About $102 billion came from its Data Centre segment alone, making up nearly 80% of total revenue. In the last quarter of that year, this segment brought in $35.6 billion, demonstrating sustained demand from cloud providers, tech firms, and AI labs.

This shift proves that Nvidia has outgrown its image as a gaming chipmaker. It now serves as the infrastructure backbone for global AI operations. The company currently holds confirmed AI chip orders worth $500 billion, showing the strength of what analysts call its ongoing “AI super cycle.”

Even more convincing is Nvidia’s disclosure of a $500 billion backlog for upcoming Blackwell and Rubin chips through 2026. This isn’t a projection; it’s locked-in demand. Based on these figures, Nvidia expects Data Centre revenue of around $300 billion for fiscal 2027. For context, its Hopper line generated about $100 billion in revenue between 2023 and 2025.

How CUDA and hardware innovation keep Nvidia ahead

Nvidia’s $5 trillion market value rests on more than strong sales. Its biggest strength is the technological moat built around its hardware, software, and networking systems.

The CUDA ecosystem

At the centre of Nvidia’s dominance is CUDA, its exclusive software platform for parallel computing. CUDA connects Nvidia’s GPUs with tools and libraries that make AI training faster and more efficient. Libraries like cuDNN, TensorRT, and NCCL help developers run large AI models without performance issues.

This system creates a powerful lock-in effect. Once developers build AI models using CUDA, switching to another platform becomes expensive and time-consuming. Over time, this has given Nvidia control of over 90% of the AI training hardware market. The company’s technology is now used in finance, healthcare, and climate science, where rewriting CUDA-based code would cost millions.

This deep integration makes Nvidia’s software just as valuable as its chips. It keeps profit margins high and supports its market leadership in the global AI chip industry.

Mellanox and data centres

Nvidia’s success also depends on how well thousands of GPUs work together. The 2019 Mellanox acquisition gave Nvidia control of this communication layer. Mellanox technology connects GPUs at high speed, reducing bottlenecks in massive data centres.

Owning this interconnect system means Nvidia can deliver a complete AI infrastructure, from silicon and software to networking, ensuring its Data Centre business, now worth over $100 billion a year, runs efficiently at scale.

Hardware and cost efficiency

Nvidia’s H100 GPU shows why its products remain the top choice for AI workloads. The H100 trains AI models up to three times faster than the older A100. Although it costs more upfront, it reduces total expenses by cutting energy use and shortening training times.

Cloud providers now find the H100 more cost-effective per task, with about 40-60% lower costs per unit of work. For companies investing in AI, that efficiency makes Nvidia chips a more brilliant long-term choice and guarantees continued demand.

Nvidia’s next frontier beyond the GPU

Nvidia’s $5 trillion value didn’t come from GPUs alone. The company has expanded into new, high-growth markets, strengthening its control over the global AI ecosystem.

Omniverse and the rise of physical AI

Through its Omniverse platform, Nvidia is bringing AI into the physical world. CEO Jensen Huang calls it the “operating system for AI factories,” targeting what he describes as a $50 trillion physical AI market.

Omniverse lets companies create digital twins —virtual versions of real factories, vehicles, or data centres — to test and optimise before construction begins. It’s already helping industries design smarter systems and improve production efficiency. While Nvidia doesn’t list Omniverse revenue separately, it fuels growth in its Data Centre and Automotive divisions, with the latter earning $1.7 billion in 2025.

By positioning Omniverse at the centre of industrial automation, robotics, and manufacturing, Nvidia is laying the groundwork for the next decade of AI-driven infrastructure.

Partnerships and global reach

Nvidia’s strategy also depends on powerful partnerships. It recently announced a significant deal with OpenAI, including a $100 billion investment that ensures long-term chip demand. The company is also building seven supercomputers for the U.S. government, strengthening its role in America’s AI development.

Partnerships with firms like Palantir Technologies expand Nvidia’s reach into defence, logistics, and healthcare. At the same time, geopolitical discussions around the sale of its Blackwell chips to China show how closely the company’s success is tied to international policy. Any relaxation in export restrictions could instantly lift Nvidia’s market value.

How Nvidia’s partnership with Intel protects the CUDA ecosystem

Nvidia’s partnership with Intel marks a new phase of strategic cooperation. Both companies are working on data centre and PC products that connect their architectures through NVIDIA NVLink.

Intel will build custom x86 CPUs and system-on-chip integrated with Nvidia’s RTX GPU chiplets, while Nvidia invests $5 billion in Intel stock. This collaboration keeps the CUDA ecosystem central to enterprise and cloud computing. It also prevents competitors from building closed systems that exclude Nvidia technology, preserving the high margins from its Data Centre business.

Valuation integrity and future projections

The key question for investors is simple: Can Nvidia’s $5 trillion value last? The answer depends on how much of the future AI infrastructure market Nvidia can own.

Is Nvidia in a bubble?

Right now, most analysts remain confident. Out of 80 analysts tracking Nvidia, 73 rate the stock a “buy.” Their view is that global AI spending is only at the beginning. Forecasts suggest annual AI infrastructure investment could reach $3 to $4 trillion by 2030, about 40% higher each year than current levels. If Nvidia continues to power most of that infrastructure, its current value could be justified by strong long-term earnings.

Still, not everyone agrees. A few analysts warn that Nvidia’s success relies heavily on a handful of large buyers — Microsoft, Google, Amazon, Meta, Oracle, and OpenAI — who together are expected to spend around $400 billion on AI this year. If any of these companies reduce spending or entirely switch to their own chips, Nvidia’s growth could slow sharply.

Projecting the next five years

Nvidia’s AI infrastructure revenue is currently around $164 billion annually. Projections suggest it could grow to between $1.3 trillion and $2.2 trillion within five years, an 8- to 14-fold increase.

If these growth rates hold, Nvidia’s stock could rise from about $183 to between $1,300 and $2,100 over the same period. In more optimistic forecasts, prices could reach $3,100, representing a compound annual growth rate above 60%. Some analysts even see the potential for an $8 trillion valuation, but that depends on Nvidia expanding into newer markets like Omniverse and Automotive AI to reduce reliance on hyperscaler demand.

Web hosting by Somee.com