On June 18, 2026, Molo MP and National Assembly Finance Committee Chairperson Kuria Kimani stood before Parliament and declared that Kenya produces more than one million smartphones per year. He said local firms are creating jobs for youth, positioning Kenya as an emerging technology production hub, and that the country has progressed from being a pure technology consumer to a participant in production and export of digital devices.It sounded like a proud moment. And in some ways, it genuinely is.But there is a word that keeps getting quietly glossed over in these announcements: assembling. Kenya is not manufacturing smartphones. It is assembling them. And while that distinction is not a reason to dismiss the progress, it matters enormously when you start asking the harder questions, like whether these phones are actually built for Kenyan needs, whether they can compete with what is already on the market, and whether the entire initiative can survive as smartphone users grow more demanding.<h2>The Real Story Behind "Made in Kenya"</h2>To understand what is actually happening, you need to look at the two main operations driving these numbers.The first is East Africa Device Assembly Kenya (EADAK), a joint venture between Safaricom, Jamii Telecom, and Chinese firm Shenzhen TeleOne Technology. The plant is located in Athi River, Machakos County, and was officially launched by President William Ruto in October 2023. At launch, the facility was targeting a production capacity of up to 3 million smartphones annually and was already producing the Neon Smarta and Neon Ultra brands. In its first year of operation, EADAK announced it had produced 360,000 phones.<ad-spot></ad-spot>The second is M-KOPA's assembly facility, operated through a partnership with HMD (the Finnish company behind Nokia phones), also known as the East Africa Device Assembly Kenya plant in Nairobi's Industrial Area. This facility opened in January 2023, and by November 2024, M-KOPA reported it had assembled over 1.5 million smartphones, making it sub-Saharan Africa's largest smartphone assembly plant. The facility was producing roughly 100,000 phones per month, and M-KOPA has set a target of 10 million locally assembled smartphones by 2027.Together, these two operations are the backbone of Kenya's "smartphone production" story. And the jobs being created are real. M-KOPA has created 325 new jobs since 2023 in its factory, primarily for young Kenyans and first-time job seekers. These are meaningful contributions.But here is the part the press releases tend to skip: every single component in those phones is imported, primarily from China. The EADAK plant assembles up to 1.4 million smartphones annually from parts imported from China. M-KOPA's X20, their flagship locally assembled device, was developed in partnership with HMD, a Finnish company, using components largely sourced from abroad. The "assembly" process in Kenya involves putting together parts that were designed and manufactured elsewhere, typically in China, and packaging them for the Kenyan market.To understand just how deep this dependency runs globally, consider that even India, which has aggressively built its own domestic manufacturing industry, still has roughly 90 percent of components feeding its assembly plants coming from China. Vietnam, the world's second-largest smartphone export hub, also sources over 90 percent of its component value from China. So Kenya is not alone in this challenge. But it does mean that the "we are exporting smartphones" narrative needs to be held alongside the reality that Kenya is importing all the ingredients that make a smartphone a smartphone: chipsets, screens, batteries, camera sensors, and circuit boards.The government has also not made this any easier. While finished locally assembled devices are zero-rated for tax purposes, imported components attract 16 percent VAT, which manufacturers must then reclaim through a refund process that can tie up working capital for months. It is the kind of structural friction that quietly erodes the competitiveness of these assembly lines even when the political will is there.Kenya imported over 2.3 million smartphones directly from China in 2024 alone, a 71.5 percent increase from 1.3 million units in 2023. The country's total imports from China for electrical and electronic equipment reached over $685 million in 2024. These numbers tell you that regardless of what is being assembled locally, Kenyans are still overwhelmingly choosing to buy phones that arrive as finished products.<h2>So What Do These Locally Assembled Phones Actually Offer?</h2><img src="https://cdn.calebms.com/articles/0e440558-168e-4332-a45d-34aa786ac160.jpg" alt="Safaricom_Neon_Ultra_PR_photo.jpg" data-caption="Neon Ultra">This is where the conversation gets more specific and, frankly, more revealing.When EADAK launched, the two phones unveiled were the Neon Smarta, retailing at KES 7,499, and the Neon Ultra at KES 8,999. The government had originally promised phones at around KES 5,484 (US$40), so the final retail price was already higher than what was advertised. The Neon Smarta, which is the model most people interact with given its lower price point, comes with a 5-inch display, 2GB of RAM, 32GB of internal storage, a 5MP rear camera, a 2MP selfie camera, and a 3,000mAh battery. It runs Android 10 Go Edition, which is a stripped-down version of Android designed specifically for low-memory, entry-level devices.The Neon Smarta 2, an updated version, brought minor improvements, notably bumping the rear camera to 8MP and updating the software to Android 13 Go Edition. The price dropped to around KES 2,999, and the processor was upgraded to a Unisoc SC9863A octa-core chip. Still 2GB of RAM. Still 32GB of storage.<ad-spot></ad-spot>M-KOPA's X20, the company's flagship locally assembled device, is a notable step up from the Neon lineup. It features 6GB of RAM, 256GB of internal storage, a 50MP dual rear camera, a 50MP front camera, a 6.56-inch HD+ display, a 5,000mAh battery with 20W fast charging, and runs Android 14. On paper, those are respectable specifications even by 2026 standards. But the M-KOPA model is also tied to the company's "Lipa Mdogo Mdogo" financing structure, meaning the phone is accessed through daily or weekly payments rather than an outright purchase. The branding itself is a barrier for some users. One reviewer on a local phone listing site put it plainly: "But the name M-KOPA, it's like kila mtu anajua nimekopa!"<h2>The Smartphone Has Evolved. Have These Phones Kept Up?</h2>There is a bigger issue sitting underneath the specs debate, and it is one that most discussions about Kenya's assembled phones conveniently avoid.We are in 2026. The smartphone is no longer just a device you use to make calls, send WhatsApp messages, and occasionally scroll Facebook. For most Kenyans, the smartphone is the primary computer, the content creation studio, the business tool, the entertainment system, and the financial services terminal, all in one pocket-sized device.Consider what a typical Kenyan smartphone user actually does in a day. A boda boda rider uses their phone to navigate Google Maps and accept orders on apps like Bolt or Uber. A small business owner photographs products, manages inventory via M-Pesa business accounts, runs WhatsApp catalogs, and posts Reels to Instagram for marketing. A student attends virtual classes, downloads lecture materials, and collaborates on shared documents. A content creator edits short-form videos directly on their phone, uploads to TikTok or YouTube, and monitors analytics. A gamer plays titles like PUBG Mobile or Free Fire for recreation. None of these use cases are unusual. They are everyday life for millions of Kenyans.And here is the challenge: these use cases demand a specific caliber of hardware. Consistent video recording requires optical image stabilization and a capable ISP (Image Signal Processor) in the chipset. Heavy app multitasking requires at least 4GB of RAM, ideally 6GB or more, paired with LPDDR4X memory (not older, slower standards). Social media apps, especially those running short video feeds like TikTok and Instagram Reels, demand faster storage speeds, specifically UFS 2.1 or higher, rather than the slower eMMC 5.1 standard used in most entry-level phones. Mobile gaming requires a capable GPU and enough thermal management to sustain performance. Running multiple financial apps, business tools, and communication platforms simultaneously requires processing headroom that entry-level chips simply do not have.The Neon Smarta, with 2GB of RAM and a Unisoc chipset running Android Go, is a phone that can handle calls, basic messaging, and light internet browsing. It is a "first smartphone" device. But in 2026, it struggles to meaningfully run TikTok, handle Instagram Stories, or stay stable with more than three or four apps open. Android Go was designed with exactly these constraints in mind, which is why apps like Google Maps and YouTube have "Go" versions optimized to run on limited hardware. These are not the full-featured apps. They are the stripped-down versions.A phone retailing at KES 6,000 to 8,000 is not automatically bad. But when you consider what that same money buys in the broader market, the gap becomes very clear.<h2>The Redmi Question: What Are You Actually Competing Against?</h2><img src="https://cdn.calebms.com/articles/80e95559-652d-4070-ae23-c02d6c08a4f5.webp" alt="Redmi-A5-1.webp" data-caption="Xiaomi Redmi A5. Image Credit: Xiaomi">The Xiaomi Redmi A5 launched in Kenya in early 2025. Starting at KES 9,499 for the 3GB/64GB variant, it sits just slightly above the KES 6,000 to 8,000 range of the locally assembled phones. But what it offers is in a different conversation entirely.The Redmi A5 comes with a 6.88-inch IPS display with a 120Hz adaptive refresh rate, a 32MP rear camera with HDR support, an 8MP selfie camera capable of 1080p video recording, a Unisoc T7250 octa-core processor, up to 6GB of RAM with memory extension, up to 128GB of storage expandable to 1TB via microSD, a 5,200mAh battery with 15W fast charging, a side-mounted fingerprint sensor, USB-C connectivity, and Android 15 with Xiaomi's HyperOS software. It also carries TUV Rheinland certifications for low blue light and flicker-free display, which matters for people who spend hours looking at their screens.For less than KES 10,000, you are getting a 120Hz screen, which makes scrolling feel noticeably smoother than the standard 60Hz displays on locally assembled devices. You are getting a 32MP camera capable of decent photography in multiple lighting conditions. You are getting a battery that realistically lasts more than a day. You are getting a brand with global service infrastructure, an established software update track record, and the kind of marketing that has made Xiaomi one of the top-selling smartphone brands in Africa.Now compare that to the Neon Smarta at KES 7,499. A 5-inch display. 2GB of RAM. 32GB of storage. A 5MP camera. A 3,000mAh battery. Android 10 Go. The specifications are not just weaker. They belong to a completely different category of device.The honest reality is that a Kenyan consumer with KES 7,500 to spend on a phone has better options from imported brands than from the locally assembled lineup. And consumers know it. A GSMA report, titled "<a target="_blank" rel="noopener noreferrer nofollow" href="https://www.gsmaintelligence.com/research/accelerating-smartphone-adoption-in-africa">Accelerating Smartphone Adoption in Africa</a>," found that Kenyan consumers perceive locally assembled phones to be of lower quality. Brand familiarity, performance expectations, and the aggressive marketing of imported brands all contribute to a continued preference for imported devices, even when locally assembled options are available at lower prices.<h2>The Bigger Question: Who Is Buying These Phones?</h2>It is important to be fair here. The locally assembled phones are not targeting the same consumer who walks into a shop looking for a Redmi or a Tecno Camon. They are primarily targeting first-time smartphone users, people transitioning from feature phones, and low-income Kenyans who need connectivity and are not necessarily evaluating camera megapixels or display refresh rates.M-KOPA's business model is specifically built around this. Their financing structure, which allows customers to pay a small deposit and then daily installments of around KES 60 to KES 120, opens smartphone ownership to people who could not afford to pay KES 7,000 or KES 10,000 upfront. The Ministry of Health purchasing 100,000 EADAK devices for Community Health Workers is another example of an institutional use case where affordability and local production incentives matter more than raw performance.These are legitimate markets. And meeting them is a worthwhile goal. Kenya's smartphone adoption stands at 62 percent, which means there are millions of Kenyans who still do not own a smartphone. Getting those people connected, even on entry-level devices, is a meaningful step forward.But the danger of framing this entire initiative as "Kenya is manufacturing smartphones" is that it conflates two very different goals. Broadening first-time access is one goal. Building a sustainable, competitive technology manufacturing industry is another. The assembly lines can serve the first goal quite adequately. The second goal, however, requires something fundamentally different: local chipset design capability, local screen manufacturing, local battery production, investment in semiconductor research, and the kind of supply chain depth that takes decades and billions of dollars to develop. Kenya is nowhere near that point, and the current assembly operations do not move the country meaningfully closer to it.<h2>Will These Assembly Lines Survive the Shift?</h2>This is the question that matters most for the long-term health of the initiative.As Kenyans become more connected, their expectations from a smartphone will inevitably rise. A first-time smartphone user who buys a Neon Smarta today will, in two or three years, want something with a better camera, faster performance, and more storage. When that upgrade moment comes, the competition is not just within the locally assembled lineup. It is against Tecno, Infinix, Itel, Xiaomi, Realme, and Samsung, all of which are aggressively pricing competitive devices in the KES 8,000 to 20,000 range with significantly better hardware and established brand trust.M-KOPA's X20 is a step in the right direction. At 6GB RAM, 256GB storage, and a 50MP camera, it can hold its own against mid-range competition on paper. But to remain relevant beyond the financing-dependent customer base, locally assembled phones need to consistently match or exceed what is available from imported brands at the same price point. That requires continuous investment in better components, closer relationships with chipset suppliers, and genuine technical innovation rather than just label assembly.<ad-spot></ad-spot>The government can help by addressing the VAT friction on imported components, investing in technical education pipelines that feed skilled labor into these facilities, and being honest about the distinction between assembly and manufacturing so that policy goals are accurately calibrated. Calling assembly manufacturing is not just semantically wrong. It sets the wrong targets, creates misplaced pride, and can delay the harder work of building real industrial capacity.Kenya's assembly story is a beginning, not a destination. And beginnings are good. M-KOPA creating 325 jobs primarily for young people, targeting 500 by end of 2025, and aiming for 10 million assembled phones by 2027 is a trajectory worth watching. EADAK's capacity to produce up to 3 million units annually, if fully utilized and supported by policy, could meaningfully serve the lower end of the market.But it would be a mistake to confuse a milestone for a finish line. The smartphone market waits for no one, and Kenyan consumers, increasingly connected and informed, will vote with their wallets. The assembly lines that survive will be the ones that evolve to give people phones that actually respect what they need from their devices: powerful enough to run the apps that run their lives, cameras good enough to capture moments worth sharing, storage fast enough to keep up with modern software, and a price point that does not ask them to compromise on all of the above.The real test of Kenya's smartphone ambition is not how many units come out of the factory. It is whether the people those phones are meant to serve actually choose them.

Kenya Says It Makes 1 Million Smartphones a Year. But Are They Actually Made for Kenyans?

The next time you take a hot shower, there's a small but growing chance the water was heated not by your gas boiler, but by a cluster of servers crunching AI inference tasks in a box attached to your wall. That is not a thought experiment. That is 2026.The artificial intelligence industry is in the middle of what can only be described as a full-blown energy emergency. The demand for compute has grown so aggressively that the traditional playbook of "find land, build a data center, plug into the grid" is no longer working fast enough. Grids can't keep up. Planning approvals take years. Transformers are on backorder. Power utilities are overwhelmed. And so the tech industry, with all its characteristic boldness (and occasional absurdity), has started looking at the rest of the physical world and asking: can we put a GPU in that?The answers it is coming up with are genuinely fascinating.<h2>The Scale of the Problem</h2>Before we get to hot water and backyard server boxes, it helps to understand just how hungry AI has become.According to the International Energy Agency (IEA), electricity demand from data centres soared by 17% in 2025, well outpacing growth in global electricity demand of just 3%. Electricity consumption from AI-focused data centres grew even faster, surging 50% in that same year. And this is not expected to slow down. The IEA's updated projections see electricity consumption from data centres roughly doubling from 485 TWh in 2025 to 950 TWh in 2030, accounting for around 3% of global electricity demand by that date.<ad-spot></ad-spot>To put that in perspective, by one estimate, the energy consumption of data centers could approach 1,050 TWh by 2026, which, if data centers were a country, would make them the fifth largest energy consumer in the world, between Japan and Russia.At the company level, the investment numbers are staggering. The capital expenditure of just five large technology companies surged to more than $400 billion in 2025 and is set to increase by a further 75% in 2026. That level of spending is now larger than global investment in oil and natural gas production.The hardware inside data centers is also becoming exponentially more power-hungry. An individual server rack within an advanced data centre is only the size of a large refrigerator, but by 2027 it could have peak power demand equivalent to that of 65 households. Between 2020 and 2025, the power density of AI servers increased by 11 times, and by 2027 it is set to see a further fourfold increase.This is the context in which companies are now showing up at people's front doors with unusual propositions.<h2>The Hot Water Play: Heata and British Gas</h2>Let us start in the United Kingdom, where a startup called Heata has come up with an idea so elegantly clever it almost sounds like a joke until you realize it actually works.Heata has created a "virtual data centre," a network of servers distributed in people's homes. Each server is attached to the home's hot water cylinder, and as they process data, the heat they generate is transferred into the water. This reduces the energy needed to heat water in the home, cutting household bills and reducing carbon emissions in the process.The concept flips the traditional data center problem on its head. In a normal data center, heat is the enemy. Enormous energy is spent on cooling systems just to keep the servers from melting. Heata essentially relocates that waste heat into a place where heat is actually valuable: your hot water tank.Each Heata unit can provide up to 4kWh of hot water per day, and Heata pays for the electricity the unit uses, meaning the household pays less to heat their water. According to the company, this technology is expected to save households up to £340 per year when offsetting electrically heated hot water, and up to £120 per year when offsetting gas heating.British Gas came on board for a trial in early 2025. As part of a three-month trial, 10 Heata units were installed in the homes of British Gas employees, with the energy provider's computing workloads processed on those units. British Gas provided free hot water for its own employees as a byproduct of its own cloud compute.Heata claims its approach reduces energy use by 80% compared to traditional data centers, and can simultaneously help people in fuel poverty. That is a remarkable dual benefit: cheaper AI compute and cheaper home energy, both from the same device sitting next to your boiler.Heata co-founder Chris Jordan described the logic plainly: "Waste heat is a big problem for data centers, leading to significant energy costs for cooling. Yet heat is valuable. On the other side of the coin, you have an energy crisis and people struggling to heat their homes. Our unique technology brings those two things together."<ad-spot></ad-spot>It is worth noting that the system does have one practical constraint. It requires a hot water cylinder, so it cannot be used in homes with combi boilers. That rules out a significant portion of UK homes. But for those with the right setup, it is a genuine win-win.<h2>The GPU Box on Your Wall: Span's XFRA</h2>If Heata is clever and subtle, what California-based Span is proposing is considerably more dramatic.Span, a startup that originally launched with smart electrical panels designed to help homeowners save money on their electricity bills, has partnered with Nvidia to develop small, fractional data centers called XFRA nodes that can be installed on the exterior walls of residential homes and small businesses.These are not small gadgets. Each XFRA node contains Dell PowerEdge servers with 16 Nvidia RTX Pro 6000 Blackwell GPUs, 4 AMD EPYC CPUs, and 3 TB of RAM, connected to a 24-port gigabit switch. This is serious professional hardware, the kind you would normally find in a hyperscale cloud facility, being mounted on the side of a house in a suburban neighborhood.Span collaborated with Nvidia to use liquid-cooled Nvidia RTX PRO 6000 Blackwell Server Edition GPUs. These require no fans, so there is no noise. That is a critical detail. One of the biggest community complaints about data centers is the constant industrial humming. The fanless, liquid-cooled design means a neighbor would likely have no idea the unit was there.The financial deal being offered to homeowners is genuinely attractive. The company charges a flat monthly fee of about $150. In return, it essentially pays the host's electricity and internet bills. The computing power generated from the nodes is distributed to customers like hyperscalers and AI companies. Span's Chief Revenue Officer went further, suggesting that homeowners in places where XFRA nodes provide the most value could receive discounted electricity up to and including free electricity and free internet access.The reason Span can offer this deal comes down to a simple observation about how homes use electricity. Residential homes operate at an average of just 40% of their peak power capacity, leaving significant untapped headroom that can be utilized for compute. The XFRA node taps that headroom without disrupting normal household activity.Span plans a 100-home pilot in 2026 and aims to scale to 80,000 nodes nationwide starting in 2027, delivering more than 1 gigawatt of distributed compute capacity. For context, that is equivalent to a large traditional data center, but distributed across tens of thousands of homes in ordinary neighborhoods.Span says it can install 8,000 units six times faster and for roughly a fifth of the cost of building a large, centralized 100-megawatt data center. That speed-to-power advantage is the whole point. The bottleneck for AI expansion is not money or hardware; it is grid access and permitting timelines. As of early 2025, the Center for Strategic and International Studies noted delays of up to 7 years in the nation's largest data center markets. Span's approach sidesteps all of that.Major homebuilder PulteGroup is already testing the concept. "There is certainly opportunity, as SPAN can provide homeowners with access to innovative technology and potential income generation that can help offset monthly energy costs," the company said. "On a larger scale, if the technology proves out, it might also keep local infrastructure from being overburdened, which could keep land open for other uses, such as building homes."<ad-spot></ad-spot><h2>The Experts Are Not Entirely Sold</h2>Not everyone is enthusiastic about the idea of suburban compute nodes.Ari Peskoe, director of the Electricity Law Initiative at Harvard Law School, called the homeowner subsidy "fascinating," but raised a practical concern about clustering. "If there's a block that has several homes with these devices, maxing out compute and energy would force a lot of power to that local area," he said. Local distribution networks were not designed for sustained heavy industrial loads dropped into residential streets.There are also broader systemic concerns. A Utah State University physics professor warned that efforts to modestly reduce the ecological harms of data center power usage could actually exacerbate the overall problem. In a preliminary analysis, he calculated that only 30 to 40 percent of homes may be suitable for mini data centers due to integration constraints, the need for stable internet, and participants being willing to have the technology installed.Security researchers have flagged physical theft and side-channel attacks as potential risks when high-value GPU hardware sits in unguarded residential locations. These are real concerns that Span will need to address before scaling to tens of thousands of homes.<h2>The Bigger Picture: Compute by Any Means Necessary</h2>Heata and Span are not isolated experiments. They are part of a broader, industry-wide scramble for power that is pushing the AI sector into increasingly unconventional territory.The most dramatic example involves a nuclear plant that is synonymous with disaster. The owner of the shuttered Three Mile Island nuclear power plant announced plans to restart the reactor under a 20-year agreement that calls for Microsoft to purchase the plant's power to supply its data centers with carbon-free energy. Constellation Energy described the agreement with Microsoft as the largest power purchase agreement the nuclear plant operator has ever signed, with the plant expected to come back online in 2028. Constellation will invest $1.6 billion to restart the plant, covering nuclear fuel, the turbine, the generator, the main power transformer, and cooling and control systems.Let that sink in. A reactor that shut down because it was economically unviable is being brought back to life, at a cost of over a billion dollars, because AI companies need more electricity.Beyond nuclear, tech companies are hunting for power wherever they can find it: buying up decommissioned factories and old cryptocurrency mining facilities because they already have high-voltage grid connections, mapping deep geothermal zones near volcanic activity, and building solar arrays in the middle of deserts. The through-line is desperation for reliable, large-scale power, accessed faster than the normal permitting process allows.The grid itself is beginning to feel the strain everywhere, not just near the data centers. In the PJM electricity market stretching from Illinois to North Carolina, data centers accounted for an estimated $9.3 billion price increase in the 2025-26 capacity market. As a result, the average residential bill is expected to rise by $18 a month in western Maryland and $16 a month in Ohio. Ordinary people who have never used an AI tool in their lives are already subsidizing the compute boom through their electricity bills.This is precisely why the distributed, residential-compute model is so conceptually interesting. If AI companies are going to consume neighborhood electricity regardless, at least Span's approach gives the homeowner a direct financial benefit from hosting the load, rather than simply absorbing a higher bill.<h2>What This Means Going Forward</h2>We are watching the physical infrastructure of AI expand beyond the boundaries of the traditional data center in real time. The home, historically a private space insulated from industrial activity, is being quietly recruited into the global compute grid.For homeowners, the deals on offer right now look generous. Free hot water, paid electricity bills, free internet. The question is what the long-term implications look like as the technology scales. Who is liable if a node malfunctions and causes a problem? How does resale value of a home change when there is a GPU cluster mounted to its exterior? What happens to the agreement if the AI company encounters financial difficulty?<ad-spot></ad-spot>These questions are not reasons to dismiss the technology. They are the normal growing pains of any infrastructure shift. The residential electricity grid, the mobile phone tower, the rooftop solar panel: all of these faced similar skepticism before becoming normalized features of everyday life.The interesting thing about this particular moment is that the financial pressure to make distributed compute work is enormous and growing. The IEA projects that electricity consumption from AI-focused data centres will triple by 2030. That trajectory guarantees more innovation, more unusual partnerships, and more proposals that will sound absurd until they are suddenly everywhere.Kenya and the broader African continent are not yet at the center of this story, but the dynamics playing out in the UK and the US are instructive. As local AI infrastructure grows and electricity reliability remains a challenge in many regions, the principle of distributing compute closer to where power exists, and compensating hosts for that privilege, could prove even more relevant here than in the markets where it is currently being tested.The era of "compute by any means necessary" is producing some genuinely creative engineering. Some of it will fail. Some of it will reshape how we think about homes, energy, and infrastructure. Either way, it is worth paying close attention.

Your Home Is Now a Data Center: How the AI Energy Crisis Is Turning Living Rooms Into Server Farms

For nearly two decades, the Kenyan telecommunications landscape has been defined by a single entity: Safaricom. Since its early dominance, the Green Giant has not just led the market, it has defined it. From the pioneering days of M-Pesa to the 5G rollout, Safaricom has been the benchmark against which all others were measured.But the Communications Authority of Kenya's Q2 2025/26 Sector Statistics Report, covering October to December 2025 and released earlier this month, tells a story that Safaricom's shareholders should be reading carefully. Not because Airtel is winning on subscriber numbers, it is not, at least not dramatically. But because Airtel is winning in the one place that actually funds Safaricom's entire operation.<h2>The Number That Actually Matters</h2>Mobile money contributes 41% of Safaricom's total revenue, Ksh 161 billion out of Ksh 388 billion in the last full financial year. M-Pesa is not just a product. It is the engine.Two years ago, Airtel Money held 5.5% of Kenya's mobile money market. By Q1 2025/26 that had risen to 10.3%. The latest Q2 report confirms it has now crossed 11%. M-Pesa's share, which once sat at 98%, now stands at 89%.<ad-spot></ad-spot>That is a 9 percentage point erosion in two years in the segment that funds almost half of Safaricom's revenue. The pace is not explosive, but it is consistent, and consistency in market share erosion is more dangerous to a dominant player than a single dramatic shock.<h2>The Subscription Picture Is More Complicated</h2>Your WhatsApp groups have probably been circulating claims about Airtel's dramatic subscriber growth. The actual CA data is more nuanced.As of December 2025, total mobile subscriptions across all operators stood at 78.4 million, growing by just 0.1% in the quarter, a sign of a maturing market that is approaching saturation at 149.5% penetration. Safaricom holds 52.4 million of those subscriptions, roughly 66.8% of the market. Airtel holds 22.9 million, or approximately 29.2%.Airtel's subscription share is not surging, it actually edged slightly lower between Q1 and Q2 as the overall market growth slowed. The subscriber story is not where the competition is being decided. The mobile money story is.What the subscription data does reveal, however, is something more behavioural and arguably more interesting than raw market share.<h2>The Active SIM Insight: Airtel Users Actually Use Their Phones</h2>Buried in the CA report is a data point that deserves more attention than it has received. Airtel users average 2.7 minutes per call. Safaricom users average 1.6 minutes.Think about what that means. Most Kenyans carry two SIMs, a Safaricom SIM for M-Pesa access and network coverage, and an Airtel SIM for cheaper calls and data. The Safaricom SIM is the wallet. The Airtel SIM is the phone.When someone calls from Airtel, they stay on the line longer because the call is affordable enough to have a full conversation. Safaricom calls are shorter because users are watching the bill tick. This call duration gap is one of the clearest signals of where pricing pressure actually lands on ordinary Kenyans.<ad-spot></ad-spot>Airtel Kenya also recorded 11.83 billion minutes of voice traffic in Q2, up 2.4% from Q1. More tellingly, off-net traffic( calls crossing from Airtel to Safaricom and other networks) grew 8.4% in the same period. People are calling across networks more than before, which means interoperability is working, and Airtel's affordability advantage is compelling users to use it as their primary communication tool even when calling someone on a different network.<h2>The Pricing War: Where Two Shillings Becomes Ksh 3,000</h2>At TechInKenya we often focus on technical specifications, but for the average Kenyan consumer the most important feature of any network is the cost of a minute or a megabyte. Airtel has played the long game on affordability and the mobile money data confirms it is paying off.Sending Ksh 1,000 to another network on Airtel Money costs Ksh 11. On M-Pesa the same transaction costs Ksh 13. A two-shilling difference sounds negligible. For a micro-entrepreneur in Kajiado making 50 transactions a day, that is Ksh 100 daily, Ksh 3,000 every month in savings, or Ksh 36,000 per year. At that scale, the "cheaper" network stops being a secondary SIM and starts becoming a business decision.This aggregation of marginal gains is precisely how Airtel is converting M-Pesa's casual users into Airtel Money's committed ones.<h2>The Interoperability Trojan Horse</h2>For years M-Pesa's dominance rested on a network effect that felt unbreakable. Everyone had M-Pesa, so you needed M-Pesa to send money to anyone. Switching to Airtel Money meant your recipients might not be able to receive it. The ecosystem locked you in more effectively than any contract.The CA's push for full interoperability dismantled that lock-in quietly but completely. Now a user can keep their balance in an Airtel Money wallet, pay lower transaction fees, and still pay at a Safaricom Buy Goods till or any Paybill number. The friction of being on a different network has effectively been removed.<ad-spot></ad-spot>This is why Airtel Money's growth from 5.5% to 11% in two years tracks almost perfectly with the interoperability timeline. The regulatory change was the door. Lower fees were the reason to walk through it.<h2>What Safaricom Is Doing About It</h2>Safaricom is not standing still. The launch of My OneApp this month ( merging the M-Pesa and MySafaricom apps into a single AI-powered platform ) is partly a response to the UX pressure from competitors. The new app's infrastructure reportedly handles up to 6,000 transactions per second, up from roughly 100 previously. The AI personalisation layer, which learns user behaviour and surfaces relevant services proactively, is a genuine attempt to deepen the ecosystem moat rather than just defend the pricing one.But My OneApp has generated its own friction. Users abroad have reported significant issues with the app's device-binding and OTP requirements that do not work well with international SIM swaps or roaming. For diaspora Kenyans ( a segment that sends significant remittance volumes through M-Pesa )being locked out of your account when you need it most is not a loyalty-building experience. Airtel's lighter, more utility-focused app has quietly absorbed some of that frustration.Safaricom also holds structural advantages that Airtel cannot easily replicate. Its IoT segment leads the market with 1.8 million active connected devices ( smart meters, fleet management, industrial monitoring ) a B2B revenue stream Airtel has not meaningfully entered. Safaricom Home Fibre remains dominant in suburban fixed broadband, where the physical fibre infrastructure provides stability that 5G routers cannot yet match for high-bandwidth households.The Ethiopia expansion is the strategic wildcard. The Safaricom Ethiopia rollout has consumed billions of shillings in capital expenditure over the past three years, which some analysts argue has constrained Safaricom Kenya's ability to respond more aggressively to domestic pricing pressure. If Ethiopia reaches profitability on its current trajectory, that capital will have been well spent and Safaricom will have built a second large market. If it does not, the domestic opportunity cost becomes harder to justify.<h2>The Broadband Front: A Third Competitor</h2>One finding in the CA data that neither Safaricom nor Airtel should ignore: Starlink has reached 0.9% of Kenya's fixed broadband market. That sounds small, but Ahadi Wireless is at 9% and Vilcom Network at 5.4% and both grew this quarter. Safaricom's fixed broadband share declined to 34.9% from 35.6%.<ad-spot></ad-spot>The fixed internet market is fragmenting. For the urban professional household that uses Home Fibre primarily for streaming, gaming, and working from home, the question of whether Safaricom's fibre or a Starlink dish is the better option is a real one in 2026. This is a slow-moving threat but it is worth watching.<h2>What Happens Next</h2>Airtel's path to further mobile money growth runs through the same mechanism that got it to 11% that is pricing, interoperability, and consistent execution. Hitting 15% mobile money share within two years is plausible if the current trajectory holds. Whether it can get to 20% or beyond depends on whether Safaricom makes a structural pricing response rather than an ecosystem one.Safaricom's most likely responses:Loyalty depth over loyalty breadth. Bonga Points have always been a thin loyalty mechanism. Deeper integrations (cashback within M-Pesa transactions, bundled data with financial products, preferential rates for high-volume users ) would raise the switching cost without requiring across-the-board fee cuts.Enterprise and IoT as the moat. The B2B segment ( IoT connections, enterprise data, government contracts, API access through Daraja) is where Safaricom's margins are most defensible and where Airtel has the least competitive presence. Expect continued investment here.Ethiopia as the long-term bet. If the Ethiopia venture starts generating positive returns in the next 18 to 24 months, Safaricom's growth story changes significantly. It stops being a single-market company competing on price and becomes a regional technology platform with a very different investment case.<h2>The TechInKenya Take</h2>The current competitive dynamics are healthy for Kenyan consumers. Safaricom at 98% mobile money market share two years ago was not a situation that produced competitive pricing or incentivised product innovation. Airtel at 11% and rising is not yet a genuine threat to Safaricom's position, but it is enough of a threat to change behaviour.The two-shilling fee difference is already saving the micro-entrepreneur in Kajiado Ksh 36,000 a year. That is real money doing real work. That is what competition is supposed to deliver.The Green Giant is not going anywhere. But for the first time in the history of Kenya's mobile money market, M-Pesa's dominance is something that is being competed away — slowly, consistently, and in the one segment that matters most to Safaricom's bottom line.All figures sourced from the Communications Authority of Kenya Q2 FY2025/26 Sector Statistics Report covering October to December 2025. Full report available at ca.go.ke/statistics.

The Airtel Rebellion: How Kenya's Underdog Is Attacking Safaricom Where It Hurts Most

Local Assembly or Global Surplus Offloading? The Uncomfortable Truth Behind Kenya's KSh 320M EV Plant

<h3>The "Involution" Crisis: Why China is Overflowing</h3>The term "Involution" (neijuan) has emerged as the defining economic struggle of 2025. It describes a state of "meaningless competition"—a cycle where an oversupply of goods meets stagnant domestic demand, forcing Chinese companies into brutal price wars that slash profit margins to the bone.<ul><li>The Surplus Problem: China’s massive manufacturing engine continues to accelerate, yet domestic consumer confidence remains at historic lows.</li><li>The Price War: In sectors ranging from heavy industry to consumer electronics, companies are selling products at or even below cost simply to defend their market share.</li><li>The Export Pressure: Facing a record $1 trillion-plus trade surplus, China is desperate to offload this inventory into global markets to prevent a total domestic industrial collapse.</li></ul><ad-spot></ad-spot><h2>The Global Shield vs. The Kenyan Open Door</h2>As China attempts to export its way out of the Involution trap, the world’s major economies are raising their shields. The US and EU have implemented aggressive "de-risking" strategies and punitive tariffs to prevent their own industries from being crushed by subsidized, ultra-cheap imports.Kenya, however, tells a different story. While the West builds barriers, our "open-door" policies have made us a primary destination for this surplus.<ul><li>The Trade Gap: The numbers are staggering. In October 2025 alone, Kenya imported $622 million worth of goods from China while exporting a mere $13.2 million in return. This creates a monthly trade deficit of $609 million, a gap that drains Kenya's foreign exchange reserves.</li><li>Automotive Dumping: As China pivots rapidly toward Electric Vehicles (EVs), it is offloading massive quantities of gasoline-powered vehicles and spare parts into African markets. In 2025, automotive spare parts dominated the illicit trade sector in Kenya, with the Anti-Counterfeit Authority reporting that four-wheeler accessories account for 81.8% of seized counterfeit goods.</li></ul><h2>The Long-Term Cost: "Premature De-industrialization"</h2>The danger isn't just the influx of cheap goods; it’s the systemic erosion of the Kenyan manufacturing dream.<ul><li>Job Losses: The Kenya Association of Manufacturers (KAM) has warned that the flood of cheap imports, combined with rising operational costs, has placed over 60,000 direct manufacturing jobs at risk as of late 2025.</li><li>Unfair Competition: Local firms are being forced to shutter or relocate because they cannot compete with "dumped" goods that are subsidized by the Chinese government.</li><li>The Stifled Dream: Kenya’s aspiration to be a regional tech hub is in a stranglehold. Local assembly for smartphones and PCs is caught between high taxes on imported components and an overwhelming surge of finished Chinese imports.</li></ul>The Shipment Surge: In 2024, Kenya imported over 2.3 million smartphones from China—a 71.5% increase from the previous year. While this benefits consumer connectivity, it prevents local assembly plants from reaching the scale they need to survive.<ad-spot></ad-spot><h2>Trade Policy: Kenya’s "Open Door" vs. The EU’s "Shield"</h2>The divergence in trade strategy between Kenya and the EU highlights a fundamental choice between immediate consumer convenience and long-term economic survival.<table style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1">Features</th><th colspan="1" rowspan="1">Kenya's Approach (2025)</th><th colspan="1" rowspan="1">EU "Special Safeguards" (2025)</th></tr><tr><td colspan="1" rowspan="1">Primary Goal</td><td colspan="1" rowspan="1">Maintaining "Free Trade" and safeguarding bilateral ties.</td><td colspan="1" rowspan="1">Protecting domestic industry and reducing "strategic dependence."</td></tr><tr><td colspan="1" rowspan="1">Mechanisms </td><td colspan="1" rowspan="1">WTO valuation and cargo verification to catch tax evasion. </td><td colspan="1" rowspan="1">"Import Surveillance Task Force" to trigger punitive duties on surges.</td></tr><tr><td colspan="1" rowspan="1">Action Taken</td><td colspan="1" rowspan="1">Zero-tariff treatment for 100% of certain tariff lines from China.</td><td colspan="1" rowspan="1">Implementing "contingency plans" including steel safeguards and EV tariffs.</td></tr><tr><td colspan="1" rowspan="1">Local Impact</td><td colspan="1" rowspan="1">Record trade deficits; Chinese exports outpace Kenyan exports by ~40:1.</td><td colspan="1" rowspan="1">Active "ReSourceEU" program to diversify supply chains.</td></tr></tbody></table>While the EU uses Article 5 of the WTO Agreement to reintroduce duties the moment an import surge threatens their local market, Kenya has leaned into a "Comprehensive Strategic Partnership." Critics argue this leaves the country without the necessary protectionist shields to nurture its own industry.<h2>Deep Dive: "Blurred Corruption" in Infrastructure</h2>"Blurred corruption" is not always about direct bribes; it is about opaque contractual structures that create a closed economic loop designed to benefit the lending country over the host nation.<h3>1. The Standard Gauge Railway (SGR)</h3>The SGR is a textbook example of an "imagined future" that has led to debt distress.<ul><li>The Loop: With 90% of the project funded by China's Export-Import Bank, contracts stipulated that the China Road and Bridges Corporation (CRBC) would handle construction and maintenance. This ensured the loan money essentially flowed from one Chinese entity to another, bypassing the Kenyan economy.</li><li>Labor Disparity: Despite the massive scale, the project primarily utilized Chinese materials and reserved high-skill engineering roles for Chinese nationals, leaving only low-paid manual labor for the local workforce.</li></ul><ad-spot></ad-spot><h3>2. The Nairobi Expressway</h3>Though framed as a Public-Private Partnership (PPP), the Expressway follows a similar logic of "access for infrastructure."<ul><li>Exchange Risk Protection: Toll fees were significantly increased in 2025 to "cushion" the Chinese investors against the depreciation of the Shilling. This shifts the entire economic burden of currency fluctuation onto the Kenyan motorist.</li><li>Financial Control: Operated by Moja Expressway (a Chinese contractor), the revenue stream from this critical Kenyan infrastructure stays within a Chinese-managed ecosystem for decades to come.</li></ul><h2>Conclusion: The Path Forward</h2>If Kenya is to avoid becoming a permanent safety valve for China’s economic "involution," it must move beyond being a passive consumer. Strengthening local manufacturing, enforcing stricter anti-dumping laws, and demanding more transparent infrastructure contracts are not just policy options—they are requirements for national economic sovereignty.

Industry

Kenya Says It Makes 1 Million Smartphones a Year. But Are They Actually Made for Kenyans?

Your Home Is Now a Data Center: How the AI Energy Crisis Is Turning Living Rooms Into Server Farms

The Airtel Rebellion: How Kenya's Underdog Is Attacking Safaricom Where It Hurts Most

Local Assembly or Global Surplus Offloading? The Uncomfortable Truth Behind Kenya's KSh 320M EV Plant

The Involution Trap: Is Kenya Becoming China’s Economic Safety Valve?