By 
By Ross Moyo
Econet is no longer just a mobile network. With 100+ AI engineers, a 100-model marketplace, and a chatbot that now speaks Shona better than most global models, it is trying to do for AI what EcoCash did for mobile money.
To understand Shona 2.0, you have to understand what Econet has become in 2026.
Econet Wireless Zimbabwe recorded revenue growth of 21 percent for the half-year ended August 31, 2026, supported by sustained demand for connectivity, insurance and financial services.
Performance across its four strategic pillars – Mobile Network Operations, Financial Technology, Econet Infrastructure and Econet AI – strengthened the group’s growth, resilience and integrated digital ecosystem.
Econet chairman Dr James Myers said artificial intelligence had become an increasingly important part of the group’s business strategy, moving from experimentation into core operations and product development.
Data traffic increased by 60 percent and voice traffic by 21 percent.
The group accelerated artificial intelligence adoption across customer service, network optimisation, operational efficiency and product development, while pursuing new revenue opportunities.
In other words, AI is no longer a pilot. It is a pillar, alongside the network itself.
The four pillars in practice
1. Mobile Network Operations: Econet ended Q4 2025 with 340 5G base stations and 7,143 total active base stations, exceeding NetOne’s 4,838 and Telecel’s 1,123 combined network footprint. Econet holds 73.75% of active mobile subscribers, 88.32% of mobile voice traffic, and about 81.2% of mobile internet and data traffic.
By Q1 2026, Econet’s mobile data usage rose by 20.14% to reach 156.4 Petabytes, bringing its mobile data market share to 87.21%. a11c
2. Financial Technology: EcoCash remains the backbone of Zimbabwe’s digital payments. Of 234.7 million electronic payments recorded in the first quarter of this year, around 208.8 million went through mobile wallets, most of them on EcoCash.
3. Econet Infrastructure: The company is building the towers, fibre and data centres that other operators and enterprises will rent.
4. Econet AI: This is the new pillar. And Yamurai is its most visible product.
The EcoCash playbook
Econet is deliberately replicating the playbook it used to turn EcoCash from a novel product in 2011 into Zimbabwe’s dominant mobile money platform now used by millions.
EcoCash worked because it solved a real problem – cash shortages, bank queues, sending money to rural areas – in a language people understood, on phones they already had, at a price they could afford.
Econet Ai is trying to do the same for AI.
“We want everyone who has a smartphone to start using AI, and that is nearly 70% of our base. Unfortunately, you can’t access AI without a proper smartphone,” said Econet Ai CEO Mr Navdeep Kapur.
Zimbabwe has over 15 million mobile subscriptions and smartphone penetration is estimated at between 65-70%, one of the highest in Southern Africa, providing a fertile ground for the campaign.
Econet says its ambition is to make Zimbabwe the most AI-savvy country in Africa, matching countries like China, Singapore and the US where up to 70% of citizens use AI on a daily basis.
Econet Group Chief Executive Officer Dr Douglas Mboweni said Zimbabwe already had what the company believes to be the highest level of daily AI interaction in Africa, supported by services such as Yamurai, EcoChat and EcoFarmer.
Zimbabweans are using AI every day through Econet products like Yamurai, EcoChat, and EcoFarmer.
Yamurai, EcoChat – Econet’s AI customer care assistant – and EcoFarmer, which provides agronomic advice to farmers, have already introduced millions of Zimbabweans to AI without them necessarily knowing it.
That is the EcoCash insight: people do not adopt “AI.” They adopt a solution that helps them check a bundle, send money, get farming advice, find a Bible verse, help a child with homework.
CAIMEX: 100 models, one platform
Central to the campaign is CAIMEX, Econet’s flagship AI marketplace platform described as the first of its kind in Africa. CAIMEX gives users access to over 100 curated global AI models – including large language models, image generators, coding assistants, translation and productivity tools – in one place.
The platform operates through a token-based payment system, potentially creating a new revenue stream for Econet while giving consumers access to multiple AI models without subscribing individually to each provider.
Users access the models by purchasing AI tokens which power their interactions with the tools.
CAIMEX also includes EcoAI Tokens, providing access to ChatGPT, Gemini, Claude and Mistral through one bundled subscription. Google Gemini was introduced free of charge to eligible customers, expanding access to productivity tools.
The logic is simple: a Zimbabwean student should not need four subscriptions – ChatGPT + Gemini + Claude + Mistral – to do homework. A small business should not need to pay in dollars for each model. One token bundle, one platform, in Shona, in Ndebele, in English.
That is also why Yamurai matters for CAIMEX. Yamurai is the on-ramp. A customer who learns to talk to Yamurai in Shona on WhatsApp is a customer who is ready to try EcoChat, EcoFarmer, CAIMEX.
The Cassava connection
Econet Ai works closely with Cassava AI, the artificial intelligence business of Cassava Technologies.
Econet AI’s partnership with Cassava AI provides access to NVIDIA accelerated computing infrastructure.
Cassava had invested in Africa’s first AI Compute Centre to serve the region. Econet AI will roll out Cassava AiCloud, a platform that allows Zimbabwean enterprises to run data-intensive workloads using high-performance computing powered by NVIDIA GPUs.
Navdeep Kapur, Econet AI CEO, said the unit will focus on scalable AI systems tailored to African markets. “Our vision is to build AI that is not only powerful, but relevant to African contexts,” Kapur said.
Kapur said their focus is not to following global hype around AI but to deploy practical solutions that reduce costs, improve efficiency and grow businesses while delivering tangible benefits to ordinary citizens.
Our promise is that we do not want Zimbabwe to be a consumer of AI. We want Zimbabwe to be a creator of AI.
That promise is being tested in the most practical way: can a Zimbabwean team, using African compute, train a model in an African language, to a standard that matches global models?
Shona 2.0 is Econet’s answer: yes, for customer service, for 80% of queries, in Shona.
What “best service in Africa” would actually mean
“Now there is real light at the end of the tunnel, as AI opens solutions at all levels of telco operations. It takes time to get them all deployed but we are confident that we are only months away from having the best service in Africa,” said Mr Kapur.
“Best service in Africa” is a bold claim. What would it look like in practice?
No waiting: If Yamurai handles 80% in Shona, and Ndebele 2.0 does the same in Ndebele, then the 600 human agents can focus on the 20% that are complex, emotional, or high-value – a lost SIM for an elderly grandmother, a network fault affecting a whole village, a fraud case.
No language barrier: A farmer in Binga can ask in Tonga. A grandmother in Plumtree can ask in Kalanga. A student in Chipinge can ask in Ndau. No one has to translate their problem into English first.
Agentic action: Yamurai does not just tell you your PUK. It retrieves it, sends it, and confirms it. It does not just tell you there is a network fault. It logs it, escalates it, and tells you when it will be fixed.
Network that heals itself: Econet has partnered with Ericsson to modernise its mobile network using AI, 5G Advanced technologies, and next-generation IoT services. The company has been steadily integrating AI and machine learning into its operations as it moves toward the creation of more autonomous networks – systems that can configure, optimise, and repair themselves with minimal human intervention.
That last point is crucial. AI at Econet is not just Yamurai. It is network optimisation, fraud detection, predictive maintenance, energy management for base stations, and customer churn prediction.
If AI can predict that a base station in Masvingo will fail tomorrow, and fix it today, that is better service than any chatbot.
The risks – jobs, privacy, and trust
No honest newspaper can write about AI in a call centre without asking about jobs.
Earlier reporting noted that Yamurai works alongside about 600 human contact-centre staff.
If Yamurai goes from 70% to 80% to 90% resolution, what happens to those 600 jobs?
Econet has not said it will cut staff. Its argument is that Yamurai “ensures that all queries are addressed” – meaning humans were missing queries before, because they could not keep up with volume. AI handles the volume, humans handle the complexity.
That argument is plausible, but it needs to be tested. Zimbabwe’s banking sector has already seen at least 516 banking jobs lost in 2025 alone, according to the Zimbabwe Banks and Allied Workers’ Union. AI is part of that story globally – Barclays said it would cut about 5,000 jobs, roughly 5 per cent of its 93,000 staff, and that 1,000 had already gone.
Econet should publish: how many contact centre staff it employs now, how many it employed before Yamurai, and how their roles have changed – from answering repetitive queries to handling complex cases, with higher pay and training.
The second risk is privacy. Yamurai learns “from listening to our customers.” That means customer conversations are training data. Econet must be transparent: are conversations anonymised? Are they used with consent? Are they stored in Zimbabwe or in South Africa at the Cassava AI Factory? Who has access? How long are they kept?
The third risk is trust. If Yamurai gets a query wrong – if it tells a customer their bundle is active when it is not, or gives the wrong PUK, or misinterprets a dialect – who is accountable? Econet says Yamurai ensures all queries are addressed. It should also say: Yamurai will tell you when it is not sure, and hand you to a human, in your language, without you having to repeat yourself.
What to watch next
1. Ndebele 2.0 before end of November: Econet says work is at an advanced stage. If it launches on time, with the same 80% claim, that will prove Shona 2.0 was not a one-off.
2. Dialect list: Which six Shona dialects? Will Ndau and Korekore get equal accuracy to Zezuru?
3. Independent benchmark: Will Econet publish a test set, or allow UZ, NUST, or an independent lab to test Shona 2.0?
4. Tonga, Kalanga, Nambya: When will they launch? Those are the languages that will prove Econet’s commitment to all 16 official languages, not just the two largest.
5. Agentic Yamurai: When will Yamurai move from answering to acting – completing SIM swaps, refunds, network fault reports end-to-end, without a human?
For now, the headline stands: a Zimbabwean company, using Zimbabwean engineers, Zimbabwean data, and African compute, has built a model that speaks Shona like a person, not like a robot.
That is not just a telco upgrade. That is a statement about whose language counts in the age of AI.
Comments