Econet InfraCo, the tower infrastructure subsidiary of Econet Wireless Zimbabwe, says it has reduced diesel costs by 30% through an artificial intelligence (AI) platform designed to optimise generator use, highlighting how digital technologies are being applied to manage energy expenditure across Africa’s telecommunications infrastructure sector.
The platform, known as AI Fuel Mode, was developed in collaboration with sister company Cassava Compute. It analyses patterns in diesel generator deployment to help the company improve fuel management, reduce unnecessary generator use and control one of its largest operating expenses. Econet InfraCo Chairman Godfrey Gomwe disclosed the savings in a statement accompanying the company’s half-year financial results for the six months ended 31 August 2026.
“By using AI to analyse how and when each diesel generator is deployed, we have developed a fuel management system that has reduced our diesel costs by 30%. This demonstrates the practical value of AI in improving operational efficiency and managing one of our largest expenses,” Gomwe said.
The reported reduction comes as energy costs remain a major operational challenge for telecommunications infrastructure companies. Mobile network towers require a continuous power supply to maintain connectivity, including during periods when electricity from the national grid is unavailable or unreliable. In these circumstances, diesel generators provide backup power, but their fuel and maintenance requirements can increase operating costs and expose companies to fluctuations in fuel prices.
Econet InfraCo’s latest interim financial statements identify higher diesel costs as a pressure on its operations during the reporting period, partly reflecting volatility in global oil prices associated with geopolitical tensions in the Middle East. The company said it was responding by accelerating site solarisation and deploying AI-enabled fuel management to improve energy efficiency and reduce dependence on diesel-generated electricity. Its power business reported a 30% reduction in fuel consumption during the period, alongside measures including solarisation, battery cycling technologies and periods of improved grid availability. The distinction matters because the company’s broader fuel-consumption results reflect a combination of interventions, rather than AI alone.
AI Fuel Mode uses generator deployment patterns to develop a fuel management matrix intended to improve how generators are operated across the company’s infrastructure. By analysing when generators are used, the system is designed to identify opportunities to reduce unnecessary runtime and coordinate power use more efficiently.
The approach could be relevant to telecommunications operators across Africa, where unreliable grid electricity and high energy costs have made backup generation an important part of network operations. Industry reporting cited in coverage of the initiative estimates that more than 500,000 generators support mobile network operations across the continent. The scale of this installed equipment points to the potential significance of improvements in generator efficiency, although savings will vary according to local grid reliability, site configuration, equipment condition and the availability of alternative power sources.
The company has not disclosed the number of tower sites covered by the AI Fuel Mode deployment, the platform’s technical architecture or the detailed methodology used to calculate the reported 30% reduction in diesel costs. It is therefore not possible to determine from the available information how much of the savings resulted directly from AI-driven optimisation, as opposed to other operational changes or movements in fuel prices.
These details will be important in assessing whether the results can be replicated across other networks. A reduction in diesel expenditure does not necessarily translate into an equivalent reduction in fuel consumption, because costs can also change with prices. Similarly, lower fuel use does not automatically establish a specific reduction in greenhouse gas emissions without information about baseline consumption, operating conditions and the period measured.
The platform was developed using Cassava Compute’s AI computing infrastructure powered by NVIDIA technology in Cape Town, South Africa. Econet InfraCo and Cassava Compute have described AI Fuel Mode as a first-of-its-kind deployment in the tower infrastructure industry. Such claims have not been independently established by the information available for this article.
Cassava AI plans to make the platform available to other operators, potentially extending its application beyond telecommunications to sectors such as mining and manufacturing. These industries can face similar energy-management challenges where equipment must remain operational despite interruptions in grid electricity or where fuel represents a substantial operating cost.
Commercial expansion will depend on whether the platform can deliver consistent savings across different operating environments. Mining operations, manufacturing plants and telecommunications networks have different power requirements, operating schedules and equipment configurations. Deploying an AI fuel management system in new sectors would therefore require adaptation to the data available, the energy systems involved and the operational constraints of each business.
The initiative also forms part of a broader effort by Econet and its technology businesses to apply AI to infrastructure management. Cassava AI, working with NVIDIA, has begun deploying a related platform known as AI Energy Mode for Econet Wireless. While AI Fuel Mode focuses on diesel generator use, AI Energy Mode is intended to optimise electricity consumption across the mobile network.
The two platforms could extend data-driven energy management across different parts of the telecommunications business. However, Econet has not disclosed measured electricity savings from AI Energy Mode, and the financial or environmental benefits of the newer platform remain to be demonstrated through operational results. The company expects both initiatives to contribute to demand for GPU computing capacity at Cassava’s AI factory, although it has not quantified the additional computing requirements.
The wider significance of the development lies in the relationship between digital infrastructure and energy efficiency. Telecommunications networks are essential to mobile banking, digital commerce, public services and business connectivity, but maintaining reliable coverage can carry substantial energy costs. Technologies that help operators manage power more effectively could improve operating efficiency while supporting the resilience of networks that businesses and communities depend on.
For tower infrastructure companies, combining AI-based energy management with solar power, batteries and more reliable grid connections could offer a more comprehensive route to reducing diesel dependence. AI can help identify inefficiencies and improve operating decisions, while alternative energy sources can reduce the need to run generators in the first place. The relative contribution of each measure will depend on local conditions and the economics of upgrading individual sites.
Econet InfraCo’s reported savings provide an example of how AI is being applied to a specific infrastructure problem rather than used solely for customer-facing digital services. Yet the wider case for adoption will require transparent performance data, including fuel consumption before and after deployment, the number of sites covered, implementation costs and the effect on network reliability.
As Cassava AI considers offering the technology to external customers, independently verifiable results will be important in demonstrating its commercial value. For telecommunications operators and other energy-intensive businesses across Africa, the opportunity is not simply to introduce AI, but to establish whether it can deliver repeatable cost savings while maintaining reliable operations.
The next stage will be determining how widely the platform can be deployed and whether the reported reductions can be sustained over time. If the results are replicated, AI-based fuel management could become one component of a broader strategy to improve energy efficiency, reduce exposure to diesel-price volatility and strengthen the resilience of critical infrastructure.
