Schneider Electric has warned that data centres worldwide are on the brink of an unprecedented shift as artificial intelligence moves from experimental deployments to full-scale integration in 2026, placing new demands on power, cooling, design and operational resilience.
The French company, which has a strong presence in Nigeria, said the disruptions seen in recent years are only the prologue to what is coming, noting that the next 12 months will push data centre operators to adapt faster than ever as AI inferencing, not just model training, becomes central to business operations across industries.
“AI is no longer just a tool; it is a transformative force that demands the right infrastructure to unlock its full potential,” Country President of Schneider Electric West Africa, Ajibola Akindele, said in a note shared with The PUNCH.
“In 2026, organisations that strategically invest in AI-ready data centres with advanced cooling, modular designs and energy-efficient solutions will gain the agility, resilience and competitive edge needed to lead in the AI-driven economy.”
AI usage has been rising rapidly since ChatGPT thrust the technology into the mainstream in 2022. According to McKinsey’s latest State of AI report, 78 per cent of organisations now deploy artificial intelligence in at least one business function, up from 72 per cent in 2024 and 55 per cent in 2023.
Sales and marketing still dominate adoption, but Schneider Electric said demand is intensifying in manufacturing, healthcare, finance and especially data centres, where operators are deploying AI-driven cooling systems, predictive analytics and energy-optimisation tools to cut waste and integrate renewable energy sources more efficiently.
Manufacturers using AI for demand forecasting have improved accuracy by a median of 30 percentage points. Hospitals are turning to predictive billing and appointment-scheduling tools, while financial institutions are enhancing fraud monitoring and risk decisioning with AI-backed systems.
As AI agents capable of operating with minimal human supervision become standard in enterprise workflows, the company said the pressure on data-centre capacity, particularly for real-time inferencing, will accelerate sharply.
Schneider Electric said the coming era will be defined by the rise of “AI factories,” specialised data centres designed to generate intelligence rather than simply store or process information. These facilities will handle model training, fine-tuning and continuous inferencing workloads that produce real-time insights for businesses.
Inferencing demands vary widely, from under 20 kW for compressed models to as high as 140 kW per rack for advanced agentic systems. By 2030, the company projects that a quarter of new data centre builds will operate below 40 kW per rack, half will fall between 40 kW and 80 kW, and the remaining 25 per cent will exceed 100 kW per rack, driven by large-scale training needs.
The shift will be supported by next-generation hardware such as NVIDIA’s Rubin CPX GPUs and Vera CPUs, integrated into ultra-dense MGX NVL144 CPX platforms capable of delivering 8 exaflops of AI compute, roughly 7.5 times more AI performance than previous systems.
The company said robots will increasingly augment data centre operations, performing activities ranging from security surveillance and cable management to server installation and maintenance of liquid-cooling systems. Beyond data centres, Schneider Electric expects robotics to expand in fields such as agriculture, firefighting, healthcare and disaster response, leveraging AI for more autonomous operations.
Digital twins, virtual replicas of physical systems, will become a defining feature of next-generation data centre design. Schneider Electric said tools such as ETAP modelling, integrated with NVIDIA Omniverse, will allow operators to simulate and optimise electrical infrastructure before equipment is physically deployed.
The company said this approach will significantly reduce design errors, improve energy performance, and help operators adapt to rising densities brought on by AI workloads.
AI rack densities are projected to reach 240 kW per rack in 2026, making traditional air cooling insufficient for the coming wave of high-performance hardware. Research is already exploring even higher densities, with 1 MW per rack expected by 2028 and future systems potentially reaching 1.5 MW.
To manage this, Schneider Electric said liquid-cooling solutions, once niche, will become the standard across new builds and retrofits. These systems offer higher efficiency and lower energy use, which is critical for supporting extreme compute demands.
With many companies unable to afford new hyperscale centres, Schneider Electric expects a surge in brownfield retrofits. These upgrades will include larger IT racks, higher-capacity PDUs and bolt-on liquid-cooling installations that allow older facilities to handle AI workloads without full rebuilds.
“This is how smaller enterprises can remain competitive in the AI era,” the company noted.
The company warned that as AI accelerates, sustainability will become an even more pressing concern. Data centre operators are increasingly adopting diversified power strategies, including natural gas turbines with carbon capture, HVO-fuelled generators, geothermal systems, battery storage and renewable energy sources.
Today, renewables provide about 27 per cent of data-centre electricity, and total renewable generation is projected to grow 22 per cent annually until 2030, enough to supply nearly half of future demand.
The multinational said 2026 marks the point where AI moves from disruption to foundation, reshaping the structure and purpose of data centres globally. From liquid-cooled AI factories to digital twins and large-scale retrofits, the company said operators must embrace global partnerships, smarter infrastructure planning and external expertise to remain competitive.
“Data centres will no longer just support technology; they will enable intelligence itself,” Akindele said.