Dell Predicts Major Data Centre Architecture Shift for AI
Dell Technologies, the global technology infrastructure provider, forecasts a transformation in data centre architecture as enterprises move from testing to implementing AI systems in 2025.
While AI systems currently represent a fraction of IT energy consumption, usage will inevitably increase with adoption rates. This growth in AI processing demands raises questions about power usage effectiveness and environmental impact.
Steve Young, UK Senior Vice President and Managing Director at Dell Technologies, reports that 70% of UK firms have secured returns on investment from generative AI (Gen AI) implementations following initial trials in 2024.
"For many, the test-and-learn phase is already starting to pay off," he says.
Dell targets AI processing requirements
Dell Technologies’ AI Factory framework aims to support organisations integrating AI across multiple infrastructure types, from edge computing to centralised data centres. This is in tandem with projections indicating AI workloads will constitute more than half of data centre processing requirements by 2026.
Organisations that fail to adopt the right AI strategy and architecture will be at a disadvantage
Steve indicates that 90% of AI processing will focus on inference operations by 2025. Such a shift necessitates a move away from traditional fixed-ratio infrastructure.
"We'll see traditional siloed data centres being replaced by disaggregated architectures, which allow computing, storage and networking systems to scale independently," Steve says.
Healthcare, education, government and retail sectors are emerging as primary targets for AI implementation. These sectors process substantial volumes of data through their operations, requiring robust infrastructure support.
Dell projects a rise in edge computing via AI-enabled PCs
The transition extends beyond data centres to endpoint devices. Dell has also forecasted increased adoption of PCs with neural processing units (NPUs) - specialised chips designed for AI workloads.
Edge computing, which processes data at the point of creation rather than in centralised facilities, will expand through these AI-capable devices.
"With AI PCs, data can be processed directly on the device to drive real-time insight, potentially reducing cost and allaying security considerations," Steve says.
The market for AI-optimised processors continues to expand. This includes central processing units (CPUs), graphics processing units (GPUs) and NPUs. In fact, Dell projects universal inclusion of dedicated AI accelerators in PCs by 2030.
The timing aligns with Microsoft's announcement regarding Windows 10 end of support, which necessitates hardware upgrades to meet Windows 11 specifications. This transition provides organisations with an opportunity to integrate AI processing capabilities into their computing infrastructure.
Infrastructure shaped by environmental concerns
The UK government's designation of data centres as critical national infrastructure (CNI) coincides with requirements for an 81% reduction in emissions by 2035.
Its move to reduce planning restrictions for data centres adds urgency to environmental considerations. Organisations face pressure to balance increased computing demands with sustainability targets.
Steve emphasises the importance of infrastructure planning: "The focus should shift to developing tailored, scalable AI applications that solve current challenges while also positioning companies for future opportunities."
Infrastructure audits are becoming essential as organisations prepare for AI integration. These assessments examine current computing capabilities against future requirements, encompassing both PC fleets and data centre resources.
Dell's AI Factory framework addresses these challenges by providing a structured approach to AI deployment. The system enables organisations to maintain flexibility while scaling their AI operations across different infrastructure types.
The focus should shift to developing tailored, scalable AI applications that solve current challenges while also positioning companies for future opportunities.
The framework also recognises the increasing complexity of data centre management as AI workloads grow. It accommodates the shift from traditional architectures to disaggregated systems, where computing, storage and networking components scale independently.
"Organisations that fail to adopt the right AI strategy and architecture will be at a disadvantage," Steve says.
Explore the latest edition of Data Centre Magazine and be part of the conversation at our global conference series, Tech & AI LIVE and Data Centre LIVE.
Discover all our upcoming events and secure your tickets today.
Data Centre Magazine is a BizClik brand
