The Shift Towards Smaller Data Centres: A New Era in AI Processing

Introduction

In a world increasingly reliant on artificial intelligence (AI), traditional data centres, often massive and costly, are being scrutinized. While companies invest heavily in these vast facilities, some experts argue that we may be moving towards a more efficient model: smaller, localized data centres that can operate in sync with the devices we already own.

The Case for Smaller Solutions

Aravind Srinivas, CEO of Perplexity, made an intriguing statement on a recent podcast: the day might come when our smartphones will house the AI processing power currently confined to data centres. This perspective isn't unfounded; companies like Apple and Microsoft have started utilizing specialized hardware within their devices to handle tasks locally. For instance, Apple's Apple Intelligence leverages its chips to enable on-device AI functionalities, thereby enhancing speed and ensuring user data remains private.

“It's long term 'if and when' powerful and efficient AI can run on local devices.” — Jonathan Evans, Total Data Centre Solutions

Despite the high-end technology we see in devices today, there is still a notable range between what is available and the capabilities required for advanced AI processing. Not every smartphone is equipped to handle such tasks. Evans points out that while demand for data centres is increasing, the solution doesn't necessarily involve larger facilities.

Transforming Spaces

Several years ago, I encountered a “data centre” the size of a washing machine warming a public swimming pool in Devon, UK. This sparked my skepticism, but it also ignited my curiosity about the potential for smaller-scale operations. Fast forward to November 2025, when a couple shared their experience of running a data centre from their garden shed, warming their home in the process.

Reimagining Infrastructure

The tech industry is overwhelmingly focused on building large, energy-intensive facilities. As of now, over one hundred new data centres are under construction in the UK alone. These facilities consume immense power and raise concerns about their environmental impact. Yet, there's a burgeoning concept suggesting that urban environments could house smaller data centres that double as heating solutions.

• Data Centre Shed Cuts Energy Costs
• Future of Data Centres in Space

The Changing Dynamics of AI Processing

Numerous tech leaders are discussing a shift towards smaller edge data centres, especially close to major populations. Mark Bjornsgaard, founder of DeepGreen, believes these smaller units should collaboratively work when needed. “London is just one giant data centre that hasn't been built yet,” he argues, promoting an idea where every public facility could contribute to the digital infrastructure.

Economic Considerations

Smaller data centres can also offer significant savings on the bottom line for businesses. Many are turning to bespoke enterprise AI tools that, while initially costlier, offer tailored solutions that don't draw from a common pool of data for learning. Dr Sasha Luccioni, from Hugging Face, describes this paradigm shift: "We are already seeing a switch between large models taking huge resources, to smaller models being more bespoke and running more locally." This transformation raises questions about the ideal size and scope of data processing capabilities.

Security Concerns

Could decentralizing data centres expose vulnerabilities? Experts debate this, with security specialists suggesting that smaller targets may mitigate significant risks. "Larger data centres can serve as substantial failure points," says Professor Alan Woodward of Surrey University, highlighting that less centralized models may enhance resilience while also providing environmental benefits.

A Glimpse into the Future

The appeal of localized data solutions is growing, backed by evidence of high performance requiring less resource expenditure. Moving away from large, energy-hungry data centres reflects a broader trend toward eco-consciousness in tech. With increasing demands for efficiency and sustainability, the challenge remains: can we adapt our infrastructure to meet tomorrow's needs?

Conclusion

This conversation about smaller data centres versus traditional giants isn't simply about space. It's a critical examination of how we conceptualize and use technology in our daily lives. As we venture forward, the infrastructure we build will shape our digital landscape, determining the efficacy and sustainability of AI.