The dominance of massive data centres may one day be challenged by everyday devices such as smartphones, according to Perplexity CEO Aravind Srinivas. In a recent podcast interview, he suggested that future AI systems will be powerful enough to operate directly on personal devices, eliminating the need to send data back and forth to large, remote servers.
Instead of relying on vast data centre networks, Srinivas believes people will use advanced, personalised AI tools running on the hardware already built into their phones, laptops, and tablets.
On-Device AI Is Already Emerging
Some technology companies have already begun moving AI processing closer to users. Apple’s AI system, Apple Intelligence, runs certain features directly on specialised chips inside its latest devices. This allows for faster performance and better protection of personal data.
Similarly, Microsoft’s Copilot+ laptops include built-in AI processing capabilities. However, these features are currently limited to high-end, premium devices. Most consumer hardware still lacks the processing power required to run advanced AI locally.
Jonathan Evans, director of Total Data Centre Solutions, says it will take time before efficient AI systems can run on standard devices. “It’s a long-term ‘if and when’ scenario,” he explained.
Traditional Data Centres Are Still in High Demand
Despite the interest in smaller and localised AI systems, the global demand for data centres continues to grow. These facilities support everything from video streaming and online banking to cloud storage and AI operations.
Large organisations often own their own data centres, while smaller businesses rent computing capacity. Almost every online service relies on data centres somewhere in the world.
The Rise of Micro Data Centres
In recent years, smaller data centres have begun appearing in unexpected places. One early example in Devon, UK, involved a compact data centre about the size of a washing machine. The heat it produced was used to warm a public swimming pool.
In November 2025, a British couple revealed they were heating their home using a small data centre installed in their garden shed. A month later, a university professor reported using a powerful GPU under his desk, which both processed AI workloads and heated his office.
These examples show how compact data centres can provide computing power while also serving practical purposes like heating.
Big Investments in Massive AI Facilities
While small-scale solutions are emerging, major technology firms are still investing billions in large data centre complexes. Around 100 new facilities are currently under development in the UK alone.
Data centres consume vast amounts of energy, raising concerns about their environmental impact. However, Nvidia CEO Jensen Huang describes them as “AI factories,” arguing that large-scale infrastructure is essential to support the rapid growth of artificial intelligence.
Small vs Large: What’s the Future?
Although compact data centres and on-device AI are gaining attention, traditional data centres remain central to the digital economy. The future may involve a combination of both — powerful central facilities supporting global services, alongside smaller, local systems handling personalised AI tasks closer to users.