• Source:JND

Perplexity CEO Aravind Srinivas predicts the next great leap forward for artificial intelligence will not come from larger data centres, but from powerful AI running directly on consumer devices. Speaking on a podcast with YouTuber Prakhar Gupta, Srinivas foresaw an AI future where intelligence shifted away from centralised servers onto local chips – fundamentally altering how AI is created, deployed, and monetised.

Why On-Device AI Threatens Traditional Data Centres

Srinivas asserts that one of the greatest risks facing today's data centre-centric AI ecosystem is the possibility of embedding intelligence on devices.

“The biggest threat to a data centre is if the intelligence can be packed locally on a chip that's running on the device, and then there's no need to run inference on all of it on one centralised data centre,” he said.

ALSO READ: Centre Issues Notice To X, Flags Misuse Of Grok For Obscene, Sexually Explicit Content; Gives 72-Hour Ultimatum To Act

At present, most AI applications rely on large cloud infrastructures for inference. This requires constant connectivity, high energy use and billions in investment; Srinivas suggested that if AI models became efficient enough to run locally on smartphones, laptops or wearables instead of large data centres, then economic justification might weaken significantly.

Transitioning towards decentralised AI would move the industry towards more widespread intelligence rather than being concentrated within just a few server farms worldwide.

Efficiency, Energy, and Human Intelligence

Srinivas made comparisons between artificial intelligence and biological intelligence, drawing attention to one key weakness of modern AI systems: energy consumption is far higher in modern data centres when measured per watt of computation power than with human brains.

He noted that human intelligence is driven by curiosity and the capacity to question assumptions and reinterpret familiar ideas in new ways, an intrinsic motivation missing in AI systems, which instead focus on optimisation and pattern recognition rather than genuine exploration.

From Chatbots to Agents and the Future of Work

Srinivas discussed more than hardware and infrastructure when discussing AI, from simple chatbots to autonomous agents capable of performing complex tasks on behalf of users. He believes highly personalised AI could have a democratising effect.

ALSO READ: AI Could Drive Over 2,00,000 Job Cuts In European Banking Sector By 2030: Report

As smartphones bridged the divide between individuals and large institutions, on-device AI could give individuals access to powerful tools without incurring costly cloud service fees. Srinivas noted that age is no barrier when adopting AI; rather, one's curiosity and willingness to experiment determine their success using these tools.

What This Shift Could Mean for the AI Industry

Srinivas' vision could bring significant alterations to the AI industry. Chipmakers, device manufacturers and software developers could gain influence, while cloud providers and hyperscale data centres might gradually diminish in influence.

At present, large models still rely heavily on centralised infrastructure; however, as efficiency improves and on-device inference capabilities expand, Srinivas believes the balance of power in AI could shift, signalling its next revolution.


Also In News