'The 1000 TOPS consumer': How the NPU-everywhere strategy could turn every user into a walking supercomputer by 2030

• Combined device AI compute could exceed 1,000 TOPS by the decade's end
• Smartphones, wearables, and earbuds are becoming key distributed AI processors
• Average users likely carry hundreds of TOPS across multiple personal devices

Personal electronics are heading toward a point where combined AI compute across everyday devices rivals systems that once filled dedicated facilities, according to a Futuresource CE analysis tracking edge AI silicon trends through 2030.

The report examines how neural processors are spreading across smartphones, wearables, and audio devices, and how performance growth across those categories could change our expectations for personal computing power.

Smartphones are, naturally, central to all this, with flagship chips from companies such as Qualcomm, MediaTek, Samsung, and Apple now delivering up to 100 TOPS of neural processing capability.

Growing NPU performance

Forecasts suggest smartphones alone could nearly triple their NPU performance by the end of the decade.

Smartwatches are no longer trailing quietly behind smartphones either, as dedicated neural processors are beginning to appear in smartwatch chips, a step beyond earlier designs that relied heavily on shared processing blocks.

Shipments of smartwatches reached about 94 million units globally in 2025, showing how widespread they now are.

Wireless earbuds are becoming ever more popular too, with 360 million units shipped annually. Each earbud carries its own chip, so the silicon footprint reaches well over 700 million units every year.

That spread of AI-capable hardware across multiple devices supports a broader vision often described as the “walking supercomputer.”

“These are not speculative scenarios,” Simon Forrest, Head of Core Technology at Futuresource Consulting, said. “They are the logical product of chip design trends already in motion. Edge AI offers real advantages in speed, privacy and cost, and traditional coded algorithms are being replaced by machine learned versions that increase efficiency while expanding capabilities. For CE brands, understanding where AI compute is heading, and what the silicon enables, is becoming a fundamental strategic necessity.”

Forrest told us that it would be feasible for someone in 2030 to be carrying personal electronics with a combined AI compute exceeding 1,000 TOPS (1 POPS), although it wouldn't be commonplace.

He said, “Futuresource forecast modelling shows the average is more likely to be in the range of 450 to 550 TOPS by the end of the decade, assuming a person carries a smartphone, laptop, smartwatch, plus smart glasses and perhaps one other wearable device. Nevertheless, this is still a significant amount of distributed AI computation capacity positioned in and around the body.”

When combined with advances in laptops, smart glasses, and wearables, the aggregate compute figure is becoming more widely discussed alongside single-device performance.

Marketing language often leans on headline TOPS figures, although raw numbers alone don’t capture real-world performance. Architecture design, memory bandwidth, and software optimization remain just as important when translating theoretical compute into practical AI tasks.

The move toward distributed processing across multiple devices is reducing reliance on cloud services, improving response times, and keeping sensitive data closer to the device itself.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.