LONDON: PCs will be the main driver of growth in the AI edge processor market over the next four years, according to new research from Omdia.
This was among the results announced by Omdia in the 2023 edition of its AI Processors for the Edge Forecast, which covers the market for artificial intelligence processors across ten device types including smartphones, PCs, tablets, smart speakers, robots, and UAVs, among others. The total market for AI processors at the edge is expected to grow from $31bn in 2022 to $60bn in 2028, a CAGR of 11%.
Several factors are leading to PCs’ growing importance in this market. Product availability is certainly a key factor with Intel’s Meteor Lake CPUs recently joining AMD’s Ryzen 7040, 8040, and 8000G processors, Apple’s M-series, and Qualcomm’s Snapdragon Elite X. Also playing a role is the expected rebound of PC market which has slowed recently after a COVID-19 pandemic related surge in 2020 and 2021. Beyond 2025, though, the machines sold during the surge will start approaching the end of their accounting lives, creating the possibility of a refresh cycle that will help with AI PC adoption.
“This is especially important as demand from smartphones – an early-adopting sector – is expected to slow” said Alexander Harrowell, Omdia’s Principal Analyst for Advanced Computing. “Saturation is certainly a factor in the smartphone market, as over two-thirds of them by volume now have some form of AI acceleration and growth is concentrated in the premium price tier.”
The forecast splits the market by processor types such as in-CPU accelerators, AI-ASICs, AI-ASSPs, GPUs, and FPGAs and power/performance bands across the 10 device classes.
As discussed in the report, systems-on-chips introduced in 2018-2019 with basic acceleration are now beginning to drop out of vendors’ product lines in favor of more powerful ones. As a result, the lowest and middle performance tiers are emptying out and the next ones up growing, with the strongest growth in the top (>30 TOPS) group. The bulk of growth will come from proprietary AI-ASSPs such as Qualcomm’s Snapdragon and Mediatek’s APU.
Harrowell added, “AI ASSPs will push from 19% to 28% of the market, largely at the expense of GPUs. PCs are beginning to look more and more like smartphones or tablets as they adopt the CPU-GPU-NPU architecture familiar from nearly all modern smartphones, while in-CPU acceleration has been unexpectedly slow to take off.”
