• Strong collaborations between hardware and software stakeholders are pivotal in creating unified propositions and fostering the development of productivity-focused applications.
• Cloud deployment will act as a bottleneck for generative AI to scale due to concerns about data privacy, latency, and networking costs.

Smartphone and PC vendors have committed significant resources to produce chipsets that can handle inference workloads previously limited to the cloud.

Paul Schell, Industry Analyst at ABI Research, said that smartphone OEMs like Vivo and Samsung have started implementing heterogenous AI chipsets and investing in generative AI applications to deploy on their devices.

Moreover, he said that chip makers like Qualcomm and MediaTek have promoted their developer and optimisation tools to kickstart application development.

The same applies to the PC market also, he said, where chip vendors like AMD and Intel  have started shipping heterogenous AI chipsets for PCs, complemented by the concerted effort between OEMs and ISVs like Microsoft  to build optimised AI software.

According to ABI Research, this will unlock AI chipset shipments and revenue growth in these industries and extend to tablets and gaming consoles, reaching over 1.3 billion shipments in 2030. 

AI to play a central role

“We can observe all corners of the ecosystem rallying behind the potential of low-latency, data-private AI applications that can scale beyond cloud environments, although we are still at a very early stage,” Schell said.

According to Counterpoint Research stats, GenAI smartphone share of overall smartphone shipments will reach 11 per cent by 2024 and 43 per cent by 2027 to pass 550 million units in 2027 with fourfold growth.

But Gartner reports that worldwide shipments of AI PCs and GenAI smartphones are projected to total 295 million units by the end of 2024, up from 29 million units in 2023.

Gartner estimates 240 million GenAI smartphones and 54.5 million AI PCs will be shipped by the end of 2024, which will represent 22 per cent of basic and premium smartphones and 22 per cent of all PCs in 2024.

Tarun Pathak, research director at Counterpoint Research, said that more than 10 OEMs have launched over 30 GenAI-capable smartphones so far. 

Mohit Agrawal, Associate Director at Counterpoint Research, said that smartphones of the future will be more personalised to cater to individual needs and preferences, and AI will play a central role in driving these personalised experiences.

Currently, he said the AI use cases include enhanced imaging capabilities, translation features and improved app experiences, content recommendations, creating more personalised content, and more.

“The use cases will evolve as the large language models (LLMs) will continue to grow in both size and efficiency. We believe that the integration of edge (mobile devices) and cloud will be the mainstream model for generative AI in smartphones, and OEMs with an equally strong play in software capabilities, along with strategic industry partnerships, are likely to stay ahead of the competition.”

Counterpoint Research expects GenAI smartphones to hit an inflection point in 2025 as the devices permeate the broader price segments, especially the $400-$599 price tier.

“The ≥$600 and $400-$599 price bands will account for 9 out of every 10 GenAI smartphones sold in 2024. We expect Qualcomm to lead in the AI chip space in 2024, capturing almost half of all GenAI smartphone shipments, followed by MediaTek with a 13 per cent share,” Agrawal said.

Heterogenous AI chipset: Highly effective architecture

Schell said that heterogenous AI chipsets capable of distributing workloads between CPU, GPU, and NPU are highly effective architectures for tackling today’s diverse AI workloads – including generative and multimodal AI – and these systems will be adopted across device markets at an accelerating rate.

“More demanding on-device AI workloads in PCs will be addressed by GPU cards, like Nvidia’s  RTX, and AMD’s high-end Radeon hardware. This is why these chipsets, which reside outside of heterogenous systems, will remain in the AI mix going forward,” he said.

ABI Research believes productivity AI applications can reduce refresh cycles of smartphones, notebooks, and desktops – as well as other AI applications in tablets and gaming consoles – and provide the incentive to bring about this uptick in demand.

Research firm IDC expects 2024 to be an expansion year with the introduction of AI PCs, which will ultimately drive the market forward to 292.2 million units in 2028 and a compound annual growth rate (CAGR) of 2.4 per cent over the 2024–2028 forecast period.

Growth is expected to slowly ramp up over the year along with the availability of AI PCs, which will coincide with the beginning of a commercial refresh cycle in 2025.

 “Commercial buyers, both enterprise and educational, are on the cusp of a refresh cycle that begins later this year and reaches its peak in 2025. Many of these buyers are expected to be among the first in terms of AI PC adoption. The presence of on-device AI capabilities is not likely to lead to an increase in the PC installed base, but it will certainly lead to a growth in average selling prices,” Jitesh Ubrani, Research Manager with IDC, said.