A coming wave of AI-capable smartphones may let vendors distinguish their devices via unique features and user experience, but it also poses challenges for privacy and potential user disillusion if there is too much hype.
It isn’t just PCs that are being gussied up with AI capabilities, so are smartphones. In fact, smartphone platforms such as those from tech giant Qualcomm have been infusing them with AI algorithms for some time for purposes such as enhancing photos. But now, full-blown generational AI models are starting to be supported.
Analyst Canalys says in a new report that it expects 16 percent of smartphones shipped globally this year to be AI-capable, and forecasts this will rise to 54 percent by 2028, driven by advances in phone chipset technology and growing consumer demand for AI-driven features.
The company defines a Gen AI-capable smartphone as one equipped with dedicated AI hardware capable of executing generative AI models and supporting real-time, on-device inferencing with minimal latency.
Qualcomm’s Snapdragon platforms with the company’s Hexagon NPU (Neural Processing Unit) fit the bill here, as does MediaTek’s APU (AI Processing Unit) and Google’s TPU (Tensor Processing Unit).
Canalys sees smartphone GenAI use cases falling into three key areas: personalizing user experiences, streamlining user experiences, and multimodal capabilities.
The first involves GenAI applications on smartphones using on-device models fine-tuned with data drawn from chat, context and geolocation to “enhance user interactions significantly” via AI agents and privacy-protected service recommendation engines.
Multimodal capabilities means using a smartphone’s sensors such as the camera and microphone to process inputs including text, images, and voice simultaneously to understand and respond to users “in a more holistic and natural manner”.
Delivering a more streamlined experience means merging various tasks into unified ones through continuous background operations, according to Canalys. It cites the example of Honor’s Magic Portal, which can simplify multi-step processes to a single action using the intent-based capabilities of generative AI models.
However, such application scenarios are still in their infancy, the analyst notes, and it expects to see AI agents gradually evolve to manage a broader range of tasks as vendors try to increase user satisfaction and boost brand loyalty.
One factor most likely to affect the success of AI smartphones is whether this is something users actually want. Canalys carried out a study resulting in what calls the “AI Inclination Index” based on six survey questions, and claimed that overall, the consumer audience is “cautiously positive” about AI capabilities.
Canalys claims that respondents with greater exposure to and engagement with technology products and services – those who regularly use multiple connected devices, upgrade their phones frequently, use premium smartphones, or have high daily screen time – were more likely to show high or very high interest in AI capabilities on smartphones.
In contrast, those who prioritize value for money and primarily purchase devices when they need to replace an old one showed less interest in AI and found few possible use cases that were relevant to them.
The figures indicate that about 25 percent of respondents showed high or very high interest, with 34 percent having “some” interest, 33 percent “little” interest, and seven percent having no interest at all. Not exactly a rousing endorsement that consumers are gung-ho for AI, in our opinion.
Canalys also found differences between various regions worldwide, which it said highlights the importance of localized marketing campaigns. “AI inclination” was strongest in places such as China and India, with 43 percent expressing high or very high interest in China.
In contrast, 45 percent expressed little interest in the United States, rising to 52 percent in Germany. The study encompassed 4,952 individuals.
Canalys also looked at whether AI capabilities had the potential to sway phone users to switch brands. The results showed geographical differences again, but broadly, users considered better specifications and value for money as more important than AI features.
In no surprise for anyone, 66 percent of Apple users said they would not consider switching for any reason, while the figure for Samsung owners was 25 percent.
As far as user concerns with AI go, the survey found privacy and data security risks were by far the biggest factor. This is likely apprehension about the potential collection, storage, and utilization of personal data by AI algorithms, Canalys says. More than half of respondents indicated they were concerned with AI access to their personal data, something vendors should take note of.
Summing up its findings, Canalys sorts consumers into four segments: tech enthusiasts, value optimizers, indifferent pragmatists and skeptics.
Indifferent pragmatists make up the largest segment at 67.9 percent, and the advice from Canalys is that “AI innovation can be an add-on in a wider proposition” for this audience, which again hardly makes it sound like buyers will be breaking down the vendors’ doors to get their hands on AI smartphones.
Meanwhile, skeptics are estimated to make up 7.2 percent of consumers, but “should not be a target group for leading vendors and will instead be an opportunity for a niche vendor to capture.” Hear that? If you aren’t convinced by all the AI flannel, you are a niche market.
Canalys says that current AI smartphone purchases are being driven by tech enthusiasts, but that a second wave of adoption, expected through 2025 and into 2026, will be driven by the value optimizers. Vendors will need to invest in their products and marketing to build confidence and awareness with these buyers as they seek to upgrade their devices.
It isn’t all going to be plain sailing for the vendors, however. Canalys warns that as devices with GenAI features start to enter the market, computing costs to run the AI model will start to ramp up exponentially for vendors. Marketing and customer support programs will be required to drive consumer interest and adoption as well as education in every step of the customer journey, it adds.
However, on-device AI is set to enable new revenue models for vendors, and ideas around these are only in the early stages, Canalys says. The primary objective for many early movers at this stage is to lay the groundwork for foundational AI capabilities, while driving user adoption of their specific AI services.
In the view of Canalys, there are at least 3 business model/product development stages for smartphone vendors: AI-as-a-feature, AI-as-a-service and AI-as-interface, depending on each vendor’s resources and strategic choices.
The first sees vendors already integrating generative AI functions directly into apps or as features on smartphones, positioning them as key marketing points. These typically involve specific use cases or tasks, citing examples such as “AI eraser,” or “circle to search.”
With AI-as-a-service, the key advance is personalization, Canalys says. By providing users and third-party developers with the right models and tools, vendors will be able to improve user engagement and “increase user willingness to pay.”
For AI-as-interface, Canalys points to personal AI agents that can perform tasks using multiple apps and services. Vendors will be able to differentiate their product based on the capabilities of their AI agents, it says. And as these AI agents are trained on each user’s actions and other contextual information, they “make it more costly for users to switch brands.”
Not forgetting the enterprise market, which Canalys also sees as an opportunity, if not as big as that for consumer AI smartphones.
To bolster AI smartphone adoption within enterprises, smartphone makers must address areas aimed at boosting user experiences and productivity, as well as fostering business growth within the mobile ecosystem.
The point here is striking a balance between data privacy and security, ensuring robust measures are in place while upholding user privacy standards.
Providing comprehensive developer support and partnerships will be crucial, and Canalys points here to the collaboration between Samsung and IBM that sees Watson’s AI capabilities integrated into Samsung’s enterprise solutions like Knox.
Overall, Canalys claims that the industry is now at a critical juncture. GenAI in smartphones promises “transformative changes” to user experiences and vendor business models, but on the flip side it also presents privacy concerns and market acceptance challenges.
“The success of AI smartphones will depend not only on technological advancements but also on the ability to ethically and effectively integrate these technologies into everyday consumer experiences,” Canalys concludes. ®
![](https://techcodex.com/wp-content/uploads/2023/11/stsmall507x507-pad600x600f8f8f8-9-e1701329771970.jpg)
Alex Mitchell is your go-to expert for all things mobile. With a passion for the latest smartphones, apps, and mobile innovations, Alex provides in-depth reviews, insightful analyses, and breaking news about the ever-evolving world of mobile technology. Stay connected with Alex to navigate the fast-paced realm of mobile devices.