Operators to play a 'tricky game' with AI-capable smartphone sales – Canalys report

While an increase in availability of AI-capable smartphones could drive up smartphone sales, this shift could also pose an immense infrastructure challenge to service providers, according to an analyst with research group Canalys.

Canalys defines a GenAI-capable smartphone as including dedicated AI hardware such as ASICs, which can efficiently run generative AI models and support real-time, on-device inferencing with minimal latency.

Future AI capabilities could provide operators with "new monetization routes" and "revenue, sharing agreements," but "it's a huge challenge for actually running this infrastructure," Runar Bjorhovde, analyst for Canalys, told Light Reading.

"Over the last few years we've seen more and more data being used [by smartphones]… and we don't expect that AI will take less of a load on the infrastructure," Bjorhovde explained.

Data traffic per smartphone averaged 21GB per month in 2023 and is predicted to reach 56GB per month by 2029, according to an Ericsson report.

Service providers are also pushing hard to have more AI-capabilities run on the device itself, versus over the Internet and public cloud. More on-device AI capabilities reduces the network traffic load, and operators have more control over privacy and security, said Bjorhovde.

"It's a tricky game to find the right setup between what happens on the device and what happens in the cloud. And currently, there's a data security angle on top of it," he said.

Respondents' highest ranked concern was the privacy and data security risk of running AI on smartphones. (Source: Canalys, May 2024)

For example, running ChatGPT across apps on a single device for hours a day "could be a tremendous extra load coming onto networks" in addition to managing energy costs, explained Bjorhovde. He predicted that service providers will tout their ability to provide more security as a way to bring consumers on board with the notion of running more AI features on the device itself.  

Demand for AI-capable smartphones

In addition to ensuring more on-device AI use, service providers will be challenged with successfully convincing consumers to upgrade to AI-enabled smartphones in the first place.

"Most service providers are also the largest channel for selling smartphones. If you look at the US, 85% of smartphones are sold through the carriers in the US. In Europe it's perhaps around 40%," said Bjorhovde.

In a report, Canalys forecasted that 16% of global smartphones shipped this year will be AI-capable, and that number will rise to 54% by 2028. AI-capable smartphones will achieve a 63% CAGR from 2023 to 2028, "driven by rapid advancements in chipset technologies and growing consumer demand for AI-driven features," according to the research firm.

Tech enthusiasts, who Canalys reports as representing 7% of AI smartphone buyers, present the best opportunity for service providers to upsell new AI-capable smartphones, said Bjorhovde. This group represents high-value customers interested in upgrading to the latest technology, and they proactively seek out information on new products.

On a regional level, smartphone buyers in China and India appear to have a "much stronger inclination toward the latest innovation. These are areas where people really, really care about what the latest tech is and it's not just a question of AI – people care a lot more about what the specs are," said Bjorhovde.

Interest in AI features by region. (Source: Canalys, May 2024)

In Europe, by comparison, "sustainability is coming up on the list instead, of things people care about," he added. The European market could be more challenging to convince when it comes to selling new AI-capable smartphones.

"It doesn't mean that there's not an opportunity," Bjorhovde said. "I just think it has to be phrased and framed very differently."