GenAI to be key design vector for smartphone OEMs

• As generative AI transitions from a novel concept to a mainstream tool, its integration into smartphones presents both opportunities and challenges for OEMs.
• GenAI-enabled smartphones to surge by 364% year-over-year in 2024 to reach 234.2m units.
• Advent of cloud-based AI solutions is expected to extend GenAI capabilities to mid-range and entry-level smartphones.

As generative artificial intelligence (GenAI) becomes increasingly mainstream, its incorporation into smartphone design is rapidly evolving into a fundamental differentiator for original equipment manufacturers (OEMs).

According to a recent forecast by IDC, the shipment of GenAI-enabled smartphones is anticipated to surge by 364 per cent year-over-year in 2024, reaching 234.2 million units.

Projections suggest this figure will rise to 912 million units by 2028, marking a remarkable compound annual growth rate (CAGR) of 78.4 per cent from 2023 to 2028.

Over the past year, Kiranjeet Kaur, Associate Research Director, Data & Analytics at IDC Asia/Pacific, said that numerous key players in the market have revealed their unique artificial intelligence (AI) tools, showcasing significant advancements in smartphone user experience (UX).

“These innovations are underpinned by various foundational models, including large and small language models (LLM and SLM), facilitating GenAI capabilities through both on-device processing and multimodal input/output.”

The OEMs employ a hybrid approach, balancing device-based AI for localised tasks with cloud-based solutions for more computationally intensive activities.

Currently, Kaur said that most AI features are primarily available on premium devices; however, the advent of cloud-based AI solutions is expected to extend these capabilities to mid-range and entry-level smartphones.

The democratisation of AI features will not only enhance affordability but also broaden accessibility, although concerns regarding privacy and latency must be carefully navigated.

Distinctive AI features

Each OEM distinguishes itself in the competitive landscape of smartphone AI through unique features tailored to the user experience. While the overarching categories of AI applications may be consistent—ranging from photo and video editing to live translation—the specific implementations reflect each brand’s identity.

For instance, prominent technologies like Google and OpenAI are being integrated to enrich user engagement.

Among the standout features gaining traction are Google’s Circle to Search and OPPO’s Eraser. The former has been highlighted by Samsung as the most utilised feature tied to Android 14, while the Eraser function reportedly sees an average usage of 15 times per day by its users.

Additionally, live translation tools have emerged as essential, effectively breaking down language barriers, even in offline settings. Increasingly, OEMs are also focusing on health and wellness applications, exemplified by features such as Google’s Sleep and Snore detection and Samsung’s Sleep Apnea detection.

Beyond functionality, Kaur said that the choice of LLMs and the training data used is pivotal in influencing GenAI performance and overall UX. For instance, Apple has developed its own language models while leveraging OpenAI for more complex tasks.

In contrast, Android-based smartphones utilise diverse models, including various sizes from Google’s Gemini suite. OPPO, for its part, has introduced its own SLM with 7 billion parameters alongside AndesGPT, complementing its use of the Gemini models. Partnerships with technology giants like Qualcomm, MediaTek, and Microsoft further enhance the capabilities of GenAI applications.

Meeting user expectations amidst AI hype

“The challenge for OEMs lies in delivering a user experience that aligns with the high expectations set by AI hype. As consumer fatigue with conventional hardware-centric enhancements grows, there is a significant opportunity to rejuvenate interest through innovative AI features,” Kaur said.

Just as dictation, text predictions, and advanced photo editing have become integral to the smartphone experience, she said the new wave of GenAI applications is set to redefine everyday smartphone interactions.

Key actions OEMs need to take

• Overall, a more personalised, intuitive and user-friendly experience that doesn’t interrupt the normal workflow can create more stickiness. While iPhone users have high loyalty, it’s not the same for Android users who easily switch between brands. A radical shift such as opting for an app-less device is not yet for the masses, while a traditional interface with an app may be a bit outdated and inefficient for an evolved user. An in-between approach with a generative user interface could take the users on the AI track more gradually.  
• Network connectivity will also play a key role to have a faster, smoother connection and a seamless experience for using AI features, especially if there is cloud processing. Together with faster processing on the device, Wi-Fi 6E/7 and 5G network connectivity will help in faster response with lower latency. This will also help with democratization of AI by enabling more smartphones with cloud-AI features.
• User privacy will remain central to these experiences, partly enabled by on-device AI but also reflected in a company’s philosophy to ensure user data remains private while also contextualizing it to provide more personal results. The side benefit of on-device AI is that by spreading out the processing to on-device AI across the installed base of devices, much less server CAPEX is required than would be if AI was only processed on servers.
• As these features go beyond the native apps and start to be integrated in the third-party apps, it will be imperative to work with the developers to bring AI features to more apps and ensuring easy ways for developers to integrate AI features into their apps.