AI Large Models Boost Hopes for Consumer Electronics Industry's Revival as Peak Season Approaches
-
After years of sustained market decline, the consumer electronics supply chain is leveraging AI large models to initiate a self-rescue effort within the industry.
On August 14, Xiaomi became the latest smartphone manufacturer to announce the integration of AI large models into its devices. Earlier, Huawei introduced its HarmonyOS 4 with the Pangu AI large model, while Honor CEO Zhao Ming hinted in late June that Honor would also bring AI large models to its devices, though the exact timeline remains unclear. Additionally, Apple, OPPO, and vivo have all identified embedding AI large model capabilities into smartphones as a key development direction. Qualcomm has also showcased its next-generation processor capable of supporting AI large model inference.
Multiple interviewees told Securities Times that embedding generative AI technology into smartphones and other consumer electronics will bring disruptive innovation to the industry, driving a new wave of hardware and software upgrades and revitalizing the long-sluggish consumer electronics market. "With the support of AI large models and the arrival of the peak season, industry leaders like Apple and Huawei are set to release flagship products this fall. Everyone is hopeful for a revival of the industry," said an executive at a Chinese A-listed consumer electronics company recently.
According to Wang Jiping, Vice President of IDC China Research, AI integration into devices has become a critical direction for AI development and implementation. By 2026, nearly half of all device processors in China will feature AI engine technology, heralding an exciting surge in growth for hardware-enabled AI processors. This trend is particularly significant in traditional hardware markets like smartphones, tablets, laptops, and desktops, serving as a key driver for the industry's recovery in the post-pandemic era.
AI Large Models Accelerate Adoption
On the evening of August 14, Xiaomi CEO Lei Jun's annual speech went viral. During his impassioned three-hour presentation, he announced that Xiaomi smartphones would integrate on-device AI large models.
Lei Jun stated that AI large models represent a major technological revolution, and Xiaomi is fully committed to embracing them. The company formed a large model team in April and plans to apply the technology across its business. Its voice assistant "Xiao Ai" will be the first to receive an upgrade, with an invitation-only beta test starting on August 14. Reports indicate that Xiaomi currently employs over 1,200 researchers in AI-related fields.
Similarly, 10 days earlier, Huawei unveiled its HarmonyOS 4 system embedded with the Pangu AI large model at its developer conference. This upgrade empowers Huawei's smartphone assistant "Xiaoyi" with AI capabilities, enhancing interaction, productivity, and personalized services. Users can now engage with Xiaoyi using natural language, and the assistant can generate essays, images, and videos.
Beyond Huawei and Xiaomi, other companies are also actively deploying AI large models. In late June, Honor CEO Zhao Ming noted that the biggest factor influencing the consumer electronics industry is the innovation cycle. The current smartphone industry is entering a new phase driven by AI and 5G+, presenting fresh opportunities. Honor plans to seize this moment by innovating in AI and other areas, aiming to be among the first to integrate AI large models into consumer devices for a more personalized experience.
Apple CEO Tim Cook also acknowledged during a recent earnings call that Apple is investing in AI and large models. "We view AI and machine learning as foundational technologies embedded in nearly every product we make. Starting this fall, the iPhone will feature real-time voicemail transcription, and we will continue to invest and innovate," he said. Meanwhile, OPPO and vivo have also established dedicated AI research teams to explore large model applications.
The integration of AI large models has even extended upstream to chips and components. At the 2023 World AI Conference, Qualcomm demonstrated the world's first on-device generative AI model, Stable Diffusion, and the fastest on-device language-vision model (LVM), ControlNet. These models, with 1-1.5 billion parameters, can complete inferences in seconds, generating new AI images from text or image inputs. Qualcomm also revealed plans to support generative AI models with up to 10 billion parameters on smartphones this year.
MediaTek announced that its next-generation flagship SoC, the Dimensity 9300, launching later this year, will integrate an upgraded APU for enhanced on-device AI capabilities, offering experiences akin to ChatGPT. "Generative AI will drive hardware upgrades across the supply chain. We are already discussing solutions with clients for embedding AI large models," said a senior executive at a listed consumer electronics company in China.
Development Directions Still Under Exploration
Although embedding AI large models in smartphones is currently a trend, introducing AI capabilities into consumer electronics like smartphones is not a new phenomenon. Mainstream smartphone brands already incorporate AI-related technologies to varying degrees. For example, the commonly used photography feature relies heavily on AI algorithms. Additionally, before the rise of AI large models, most smartphone manufacturers had their own AI voice assistants, such as Apple's Siri, Huawei's Xiaoyi, Xiaomi's Xiaoai, Honor's YOYO, OPPO's Breeno, and vivo's Jovi. These assistants already possess certain AI capabilities, and with the trend of embedding large models, they are expected to undergo significant upgrades.
"AI technology has been applied in smartphones for many years. With the trend of embedding large models, smartphone companies are expected to deepen their existing AI capabilities while expanding their applications and scenarios. This will lead to more differentiated competition in the industry, revitalizing the stagnant smartphone market," said Ding Shaojiang, a senior industrial economist and founder of Ding Technology.
"Currently, neither smartphone manufacturers nor supply chains have a clear idea of how to use these large models on devices. There are two possibilities: one is to replace existing AI voice assistants with more intelligent, locally trained models that better understand users—Huawei is pursuing this direction. The other is applying AI in educational scenarios to enable personalized learning," said Guo Tianxiang, a senior analyst at IDC China.
Industry players have also adopted different approaches to deploying AI large models. Some, like Xiaomi, focus on on-device deployment, where all data, inference, training, and operations are handled locally. This ensures data privacy and reduces latency but increases hardware demands, potentially raising costs. Others, like Huawei, opt for cloud-based deployment, where processing occurs remotely, easing hardware requirements but raising concerns about data security and transmission delays. A hybrid approach, combining cloud and on-device models, is also emerging, as hinted by Honor's strategy.
Xiaomi's CEO Lei Jun emphasized their focus on lightweight, on-device models, diverging from internet companies' cloud-centric approaches. Huawei's consumer business CEO Yu Chengdong highlighted their commitment to privacy and security in AI deployments. Meanwhile, regulatory measures, such as China's new policies on AI and electronics consumption, aim to foster innovation while safeguarding user rights.
The integration of AI large models into smartphones not only aligns with technological trends but also serves as a catalyst for industry revival and innovation.
As is widely known, the global consumer electronics market, primarily driven by smartphones, has experienced continuous decline in recent years, with consumers extending their device replacement cycles. According to the latest research from Counterpoint, global smartphone sales in Q2 2023 decreased by 8% year-on-year and 5% quarter-on-quarter, marking the eighth consecutive quarter of annual decline, though the rate of contraction has slowed.
Financial reports from listed companies also indicate a gradual recovery in consumer electronics since the beginning of the year. For instance, Universal Scientific Industrial (Shanghai) Co., Ltd. reported H1 2023 revenue of RMB 26.87 billion, down 7.2% year-on-year, with consumer electronics revenue growing 3.2% to RMB 8.32 billion. Ofilm Group projected an H1 loss of RMB 320-400 million, significantly narrower than the RMB 874 million loss in the same period last year. Ofilm attributed this improvement to increased order volumes in Q2 and cost-cutting measures.
Sunlord Electronics noted on an investor platform that orders for mobile phones and consumer electronics have shown signs of slow recovery, with capacity utilization rates for communication and consumer products gradually improving both quarter-on-quarter and year-on-year. The company expects further recovery as the market warms up and new products are launched.
Universal Scientific Industrial stated during an institutional research session that revenue is expected to grow quarter-by-quarter in H2, with all business segments outperforming H1. Smartwatches, TWS earbuds, and VR/AR/MR devices are anticipated to drive over 30% cumulative growth in the consumer electronics sector over the next three years. Companies like Dongxu Optoelectronic Technology and Hongxin Electronics echoed similar optimism.
Counterpoint analysts predict that new smartphone launches in H2 will attract upgrades and shorten replacement cycles, leading to a slow market recovery. The traditional peak season for consumer electronics, highlighted by flagship releases from Huawei and Apple in September, combined with AI advancements, is expected to revitalize the industry.
TF Securities research team emphasized that technological advancements, particularly AI integration, will redefine user experiences across smartphones, smart speakers, and IoT devices. AI is projected to enhance hardware value, spur software innovation, and accelerate industry renewal.
Industrial Fulian Technology cited IDC forecasts in its interim report, predicting that generative AI will reshape mobile and smart living, with global smartphone and wearable shipments rebounding to 1.862 billion units by 2025.
However, some industry experts remain cautious. Analyst Guo Tianxiang noted that July retail data still showed year-on-year declines, indicating persistent weak consumer demand for smartphones.