The first quarter of 2024 marked a significant turning point for various intelligent Internet of Things (IoT) chip companiesAs advancements in Artificial Intelligence (AI) have been integrated into industrial applications, System on Chip (SoC) solutions have emerged as pivotal components, functioning as the "brains" of smart terminal devicesThis shift has facilitated the transition into the era of intelligent IoT, enhancing efficiencies and enabling smart interactions integral to modern consumer experiencesAn accelerated drive in the smart IoT landscape is henceforth expected to support the cyclical prosperity of semiconductor manufacturing sectors.

Conceived from the integration of AI and IoT technologies, termed Artificial Intelligence of Things (AIoT), the latter half of 2023 witnessed recovery for chip manufacturers involved in AIoT

Thanks to the progress in high-performance SoCs, Wi-Fi chips, and high-definition video decoding chips, many companies have reported significant performance rebounds as of Q1 2024. Some publicly listed firms are on the verge of recuperating their revenues to historical high levels, indicating a resurgence in industry sentiment.

Generative AI systems exemplified by ChatGPT have revolutionized interactions within smart homes, paving the way for enhanced functionalities in smart home products such as smart speakers and wearablesThis transformative technology vastly improves user feedback in terms of both richness and accuracy, allowing for a more profound understanding of user intent and increasing consumer satisfactionProjections suggest that the Chinese AIoT SoC chip market could reach ¥279.4 billion by 2027.

The chip industry faced challenges in 2022, particularly within the consumer electronics market, where a decline in demand pressured profitability across relevant chip products

• The Japanese Stock Market May Be Overbought
• Can the U.S. Treasury Expect a Turnaround?
• Slight Growth in the U.S. Economy
• The Definitive Thread of Economic Prosperity
• Will Gold Continue Its Bull Run Next Year?

Consequently, while the industry found itself in a cyclical downturn, design companies opted for ongoing investment in research and development to drive the creation of next-generation chipsAIoT SoC chips, serving as the central processing units in AIoT devices, integrate core computational components including CPU, GPU, neural network processor, video/audio codecs, display controllers, and memory systemsThese multifunctional modules cater not only to the demands of consumer electronics but also enable advanced AI processing, audio-visual handling, and sophisticated voice recognition capabilities.

Starting from the latter part of 2023, with the clearing of excess inventory and the rollout of high-performance chips in commercial applications, AIoT SoCs have begun to recoverFor instance, Lexin Technology reported a 12.7% increase in operating income year-over-year for 2023, with a striking 63.6% rise in net profits after deducting non-recurring items

Today, the global shipment of IoT chips has surpassed one billion units, while the company has expanded its portfolio from Wi-Fi MCUs into the AIoT SoC arenaSimilarly, Hengxuan Technology experienced a substantial 46.6% growth in revenue on a year-over-year basis, with non-recurring net profit up an impressive 136%. Allwinner Technology also marked a 12.4% year-over-year achievement in revenue from smart terminal application processors.

The local deployment of AI models at the edge and terminal levels is becoming increasingly evidentFor instance, Rockchip saw significant growth in automotive electronics, industrial applications, and machine vision sectors in Q1 2024, with revenue growth estimated between 64% and 67%, leading to a net profit forecast of between ¥60-70 millionIn contrast, the company faced profitability struggles from 2021 to 2023, achieving net profits of ¥110 million, ¥84.15 million, and a loss of ¥1.838 million in respective quarters

Additionally, Amlogic generated operating revenues of ¥1.38 billion in Q1 2024, reflecting a year-on-year rise of 33.1%, with only a slight difference from its historic revenue high of ¥1.48 billion in Q1 2022.

The features of SoCs, including high integration, low power consumption, minimal heat generation, and robust compatibility, allow for extensive format compatibility particularly in video codec and visual processingSpecifically, the market for SoCs in wearables like TWS earbuds and smartwatches is characterized by numerous smaller, fragmented customer segments that larger chipmakers may find less appealing due to their focus on vast single-category markets like smartphone SoCs, PC CPUs, or GPUsAs such, the priority for SoC applications lies in the benefits gained from extended battery life, creating room for domestic firms to thrive.

In terms of video technology, the rise of 8K-capable smart boxes has started in consumer markets, with standards for ultra-high-definition videos being implemented as of November 2023. The demand for 8K displays in sectors such as healthcare, security, industrial visualization, and smart traffic is increasing, facilitated by improvements in corresponding technologies and solutions

The industry chain is propelling toward maturity, evidenced by a rising shipment of related chips.

Wi-Fi technology serves as the cornerstone for the development of smart cities and homes, previously constrained to 2.4GHz and 5GHz frequency bandsHowever, the rise of ultra-high-throughput and low-latency applications now compels continuous evolution toward faster speeds, lower latencies, and higher data transfer quality across industrial IoT marketsGiven the competitive landscape, significant IoT device manufacturers converge on similar market shares and product strategies, resulting in substantial price competition where domestic manufacturers leverage cost advantages to address price reductions born from ongoing AIoT penetration.

The innovative development of smart homes is evident in the evolution of robotic vacuum cleaners, wearables, smart speakers, and smart TVs

In 2023, the vacuum cleaner market swiftly recovered, with domestic brands successfully mass-producing independently navigating products equipped with AI object recognition and avoidance featuresAddressing this trend, domestic chip manufacturers have designed a new generation of high-performance octa-core AI chips and solutions tailored for mid-to-high-end vacuum cleaners.

Wearable device manufacturers are increasingly implementing generative AI technologiesSmartwatches, for instance, can employ deep learning and natural language processing to better understand user language and behaviors, yielding personalized suggestions and servicesSimilarly, advanced AI algorithms are requisite for developing sophisticated products such as robotic lawnmowers, biomimetic robots, and delivery pods, necessitating continuous development in sensor technologies like laser, vision, ITOF, and stereo camera technologies from chip companies.

The surge in sales of smart speakers was primarily seen in 2018 and 2019. However, the slow pace of intelligent upgrades led to severe homogenization issues, hampering cross-device connectivity and adversely affecting user experience

As noted by Guotai Junan Securities, domestic smart speaker sales peaked at 37.85 million units in 2020, but began a rapid decline from 2022, experiencing a 28% year-over-year drop in that yearTraditional domestic IoT markets thus face significant growth bottlenecks, prompting leading manufacturers to incorporate large models into their smart speakers, which may stimulate new demandProjections indicate smart speaker annual sales will stabilize around 20 million units from 2024 onward.

Presently, giants like Baidu, Xiaomi, and Tmall Genie have developed extensive ecosystems over several yearsThe multi-modal capabilities exhibited by models like GPT-4 are anticipated to significantly enhance interactivity, boosting voice and video functionalities in smart speakersDomestic chips are now capable of customizing scenario-based smart assistant experiences, illustrated by the Tmall Genie SOUND Series connecting seamlessly to large model terminal operating systems.

Smart TVs are at the nascent stage of a profound transformation driven by AI technology, necessitating enhancements in CPU capability within SoCs and the incorporation of dedicated AI cores and DSPs

There is also an increased demand for extensive audio-visual expansion interfaces and a reduction in power consumption to enhance battery lifeEnvironmental intelligence will emerge as a pivotal area for the upgrade of smart home platformsBy consolidating information related to air quality, light exposure, and user activity, smart home systems will gradually evolve to perceive and anticipate user needs, facilitating more intuitive human-computer interaction and personalized services.

The growing momentum around edge AI heralds a new era, amplified by RISC-V, which has sparked interest in open-source hardware and chip designThe openness of computational architecture fosters innovations in hardware, while the Hongmeng (Harmony) operating system, noted for its cross-device and distributed characteristics, permits users to form cluster computing capabilities without a cloud intermediary

This shift significantly eases the demands placed on edge hardware and its computing power.

In this contemporary age characterized by the fusion of cloud computing and IoT, AI chips deployed on cloud and edge computing servers, exemplified by GPUs, enjoy robust versatilityIn contrast, terminal AI chips typically interlink AI functionalities with traditional CPU, GPU, and ISP modules into an SoC format, enabling custom designs tailored to specific applications like voice and facial recognition while striking a balance between performance and power consumptionThe ever-present proliferation of IoT devices presents a primary venue for the application of edge AI, with AI SoCs poised to grow in tandem with the intelligent evolution of IoT devices.

As it stands, the global market for edge AI chips was under $40 billion in 2022, with expectations to exceed $60 billion by 2025. In China, the domestic market is forecasted to rise from $5 billion in 2022 to $11 billion by 2025, corresponding to an approximate 30% annual growth rate

Furthermore, estimates for 2023 indicate global edge AI chip shipments will reach 1.31 billion units, which encompasses around 190 million for voice processing, 900 million for machine vision, and 140 million for sensor data analysis, with projections suggesting a dramatic surge to 2.29 billion units by 2028.

In a world increasingly interconnected where user demands diversify, the integration of expansive models and generative AI is confronting a rapid transformation of content and service deliveryThe era of AIoT features rapid updates in smart hardware technology and shortened product life cycles supporting continuous upgrades of AIoT chips, with functionalities evolving swiftlyAs the number of IoT devices continues to surge and AI technologies are embraced at a wide scale, the market for AIoT chips is anticipated to persistently expand.