In 2025, China’s A-share market surged like ocean tides, revealing deep-seated momentum amid its peaks and troughs. Total trading volume on the A-share market surpassed RMB 405 trillion for the first time in history, pushing market activity to a new high. The Shanghai Composite Index showed a structurally upward trend throughout the year, while the STAR Market stood out particularly brightly—the STAR Composite Index briefly led all major indices, posting an annual gain of approximately 22%, underscoring the dominant role of innovation-driven growth in the market. The ChiNext Board and growth-oriented sectors continued to attract capital inflows, becoming one of the core drivers of investor sentiment for the year.

Whether measured by index movements or capital flows, 2025 continuously redrew the A-share landscape—growth driven by technological innovation and structural opportunities has become the market’s guiding beacon.

Looking ahead to 2026, A-share investment logic is shifting from “hype” to “depth.” AI hardware (chips, computing platforms, infrastructure) and AI software (intelligent applications, enterprise systems, algorithm services) will form more solid growth trajectories as their fundamentals and profitability models become increasingly clear. Meanwhile, cross-industry digital transformation and industrial upgrading will offer investors fresh perspectives and opportunities.

The next chapter of the market will not merely be a race among sectors—it will be a test of value creation and innovation reshaping.

At the end of 2025, the consumer electronics market was ignited by a new terminal device called the “Doubao Phone.” This AI-native smartphone delivered a disruptive user experience by “breaking down app silos,” delivering a “small shock” to the long-stagnant internet industry: users no longer needed to switch repeatedly between apps from Tencent, Alibaba, and others. Instead, they could complete multi-platform tasks—such as shopping, social sharing, and financial payments—simply through natural voice commands.

The breakout success of the Doubao Phone was no accident. It acts like a mirror, clearly reflecting the market’s urgent demand for “AI-native hardware.” Three years after ChatGPT sparked a global AI boom, the industry has finally realized: while computing power is the foundation of AI development, terminal applications are the key to unlocking its true value.

In 2026, with deeper strategic commitments from major tech firms and maturing technologies, AI hardware terminals will officially move beyond the “concept-testing” phase and enter a period of large-scale growth and market breakthrough.

AI Terminals Enter a High-Growth Phase

Since the release of ChatGPT 3.0 in November 2022, global capital markets have engaged in a three-year frenzy centered on AI computing power, with upstream infrastructure—such as GPUs and data centers—drawing intense focus. However, a growing consensus is emerging: computing power detached from real-world applications remains a “castle in the air.” The ultimate value of AI must reach end consumers through terminal devices to truly take root.

Underlying this shift is a fundamental transformation in human-computer interaction. In the traditional internet era, users had to actively open apps and manually input commands to trigger functions—the device was merely a “passive execution tool.” In the AI Agent era, however, terminals are endowed with “proactive service capabilities.” Integrated with multimodal large models, these devices can listen, see, understand user intent, and even anticipate needs based on user habits—becoming genuine “personal AI assistants.”

This paradigm shift has drawn tech giants into the arena en masse. ByteDance’s Doubao Phone, Alibaba’s Quark AI Glasses, Li Auto’s Livis AI Glasses—though diverse in form, all aim for the same goal: capturing the core traffic gateway of the AI era. Just as smartphones replaced feature phones to become the carriers of mobile internet, AI hardware terminals are now emerging as the central hubs of the next-generation intelligent ecosystem.

Data confirms the growth potential of this sector. According to Frost & Sullivan, the global AI edge hardware market is projected to surge from RMB 321.9 billion in 2025 to RMB 1.22 trillion by 2029, with a compound annual growth rate (CAGR) of 40%—far outpacing traditional consumer electronics categories. Among these, AI smart glasses, as the most mature form factor today, are entering a phase of explosive growth.

Research from WellsennXR shows that global AI smart glasses shipments reached 1.55 million units in 2024, largely driven by Ray-Ban Meta smart glasses. In 2025, that figure skyrocketed to 7 million units—a year-over-year increase of over 350%. In 2026, high growth is expected to continue, with shipments projected to exceed 15 million units. China’s performance has been even more impressive: according to IDC, Chinese smart glasses vendors shipped over 1 million units in the first half of 2025, a 64.2% year-over-year increase, capturing 26.6% of the global market share. Looking further ahead, global smart glasses shipments are expected to surpass 40 million units by 2029, with China’s five-year CAGR forecast at 55.6%—the highest globally—making it the core engine of global AI terminal development.

The boom in AI hardware will directly fuel prosperity across the upstream supply chain. Take the Oakley Meta HSTN AI glasses as an example: its bill of materials (BOM) costs approximately $165.47, with core components including SoC chips, power management ICs, cameras, sensors, and lithium batteries. Chips account for nearly 50% of total costs, making them the linchpin of the supply chain. As terminal sales continue to grow rapidly and market ceilings rise, upstream component suppliers are entering a golden cycle of revenue expansion.

SoC Chips: The Compute Foundation for On-Device AI

If large AI models are the “brain” of the terminal, then SoC chips are the “computing heart” that powers it. A notable trend among AI glasses launched by major Chinese firms in 2025 is that flagship products—including Quark AI Glasses, Xiaomi AI Glasses, and Li Auto’s Livis—all selected BES2800 from BES Technology (SHA: 688608) as their main controller chip. This “collective choice” stems from BES’s clear technological edge.

For wearable devices like AI glasses, battery life is the top user pain point. Unlike Bluetooth earbuds, AI glasses are worn throughout the day (except during sleep), demanding far stricter low-power performance. The most direct and effective way to reduce chip power consumption is through advanced manufacturing processes.

BES’s BES2800 series precisely addresses this need: built on a 6nm FinFET process, it achieves a qualitative leap in power efficiency compared to domestic competitors still using 22nm processes. Currently, BES’s domestic rivals in wearable chips—such as Zhuhai Jielix, Bluetrum (SHA: 688332), and Actions Semiconductor (SHA: 688049)—remain stuck at the 22nm node for AI glasses solutions, creating a clear generational gap with BES.

Even more promising is BES’s next-generation chip, the BES6000, which has already entered the sampling phase and is expected to enter mass production in the first half of 2026. Industry sources indicate that the BES6000 delivers a quantum leap in performance over the BES2800—with NPU compute power alone increasing tenfold. This upgrade will resolve the current compute bottleneck in AI glasses: previously, due to insufficient BES2800 performance and lack of an integrated ISP, major brands commonly adopted a dual-chip solution pairing Qualcomm’s Snapdragon AR1 with a BES chip as a co-processor, requiring an external third-party ISP for camera functionality. The BES6000 integrates an ISP module, enabling high-definition imaging without external components—reducing system complexity and lowering terminal costs.

Strategically, the BES6000 marks BES’s “transformation milestone.” ARM’s IP cores fall into three categories: A-series for high-performance applications like smartphones, M-series for ultra-low-power devices like earbuds, and R-series for high-reliability uses like aerospace. Historically, BES focused exclusively on M-series cores for the Bluetooth audio market. The BES6000 is the company’s first A-series-based chip, signaling its official transition from a “Bluetooth audio chip vendor” to a “smart terminal chip provider.” This pivot unlocks vastly broader market opportunities—the combined global markets for smart glasses and smartwatches far exceed that of Bluetooth earbuds.

This technological advantage has already translated into strong financial performance. In the first three quarters of 2025, BES reported an 18.61% year-over-year revenue increase, nearly doubled adjusted net profit, and improved gross margin by nearly 5 percentage points. With major brands launching more mature AI glasses in 2026, economies of scale will further reduce costs, leaving ample room for continued earnings growth.

Acoustic Components: The “Key” to AI Interaction—Multi-Mic Arrays Become Standard

One core difference between AI terminals and smartphones lies in interaction methods: voice replaces touchscreens as the primary human-computer interface. This shift elevates acoustic components to unprecedented importance—only by accurately capturing voice commands can the embedded AI model “understand user needs,” making the terminal’s intelligence truly functional.

To deliver exceptional voice interaction, leading brands are heavily investing in acoustic hardware. The Quark AI Glasses, for instance, feature a premium configuration of five high-performance microphones plus one bone conduction microphone (VPU). This multi-mic array uses beamforming technology to precisely locate the user’s voice, effectively filtering ambient noise—even whispered commands can reliably wake the assistant while significantly reducing false triggers.

Compared to smartphones, AI glasses represent a “step-function” improvement in microphone deployment. Most smartphones have only 2–4 microphones: entry-level models typically include two, mid-to-high-end models three, and only flagship devices like the Xiaomi 17 or OPPO Find X9 offer four. Given AI glasses’ greater reliance on voice interaction, increased microphone counts are becoming an industry standard, driving explosive demand for acoustic components.

In this segment, AAC Technologies (HKEX: 2018), China’s acoustic component leader, has secured a first-mover advantage. The Quark AI Glasses use AAC’s microphone solution, which has become a core supplier to major brands thanks to its “high recognition accuracy + low false-trigger rate.” Moreover, AAC is not limited to acoustics—it also supplies optical modules, speakers, thermal solutions, and actuators, forming a comprehensive consumer electronics component portfolio.

To deepen its commitment to the AI glasses market, AAC recently acquired Dispelix, a leader in AR diffractive waveguide technology. Diffractive waveguides are critical display components for AR glasses, and this acquisition marks AAC’s evolution from a “component supplier” to a “full-system solution provider,” further solidifying its position in the AI glasses supply chain.

Strong technical capabilities and strategic positioning have translated into robust financial results. In the first half of 2025, AAC reported an 18.42% year-over-year revenue increase and a 63.06% jump in net profit, demonstrating high-growth resilience. Notably, as of year-end 2025, the company traded at a P/E ratio below 20x—still at a historical valuation low—leaving significant room for re-rating as the AI terminal market expands.

2025 marked the “inaugural year” for Chinese tech giants entering the AI hardware terminal space, with products largely serving as trials. In 2026, with maturing technology, falling costs, and ecosystem improvements, AI hardware terminals will enter a “full-scale breakthrough” phase. Beyond smart glasses achieving mass adoption, multiple product categories—including AI phones, AI watches, and AI in-car terminals—will collectively drive a “blossoming” market landscape. Technologically, advances in chip compute power, optimized acoustic interaction, and multimodal integration will further refine user experiences. Geographically, China—as the world’s largest consumer electronics market—will serve as the core growth engine for AI hardware terminals, simultaneously elevating the global competitiveness of its upstream supply chain.

In 2025, China’s A-share market surged like ocean tides, revealing deep-seated momentum amid its peaks and troughs. Total trading volume on the A-share market surpassed RMB 405 trillion for the first time in history, pushing market activity to a new high. The Shanghai Composite Index showed a structurally upward trend throughout the year, while the STAR Market stood out particularly brightly—the STAR Composite Index briefly led all major indices, posting an annual gain of approximately 22%, underscoring the dominant role of innovation-driven growth in the market. The ChiNext Board and growth-oriented sectors continued to attract capital inflows, becoming one of the core drivers of investor sentiment for the year.

Whether measured by index movements or capital flows, 2025 continuously redrew the A-share landscape—growth driven by technological innovation and structural opportunities has become the market’s guiding beacon.

Looking ahead to 2026, A-share investment logic is shifting from “hype” to “depth.” AI hardware (chips, computing platforms, infrastructure) and AI software (intelligent applications, enterprise systems, algorithm services) will form more solid growth trajectories as their fundamentals and profitability models become increasingly clear. Meanwhile, cross-industry digital transformation and industrial upgrading will offer investors fresh perspectives and opportunities.

The next chapter of the market will not merely be a race among sectors—it will be a test of value creation and innovation reshaping.

At the end of 2025, the consumer electronics market was ignited by a new terminal device called the “Doubao Phone.” This AI-native smartphone delivered a disruptive user experience by “breaking down app silos,” delivering a “small shock” to the long-stagnant internet industry: users no longer needed to switch repeatedly between apps from Tencent, Alibaba, and others. Instead, they could complete multi-platform tasks—such as shopping, social sharing, and financial payments—simply through natural voice commands.

The breakout success of the Doubao Phone was no accident. It acts like a mirror, clearly reflecting the market’s urgent demand for “AI-native hardware.” Three years after ChatGPT sparked a global AI boom, the industry has finally realized: while computing power is the foundation of AI development, terminal applications are the key to unlocking its true value.

In 2026, with deeper strategic commitments from major tech firms and maturing technologies, AI hardware terminals will officially move beyond the “concept-testing” phase and enter a period of large-scale growth and market breakthrough.

AI Terminals Enter a High-Growth Phase

Since the release of ChatGPT 3.0 in November 2022, global capital markets have engaged in a three-year frenzy centered on AI computing power, with upstream infrastructure—such as GPUs and data centers—drawing intense focus. However, a growing consensus is emerging: computing power detached from real-world applications remains a “castle in the air.” The ultimate value of AI must reach end consumers through terminal devices to truly take root.

Underlying this shift is a fundamental transformation in human-computer interaction. In the traditional internet era, users had to actively open apps and manually input commands to trigger functions—the device was merely a “passive execution tool.” In the AI Agent era, however, terminals are endowed with “proactive service capabilities.” Integrated with multimodal large models, these devices can listen, see, understand user intent, and even anticipate needs based on user habits—becoming genuine “personal AI assistants.”

This paradigm shift has drawn tech giants into the arena en masse. ByteDance’s Doubao Phone, Alibaba’s Quark AI Glasses, Li Auto’s Livis AI Glasses—though diverse in form, all aim for the same goal: capturing the core traffic gateway of the AI era. Just as smartphones replaced feature phones to become the carriers of mobile internet, AI hardware terminals are now emerging as the central hubs of the next-generation intelligent ecosystem.

Data confirms the growth potential of this sector. According to Frost & Sullivan, the global AI edge hardware market is projected to surge from RMB 321.9 billion in 2025 to RMB 1.22 trillion by 2029, with a compound annual growth rate (CAGR) of 40%—far outpacing traditional consumer electronics categories. Among these, AI smart glasses, as the most mature form factor today, are entering a phase of explosive growth.

Research from WellsennXR shows that global AI smart glasses shipments reached 1.55 million units in 2024, largely driven by Ray-Ban Meta smart glasses. In 2025, that figure skyrocketed to 7 million units—a year-over-year increase of over 350%. In 2026, high growth is expected to continue, with shipments projected to exceed 15 million units. China’s performance has been even more impressive: according to IDC, Chinese smart glasses vendors shipped over 1 million units in the first half of 2025, a 64.2% year-over-year increase, capturing 26.6% of the global market share. Looking further ahead, global smart glasses shipments are expected to surpass 40 million units by 2029, with China’s five-year CAGR forecast at 55.6%—the highest globally—making it the core engine of global AI terminal development.

The boom in AI hardware will directly fuel prosperity across the upstream supply chain. Take the Oakley Meta HSTN AI glasses as an example: its bill of materials (BOM) costs approximately $165.47, with core components including SoC chips, power management ICs, cameras, sensors, and lithium batteries. Chips account for nearly 50% of total costs, making them the linchpin of the supply chain. As terminal sales continue to grow rapidly and market ceilings rise, upstream component suppliers are entering a golden cycle of revenue expansion.

SoC Chips: The Compute Foundation for On-Device AI

If large AI models are the “brain” of the terminal, then SoC chips are the “computing heart” that powers it. A notable trend among AI glasses launched by major Chinese firms in 2025 is that flagship products—including Quark AI Glasses, Xiaomi AI Glasses, and Li Auto’s Livis—all selected BES2800 from BES Technology (SHA: 688608) as their main controller chip. This “collective choice” stems from BES’s clear technological edge.

For wearable devices like AI glasses, battery life is the top user pain point. Unlike Bluetooth earbuds, AI glasses are worn throughout the day (except during sleep), demanding far stricter low-power performance. The most direct and effective way to reduce chip power consumption is through advanced manufacturing processes.

BES’s BES2800 series precisely addresses this need: built on a 6nm FinFET process, it achieves a qualitative leap in power efficiency compared to domestic competitors still using 22nm processes. Currently, BES’s domestic rivals in wearable chips—such as Zhuhai Jielix, Bluetrum (SHA: 688332), and Actions Semiconductor (SHA: 688049)—remain stuck at the 22nm node for AI glasses solutions, creating a clear generational gap with BES.

Even more promising is BES’s next-generation chip, the BES6000, which has already entered the sampling phase and is expected to enter mass production in the first half of 2026. Industry sources indicate that the BES6000 delivers a quantum leap in performance over the BES2800—with NPU compute power alone increasing tenfold. This upgrade will resolve the current compute bottleneck in AI glasses: previously, due to insufficient BES2800 performance and lack of an integrated ISP, major brands commonly adopted a dual-chip solution pairing Qualcomm’s Snapdragon AR1 with a BES chip as a co-processor, requiring an external third-party ISP for camera functionality. The BES6000 integrates an ISP module, enabling high-definition imaging without external components—reducing system complexity and lowering terminal costs.

Strategically, the BES6000 marks BES’s “transformation milestone.” ARM’s IP cores fall into three categories: A-series for high-performance applications like smartphones, M-series for ultra-low-power devices like earbuds, and R-series for high-reliability uses like aerospace. Historically, BES focused exclusively on M-series cores for the Bluetooth audio market. The BES6000 is the company’s first A-series-based chip, signaling its official transition from a “Bluetooth audio chip vendor” to a “smart terminal chip provider.” This pivot unlocks vastly broader market opportunities—the combined global markets for smart glasses and smartwatches far exceed that of Bluetooth earbuds.

This technological advantage has already translated into strong financial performance. In the first three quarters of 2025, BES reported an 18.61% year-over-year revenue increase, nearly doubled adjusted net profit, and improved gross margin by nearly 5 percentage points. With major brands launching more mature AI glasses in 2026, economies of scale will further reduce costs, leaving ample room for continued earnings growth.

Acoustic Components: The “Key” to AI Interaction—Multi-Mic Arrays Become Standard

One core difference between AI terminals and smartphones lies in interaction methods: voice replaces touchscreens as the primary human-computer interface. This shift elevates acoustic components to unprecedented importance—only by accurately capturing voice commands can the embedded AI model “understand user needs,” making the terminal’s intelligence truly functional.

To deliver exceptional voice interaction, leading brands are heavily investing in acoustic hardware. The Quark AI Glasses, for instance, feature a premium configuration of five high-performance microphones plus one bone conduction microphone (VPU). This multi-mic array uses beamforming technology to precisely locate the user’s voice, effectively filtering ambient noise—even whispered commands can reliably wake the assistant while significantly reducing false triggers.

Compared to smartphones, AI glasses represent a “step-function” improvement in microphone deployment. Most smartphones have only 2–4 microphones: entry-level models typically include two, mid-to-high-end models three, and only flagship devices like the Xiaomi 17 or OPPO Find X9 offer four. Given AI glasses’ greater reliance on voice interaction, increased microphone counts are becoming an industry standard, driving explosive demand for acoustic components.

In this segment, AAC Technologies (HKEX: 2018), China’s acoustic component leader, has secured a first-mover advantage. The Quark AI Glasses use AAC’s microphone solution, which has become a core supplier to major brands thanks to its “high recognition accuracy + low false-trigger rate.” Moreover, AAC is not limited to acoustics—it also supplies optical modules, speakers, thermal solutions, and actuators, forming a comprehensive consumer electronics component portfolio.

To deepen its commitment to the AI glasses market, AAC recently acquired Dispelix, a leader in AR diffractive waveguide technology. Diffractive waveguides are critical display components for AR glasses, and this acquisition marks AAC’s evolution from a “component supplier” to a “full-system solution provider,” further solidifying its position in the AI glasses supply chain.

Strong technical capabilities and strategic positioning have translated into robust financial results. In the first half of 2025, AAC reported an 18.42% year-over-year revenue increase and a 63.06% jump in net profit, demonstrating high-growth resilience. Notably, as of year-end 2025, the company traded at a P/E ratio below 20x—still at a historical valuation low—leaving significant room for re-rating as the AI terminal market expands.

2025 marked the “inaugural year” for Chinese tech giants entering the AI hardware terminal space, with products largely serving as trials. In 2026, with maturing technology, falling costs, and ecosystem improvements, AI hardware terminals will enter a “full-scale breakthrough” phase. Beyond smart glasses achieving mass adoption, multiple product categories—including AI phones, AI watches, and AI in-car terminals—will collectively drive a “blossoming” market landscape. Technologically, advances in chip compute power, optimized acoustic interaction, and multimodal integration will further refine user experiences. Geographically, China—as the world’s largest consumer electronics market—will serve as the core growth engine for AI hardware terminals, simultaneously elevating the global competitiveness of its upstream supply chain.