In this issue, the EETimes analyst team provides in-depth trend analysis and market outlook on key topics including storage chips, advanced packaging, embodied intelligence, custom chips, optical interconnection, SiC/GaN power devices, RISC-V, solid-state batteries, AI Agent, and smart manufacturing.

In 2026, artificial intelligence (AI) will further accelerate penetration across industries, driving profound changes in industrial structure. Multiple sectors will be directly or indirectly driven by AI, ushering in new development opportunities in the new year.

Trend 1: AI Drives "Structural Shortage" of Storage Chips

Driven by surging AI demand, capacity restructuring, and accelerated technological iteration, the global storage chip market will remain in a phase of "structural shortage" in 2026. Currently, AI technology is evolving from the cloud to the edge, and with the commercialization of AI models across industries, edge AI has achieved initial deployment. This context requires more storage to meet the demands of both cloud and edge computing.

High Bandwidth Memory (HBM) is a core component of AI servers, primarily deployed near GPUs/accelerators to support high-performance computing by breaking through memory bandwidth bottlenecks. Due to its 3D stacking technology that integrates multiple DRAM chips for higher bandwidth and capacity, HBM consumes over 3 times the wafer capacity of standard DRAM. DDR5, with its advantages of high bandwidth, large capacity, and energy efficiency, is particularly suitable for large-scale model training and high-throughput tasks. However, DDR5 modules have higher costs than DDR4 due to stricter requirements on process, integration, error correction mechanisms, and electrical performance.

Amid sustained AI demand growth, major storage vendors are prioritizing capacity allocation to HBM and DDR5, leading to tight supply of traditional products like DDR4. Additionally, starting from 2024, storage giants such as Samsung, SK Hynix, Micron, and Changxin Memory have reduced DDR4 production, planning to phase out DDR4 supply between late 2025 and mid-2026. Although Samsung and SK Hynix later extended the lifecycle of DDR4 production lines to the end of 2026, demand resilience exceeded expectations—some equipment manufacturers remain highly dependent on DDR4, have low cost sensitivity, and lack motivation to migrate to DDR5. As a result, DDR4 shortages are expected to persist at least until mid-2026.

NAND flash is also facing supply constraints. NAND prices have continued to decline since Q3 2024, prompting flash giants to cut production in Q2 2025 to boost prices. Meanwhile, strong demand for AI server deployment and general-purpose server expansion by North American cloud service providers has led international storage vendors to shift more NAND capacity to enterprise-grade SSDs (high-capacity QLC), further exacerbating supply tightness. NAND prices are projected to rise further in 2026, with enterprise-grade SSD prices continuing their upward trend.

Currently, the market supply-demand structure is undergoing profound changes, with the storage chip industry exhibiting "structural shortage" characteristics: on one hand, high-end products are in short supply due to limited advanced process capacity, driving price increases; on the other hand, mid-to-low-end products face capacity mismatches, with surplus mature process capacity failing to meet actual demand, leading to reduced resource utilization efficiency. Overall, the misalignment between supply and demand structures will be the core driver of this round of price increases.

Trend 2: Advanced Packaging Market Expands Significantly, Standards Converge

Narrowing "advanced packaging" to 2.5D/3D packaging, the market size is expected to grow substantially in 2026. In fact, 2.5D/3D advanced packaging solutions have already been popularized in consumer electronics over the past two years—for example, AMD Ryzen processors use 3D V-Cache, Apple M-Series Ultra processors "stitch" two Max chips, and Intel Core Ultra processors adopt modular design. Unsurprisingly, the technology is widely used in data center HPC and AI chips.

Two landmark events highlight the expanded application scope of advanced packaging in the 2.5D/3D space:

1. Technology Innovation: Intel’s upcoming Xeon 6+ processors, scheduled for launch in H2 2026, will debut Intel’s own 3D hybrid bonding technology. While AMD has already adopted TSMC’s similar technology in Epyc processors, Intel’s large installed base in the data center server market is expected to drive widespread adoption of this cutting-edge packaging technology.
2. Standardization & Ecosystem: As of 2025, major chip design and manufacturing giants—including Intel, AMD, TSMC, Samsung, Arm, Google, and Microsoft—have endorsed the UCIe standard, aiming to achieve Chiplet interoperability similar to PCIe’s role in board-level systems. UCIe 2.0 and later versions are expected to be widely adopted in 2026, focusing on HPC, AI accelerators, and more data center chips. Regardless of UCIe, the standardized Chiplet ecosystem will converge in 2026—a consensus among mainstream enterprises across cost, time-to-market, and other commercial dimensions.

Notably, TSMC’s Foundry 2.0 strategy (proposed in recent years) positions the company to become the world’s largest packaging service provider after 2025, with its advanced packaging capacity expected to reach 10 times the 2023 level by 2026. Intel Foundry also announced the full launch of its OSAT model in 2025, offering backend packaging services including EMIB and Foveros to meet market capacity needs.

In terms of market potential, advanced packaging is expanding its technical and application scope. Technically, this includes hybrid bonding for HBM, Co-Packaged Optics (CPO), panel-level packaging, and more forward-looking glass substrates and interposers. Broadening the definition beyond cutting-edge technologies requiring advanced manufacturing processes and ultra-fine bonding/pitch, advanced packaging also presents opportunities in 5G/6G communications (antenna and RF front-end packaging) and automotive chips (sensor-computing unit integration) beyond traditional data center, PC, and mobile applications.

Trend 3: Embodied Intelligence: Scene Deployment Drives Trillion-Yuan Market Explosion

Shifting from "disembodied thinking" to "embodied action," embodied intelligence is ushering in a new cycle for the AI industry. As a key future industry prioritized in the 15th Five-Year Plan, its market size is expected to exceed 1 trillion yuan for the first time in 2026, with a CAGR of approximately 25% over the next five years—driven primarily by deep penetration of application scenarios.

The industrial sector is poised to become the first major battlefield for large-scale deployment of embodied intelligence. Labor shortages and efficiency demands in manufacturing industries such as automotive and electronics are driving industrial robots to evolve from "mechanical execution" to "autonomous decision-making." Humanoid robots like Tesla Optimus G3, Ubtech Walker S, and Figure 02, equipped with multi-modal perception and environmental adaptability, can perform complex tasks such as precision assembly and bin picking, and have already begun trial operations in multiple automaker factories.

Specialized scenarios are achieving breakthrough applications, filling human operational blind spots. In high-risk fields such as energy exploration and disaster rescue, embodied intelligence devices have demonstrated irreplaceable value. Currently, 70% of explosion-proof robots adopt quadruped (dog-like) forms, and the German Aerospace Center is testing humanoid robots for space operations. These devices can withstand harsh conditions such as extreme temperatures and toxic environments, liberating humans from high-risk settings in scenarios like chemical inspections and earthquake search and rescue, and are becoming standard equipment in specialized industries.

Service and logistics scenarios are accelerating civilian penetration. Data from China Post shows that as of 2024, the cumulative deployment of unmanned delivery vehicles in express logistics exceeded 6,000 units. Currently, collaboration between humanoid robots and drones is addressing the "last 10 meters" delivery challenge.

B2B service scenarios such as mall delivery and store reception have taken the lead in deployment, with service robots from companies like Galaxy General already in operation. As costs decline, C2C scenarios such as elderly care and household assistance will gradually become widespread, alleviating the national pain point of a 5 million shortage in home caregivers.

It is foreseeable that over the next five years, embodied intelligence will follow the path of "industrial leadership, specialized breakthroughs, and service popularization," penetrating from high-end manufacturing to daily life, and reshaping the global intelligent industry pattern with the "Embodied China" development model.

Trend 4: Three Engines (AI/Automotive/Consumer Electronics) Drive Custom Chip Growth

Combined data from McKinsey and the Semiconductor Industry Association (SIA) shows that as early as 2024, the global semiconductor design services market (including general-purpose + custom) reached approximately 38–42 billion US dollars, with custom design services (primarily targeting AI, HPC, automotive, and IoT customers) accounting for 35%–40%, or 13–17 billion US dollars. A new trend is emerging: custom chips are shifting from an exclusive choice of tech giants to a rigid demand across industries.

AI and computing scenarios have become the largest incremental market for custom chips. The explosive growth of generative AI has driven exponential increases in computing power demand, and general-purpose GPUs struggle to match the efficiency requirements of specific models. In response, global tech giants have taken the lead in deployment—OpenAI collaborating with Broadcom to develop custom chips, Amazon advancing the deployment of Trainium2 chips, and Google making TPU available for external sales. Starting from 2026, the ASIC custom services market in this field will surge from 17.7 billion US dollars, as differentiated AI workload demands drive end-to-end custom solutions from training to inference. Combined with modular design based on Chiplet architecture, precise matching of computing power and power consumption can be achieved.

The automotive electronics sector is experiencing an inflection point for custom chip growth. Traditional general-purpose automotive chips cannot balance the low latency, automotive-grade safety, and multi-sensor fusion requirements of autonomous driving. With the upgrade of intelligent driving levels and the popularization of connected cars, the transition from standardized to custom automotive chips has become an inevitable trend. It is expected that from in-vehicle infotainment systems to LiDAR signal processing, custom chips will become a core tool for automakers to build differentiated competitiveness. The trend of regional manufacturing further enables automakers to iterate chip designs quickly to meet local market needs.

Personalized demands in consumer electronics and IoT are activating niche markets. The extreme pursuit of low power consumption in wearable health monitors and scenario-specific functional requirements in smart home devices are driving custom chips toward low-volume, high-precision development. With the deep popularization of 5G technology, edge computing scenarios place higher demands on local chip processing capabilities. Combined with data security regulations, custom security chips will become a new growth point—compliance requirements such as GDPR are forcing enterprises to adopt exclusive encryption chip solutions.

Overall, the global custom chip services market is developing rapidly, driven by both demand and supply sides: differentiated demands for chip efficiency and security in scenarios such as AI computing power outbreaks, automotive electronics intelligence, and consumer electronics personalization are becoming increasingly urgent. Meanwhile, technological breakthroughs such as Chiplet modular architecture and AI-assisted design, along with turnkey service models, have significantly lowered customization thresholds, jointly driving high-speed market growth.

Trend 5: Silicon Photonics Penetrates Computing: Optical Interconnection "Reshapes" AI Data Flow

Currently, silicon photonics technology is accelerating its strategic penetration from its traditional strength—data communications—to emerging frontiers: high-performance computing (HPC) and artificial intelligence (AI), to meet the efficient execution of model training and inference tasks driven by massive data. One of the most representative trends is the industry’s rapid shift from pluggable optical modules to a more integrated solution: Co-Packaged Optics (CPO). By co-packaging optical engines (including lasers, modulators, detectors, etc.) with computing chips such as switch ASICs, GPUs, and CPUs on the same substrate, this technology greatly shortens electrical signal transmission distance, significantly reduces power consumption and latency, and substantially increases bandwidth density.

Compared with traditional pluggable solutions, CPO technology’s core advantages lie in three dimensions: power consumption, bandwidth density, and latency. For example, Nvidia’s Spectrum X CPO switch, launched in March 2025, offers a total throughput of 400 Tbps (high-end configuration) with significantly reduced port power consumption, achieving a 3.5x improvement in overall energy efficiency, 10x higher network reliability, and 1.3x shorter deployment time. Data shows that as bandwidth evolves from 400G/800G to 1.6T, CPO’s power consumption advantage becomes increasingly prominent, far outperforming traditional pluggable modules. This partially addresses the challenges of "GPU computing power idling" and the substantial growth in AI data center energy consumption.

However, despite CPO’s potential advantages in increasing bandwidth, reducing power consumption, and simplifying system architecture, large-scale commercialization requires simultaneous breakthroughs in key bottlenecks: optical alignment and packaging precision, thermal management, standardization and ecosystem development, and cost and commercialization challenges. Currently, CPO technology is entering a critical phase of rapid transition to large-scale commercialization, with its technical route evolving toward higher integration and deep optoelectronic integration, and speed targets advancing from 800G/1.6T to 3.2T and beyond. Energy efficiency improvement, thermal management, standardization, and ecosystem building are also core elements for commercialization.

It is foreseeable that in 2026, optical interconnection technologies represented by CPO will reach an inflection point for large-scale deployment. A key technological driver is that industry giants like Nvidia plan to fully integrate silicon photonics interconnection into their next-generation AI platforms in 2026. With the advancement of large-scale commercialization in 2026 and 2027, CPO will become one of the core technologies for efficient interconnection in AI supercomputers and data centers, driving rapid development of upstream and downstream industrial chains such as photonics, packaging, and thermal management. Notably, beyond AI/HPC clusters, silicon photonics interconnection technology will also demonstrate application potential in automotive LiDAR, biosensing, quantum computing, and consumer electronics.

Trend 6: Three Engines Drive SiC/GaN Power Devices to High-Volume Inflection Point

In 2026, wide-bandgap semiconductor silicon carbide (SiC) and gallium nitride (GaN) power devices are facing unprecedented market opportunities, standing at the inflection point from technical verification to large-scale volume production. Core growth drivers come from three major areas:

(1) New Energy Vehicles (NEVs): The accelerated popularization of 800V high-voltage platforms is driving rapid penetration of SiC MOSFETs in main drive inverters. SiC devices can reduce energy loss by 50%, increase EV range by 5%–10%, and support the popularization of 350kW ultra-fast charging technology. In 2025, the penetration rate of 800V+SiC solutions in vehicles exceeded 15%, and the global SiC power semiconductor market for NEVs is expected to approach 3.5 billion US dollars in 2026.

(2) Energy Infrastructure: In photovoltaic inverters, SiC device penetration has increased from 5% in 2020 to over 25% in 2025, with system efficiency exceeding 99%. GaN power devices in data center power supplies can reduce PUE values and carbon emissions, aligning with global carbon neutrality goals.

(3) Consumer Electronics & Industrial Power Supplies: GaN technology has established a dominant position in fast charging, with power density reaching 1.03W/cc and charging efficiency up to 98%. In industrial UPS systems, the use of all-SiC modules can improve conversion efficiency by 3% and save 50% or more in power costs.

In terms of wafer size transition, 8-inch SiC wafers have become a focus of global leading enterprises. International giants such as Infineon, STMicroelectronics, and ON Semiconductor are actively deploying 8-inch SiC wafer production lines, while Chinese local enterprises like San’an Optoelectronics have put 8-inch SiC substrate mass production lines into operation. The core driver behind the transition from 6-inch to 8-inch wafers is cost-effectiveness—8-inch wafers can reduce chip unit costs by approximately 30%.

Meanwhile, the regional pattern of the SiC/GaN power device supply chain is showing a trend of "East Rising, West Declining." Internationally, traditional power semiconductor giants such as Infineon, STMicroelectronics, and ON Semiconductor maintain advantages in the high-end market. However, Wolfspeed fell into financial distress and filed for bankruptcy protection due to overly aggressive expansion strategies and intensified market competition; although the company later restructured, it still faces ongoing competition from Chinese manufacturers. In contrast, Chinese local power semiconductor enterprises have achieved rapid development in SiC through advantages in the local industrial chain and policy support—for example:

• Building a complete ecosystem through mergers and acquisitions, such as Dongwei Semiconductor acquiring Dianzheng Technology to expand power module business;
• BYD Semiconductor integrating substrate, epitaxy, and device links to reduce supply chain risks.

Overall, SiC and GaN power devices are in a transition period from "scale expansion" to "quality leap." 2026 will be a key node for industrial development—with the commissioning of 8-inch SiC wafer production lines, further cost reduction, and continuous expansion of application scenarios, third-generation semiconductor power devices are expected to play a more important role in the global energy transition and digitalization process. For Chinese enterprises, converting cost advantages into standard-setting power is crucial to achieving the leap from "follower" to "peer" and even "leader."

Trend 7: RISC-V Ecosystem Enters Explosive Growth, Automotive-Grade Chips Mass Produced

The RISC-V instruction set architecture is accelerating its entry into core automotive electronics fields. Its open and customizable characteristics are highly aligned with the demands for diversified computing power and independent and controllable supply chains in intelligent connected vehicles, indicating that the automotive-grade chip market pattern will undergo a new round of transformation.

RISC-V’s value in the automotive sector is reflected in five key aspects:

1. Open and Free Architecture: Enterprises can modify and use it flexibly, facilitating technological independence and cost optimization;
2. High Flexibility and Scalability: Supports customized instruction sets for specific applications;
3. Security Advantages: Open-source nature enables timely discovery and patching of vulnerabilities;
4. Outstanding Energy Efficiency Ratio: RISC-V cores can achieve high performance above 2.5GHz while maintaining low power consumption through a streamlined architecture;
5. Scalability Aligned with Software-Defined Vehicles (SDVs): Meets the hardware flexibility requirements of SDVs.

In March 2025, Infineon officially launched its automotive-grade RISC-V MCU product family under the AURIX series, covering multiple MCU categories from entry-level to high-performance. This move marks the substantive advancement of RISC-V in the automotive electronics sector. Nuclei Technology further pointed out that RISC-V automotive-grade CPU IP has "achieved breakthroughs in key areas."

Currently, RISC-V applications cover a wide range of in-vehicle scenarios including battery management, gateway control, body and vehicle control, and chassis control. Except for autonomous driving SoC main controllers and intelligent cockpit main controllers, which have not yet widely adopted RISC-V architecture, the majority of automotive chips already offer RISC-V solutions. Meanwhile, ecological supporting facilities are improving rapidly, with significant progress in AUTOSAR adaptation, compilers, toolchains, and operating system support, gradually building the system foundation for RISC-V’s automotive deployment.

Despite strong growth momentum, RISC-V automotive-grade chips still face challenges in ecosystem maturity—particularly in in-vehicle middleware (e.g., AUTOSAR) adaptation, toolchain completeness, and cross-platform software compatibility. Additionally, automotive-grade chips require long-cycle certification, and the automotive industry attaches great importance to supply chain stability, posing dual pressures on emerging chip enterprises.

Looking ahead, with the EU’s 2030 domestic chip capacity target, explicit support for RISC-V in China’s 15th Five-Year Plan, and the AI-driven trend of customized computing power, RISC-V is expected to occupy a significant share of the automotive chip market by 2030, gradually forming a new industrial pattern of "three-way division" with x86 and ARM.

Trend 8: Solid-State Batteries: Starting with Safety, Expanding to Diverse Scenarios

For a long time, the battery industry was regarded as mature and stable, even labeled a "sunset industry." As early as before 2011, lithium-ion batteries were already widely used in digital products such as MP3 players, MP4 players, cameras, and mobile phones. Later, the application of lithium-ion batteries in road vehicles pressed the "accelerator" for the automotive industry—China’s new energy vehicle (NEV) market has developed by leaps and bounds, accounting for over 60% of global NEV production and sales since 2023.

The call for solid-state batteries in vehicles initially emerged due to safety concerns. A series of electric vehicle fire and explosion accidents in 2025 accelerated the market’s shift toward safer solid-state battery technology. Currently, mainstream lithium-ion batteries use liquid electrolytes with lithium salts; in contrast, all-solid-state batteries do not require any liquid electrolytes, instead using pure solid electrolytes, which significantly improve battery safety and energy density, making them widely recognized as the next-generation battery technology.

Currently, the industrialization of solid-state batteries is at a critical stage of transitioning from laboratory to mass production and vehicle installation. In 2025, semi-solid-state batteries have taken the lead in small-batch vehicle deployment; mass production of semi-solid-state batteries is expected in 2026, while all-solid-state batteries will enter the phase of small-batch demonstration vehicle installation.

In response, major battery manufacturers and automakers are accelerating the industrialization of solid-state batteries through various strategies:

• CATL aims to achieve small-batch production of all-solid-state batteries by 2027;
• BYD plans to launch batch demonstration vehicle installation of solid-state batteries around 2027, with large-scale mass production by 2030;
• Automakers: GAC Group plans to mass-produce models equipped with all-solid-state batteries in 2026; Changan Automobile launched a solid-state battery with an energy density of 400Wh/kg, planning vehicle verification in 2026 and mass production in 2027; SAIC Group plans to achieve solid-state battery vehicle installation in 2027.

Greater "surprises" lie in the expansion of downstream applications. With technological maturity and cost reduction, solid-state batteries will penetrate from high-end niche markets to the mass market. It is expected that by 2030, global solid-state battery shipments will exceed 600GWh, accounting for a certain share of the high-end electric vehicle market, and gradually expanding to three-dimensional application scenarios including ground energy storage, humanoid robots, low-altitude flight, and electric aviation, ultimately driving the comprehensive upgrading of energy storage technology.

Trend 9: Toward an AI Agent-Driven Era of Unified Interaction

An AI Agent refers to an intelligent entity with autonomous decision-making, task execution, and multi-modal interaction capabilities—more like a "virtual employee" that can understand goals, plan steps, and call external resources to complete tasks.

In 2025, the AI Agent concept gained momentum and achieved pilot applications in some industries. That year, AI Agents entered flagship smartphones—for example:

• Samsung’s Galaxy S25 series deeply integrated multi-modal AI Agents into its One UI 7 system;
• Apple’s iPhone 17 series upgraded Apple Intelligence to enable context understanding and multi-turn conversation capabilities;
• Google’s Pixel 10 series deeply integrated the Gemini AI assistant, supporting cross-application operation functions.

AI Agents’ ability to autonomously plan tasks and call multi-application resources has become a core selling point of current flagship phones.

Beyond consumer products, AI Agents are also gaining popularity in enterprise applications. Gartner predicts that by the end of 2026, the proportion of enterprise applications integrating task-specific AI Agents will increase from less than 5% (August 2025) to 40%. EETimes also believes that with the acceleration of global enterprise digital transformation, AI Agents in enterprise applications will go beyond personal efficiency improvement, setting new standards for team collaboration and workflows through this more intelligent human-machine interaction model.

In 2027, approximately one-third of AI Agent deployments will adopt multi-agent collaboration models (combining agents with different skills to manage complex tasks in application and data environments) to handle complex tasks in application and data environments. By 2028, the AI Agent ecosystem will mature, with multi-agent networks capable of dynamic collaboration across multiple applications and business functions—users will be able to achieve goals without interacting with each application individually. By then, approximately one-third of user experiences will shift from traditional native application interfaces to "agent frontends," driving new business models and pricing structures. By 2029, at least half of knowledge workers will master new skills to use, manage, or create AI Agents on demand to perform complex tasks.

Currently, personal computers, smartphones, and smart home devices are integrating AI Agents. In the next few years, an era of unified interaction driven by AI Agents will arrive—users will only need to tell the AI Agent, for example, "Arrange my meetings and book flights for next week," and it will automatically complete the tasks across multiple applications and services. That will be the true "era of AI Agent as an entry point," where it is no longer just a tool but becomes the core interface for users to interact with the digital world, reshaping the application ecosystem and business models.

Trend 10: AI Empowers Manufacturing Transformation, Flexible Leasing Reshapes Production Line Models

In August 2025, the State Council issued the "Opinions on Further Implementing the 'AI+' Initiative," proposing to promote the intelligent development of manufacturing and support enterprises in applying general-purpose large models, industry-specific large models, and industrial agents in key scenarios. Breakthroughs in technologies such as large models, generative AI, machine vision, and deep learning are driving manufacturing from "automation" to "intelligence."

Currently, the manufacturing industry is in an accelerated transformation phase—artificial intelligence, as a core driver, is deeply integrating into smart manufacturing, promoting comprehensive upgrading of the industrial chain. In terms of industrial trends: while AI applications were previously limited to laboratories or single-point pilots, enterprises are now scaling deployment across production, quality inspection, and supply chain links. This trend is expected to become more pronounced in the industrial sector in 2026, with the penetration rate of AI in China’s manufacturing industry growing at a CAGR of 10% by 2027.

Specifically, AI technology enables efficient, flexible, and refined management of manufacturing processes through production process automation, predictive maintenance, quality inspection, and intelligent scheduling, significantly reducing costs and increasing capacity. Meanwhile, the synergistic application of AI with industrial internet and 5G technologies is transforming manufacturing from traditional mass production to personalization, customization, and green development, building digital competitive advantages and forming a new industrial pattern.

The deep integration of industrial robots with AI technology is becoming a key application carrier for implementing the "AI+" Initiative. For example, in the automotive manufacturing sector, industrial robots equipped with visual recognition and precision control technologies can efficiently complete complex tasks such as welding and assembly; in home appliance manufacturing, AI-enabled robots have improved production efficiency and product quality.

Currently, the order structure of the manufacturing industry is showing a trend of "small batches, multiple varieties, and short delivery cycles," leading to a significant increase in enterprises’ demand for production line flexibility. However, the high cost of intelligent equipment and rapid technological iteration pose high capital expenditure and depreciation risks for direct procurement. Against this backdrop, the equipment leasing model has emerged—with its low threshold, high flexibility, and controllable costs, it has become a practical choice for enterprises to "use instead of purchase," deriving innovative solutions such as "pay-per-use" and "pay-per-result." By 2026, equipment leasing is expected to gradually penetrate the manufacturing industry, helping the market transition from the traditional "fixed cycle + fixed fee" model to a more flexible and refined service model.

The expansion of equipment leasing relies on the continuous evolution of robot capabilities. With technological progress, light-load, low-complexity scenarios such as entertainment, guided tours, and emotional companionship are gradually opening up. However, widespread application in the manufacturing sector still requires breaking through the key bottleneck of "operational intelligence." In the future, the increase in leasing penetration will be a long-term trend driven by strong demand for flexible deployment on the demand side and accelerated service-oriented transformation on the supply side.