The global market research organization TrendForce has identified the top ten key trends for the development of the technology industry in 2025. Details are as follows:
Generative AI Driving Innovation:
Humanoid and service robots are set to undergo significant upgrades.
With the Advancements in AI and Robotics Technology, Market Focus on Robots Will Continue in 2025
As AI and mechanical power technologies mature, coupled with the aggressive efforts of major companies like NVIDIA and Tesla, robotics will remain a key focus in 2025. In terms of technological development, software platforms are concentrating on machine learning training and digital twin simulations. Meanwhile, complete robot systems are emphasizing collaborative robots, mobile robotic arms, and humanoid robots, designed to adapt to various environments and human-robot interactions.
Notably, with active investments from companies in the U.S. and China, humanoid robots are expected to gradually achieve mass production starting in 2025. The global humanoid robot market is projected to grow at a compound annual growth rate (CAGR) of 154% from 2024 to 2027, with market value likely exceeding $2 billion.
Looking at overall application scenarios, industrial robots will primarily focus on arm-based picking tasks. In contrast, service robots, powered by generative AI, will enable multimodal interactions, information retrieval, text summarization, and schedule planning. These features highlight their high mobility, strong companionship, and broad functionality, making them a central focus for robotics development in the coming years.
Technological Innovation Drives Market Standardization: AI Laptops to Reach 21.7% Penetration Rate by 2025
With rapid advancements in technology, AI-enabled laptops are expected to become the standard in the market over the next few years. The penetration rate of AI laptops is projected to reach 21.7% by 2025 and climb to nearly 80% by 2029. This growth in AI laptops is also a key driver for the rising adoption of Arm-based architectures. Compared to traditional x86 architectures, Arm offers higher energy efficiency and greater scalability. As the demand for on-device inference increases, energy-saving considerations will drive the market share of Arm-based laptops upward each year. Additionally, the growing adoption of Windows on Arm systems will allow more consumers to experience these high-performance, low-power AI laptops.
While AI applications currently rely heavily on cloud computing, TrendForce anticipates that breakthrough advancements in Edge AI will become a crucial factor in driving the adoption of AI laptops. Edge AI shifts computation from the cloud to local devices, enabling laptops to process real-time applications such as voice commands and image recognition more quickly and efficiently, thereby enhancing the user experience. This localized processing also ensures user privacy, making it well-suited for handling sensitive data and building consumer trust in AI laptops.
As AI technology matures, Edge AI is expected to unlock new productivity potentials for laptops, including smart office solutions and automated process management, catering to diverse user needs.
AI Server Shipments Expected to Grow Over 28% in 2025, with Focus on Improving HBM 12-Hi Yield Rates
Driven by the demand for AI infrastructure from cloud service providers (CSPs) and enterprise clients, global AI server shipments (including systems equipped with GPUs, FPGAs, and ASICs) are projected to grow by 42% in 2024. In 2025, sustained high demand from CSPs and sovereign cloud providers is expected to result in an annual growth rate exceeding 28%, with AI servers accounting for 15% of total server shipments.
With NVIDIA’s B300 and GB300 adopting HBM3e 12-high (12-hi) stacks, the 12-hi configuration is set to become the mainstream stacking standard in the industry starting in 2025. SK hynix is employing Advanced MR-MUF technology for its 12-hi generation, incorporating a mid-temperature pre-bonding process for each layer of die stacking and improving MUF materials. This approach extends the processing timeline to effectively control die warping.
Samsung and Micron, meanwhile, are continuing to use TC-NCF stacking for their 12-hi generation. While this technique offers advantages in die warping control, it faces challenges such as longer processing times, greater accumulated stress, and weaker thermal dissipation, leading to significant uncertainty in ramping up production yields.
As the adoption of 12-hi stacks is expected to extend from HBM3 to HBM3e, HBM4, and HBM4e (2027–2029), the extended production timeline poses challenges. Ensuring stable and improved mass production yield rates for 12-hi stacks will be a critical focus for suppliers in the coming year.
Focus on 2025: Semiconductor Technology and CoWoS Demand Surge Underpinned by Advanced Processes and AI
With the introduction of EUV lithography technology at the 7nm node, FinFET structures have begun to encounter physical limits starting at 3nm, leading to a divergence in advanced process technologies. TSMC and Intel continued FinFET development, mass-producing 3nm products in 2023. While Samsung attempted to adopt MBCFET (Multi-Bridge Channel Field-Effect Transistor) based on GAAFET (Gate-All-Around Field-Effect Transistors) at 3nm in 2022, the process has yet to achieve high-volume production.
By 2025, TSMC is set to transition to nanosheet transistor architecture for its 2nm process, while Intel will introduce RibbonFET at its 18A node. Samsung remains focused on improving its 3nm MBCFET process, aiming to achieve large-scale production by 2025. These three companies will officially compete in the GAAFET architecture arena, aiming to provide chips with better performance, lower power consumption, and higher transistor density per unit area through enhanced gate control via four-sided contact.
The growing demand for custom chips and larger packaging sizes driven by AI applications is simultaneously propelling CoWoS (Chip-on-Wafer-on-Substrate) demand in 2025. Key developments in the CoWoS market for the year include:
Increased NVIDIA Demand for TSMC CoWoS: NVIDIA’s share of TSMC’s CoWoS capacity is expected to rise to nearly 60%, driving TSMC’s monthly CoWoS production capacity to nearly double, reaching 75–80K units by year-end.
Shift Toward CoWoS-L: The ramp-up of NVIDIA’s Blackwell platform in the first half of 2025 will drive demand for CoWoS-L to surpass CoWoS-S, with CoWoS-L expected to account for over 60% of total demand.
CSP Investments in ASIC AI Chips: Cloud service providers (CSPs), including AWS, are increasing their investments in ASIC AI chip development, significantly boosting CoWoS demand in 2025.
Cybersecurity in 2025: AI as a Double-Edged Sword for Strengthening Defense and Threat Detection Against Complex Attacks
The global focus of cybersecurity remains on software and hardware tailored to the cloud and IoT era. However, as technologies continue to advance, the complexity of both defensive and offensive measures has significantly increased. This has prompted a shift in focus from IoT-centric approaches to AI-driven solutions.
In the realm of generative AI (Gen AI), two major trends in enhancing cybersecurity defense have emerged:
Empowering Operators: Gen AI enables operators to use natural language for risk analysis and response by providing automatic translation and data summarization tools, making it easier to identify and address critical threats.
Accelerating Threat Detection: It helps users swiftly identify vulnerabilities and offers actionable recommendations to reduce detection cycles.
However, Gen AI is also leveraged by attackers, enhancing methods such as enumeration and phishing.
Analyzing the risks associated with large language models (LLMs), potential vulnerabilities include generating erroneous outputs from manipulated inputs, introducing flaws during the training phase, lacking comprehensive access controls, and granting excessive autonomy to AI systems. These risks will be critical cybersecurity challenges for enterprises developing AI-based products and services in 2025.
AMOLED Expands into Mid-Sized Applications, Driving Laptop Market Penetration to 3%
In 2024, Apple is set to launch its iPad Pro series featuring RGB AMOLED panels, signaling the expansion of RGB AMOLED technology into mid-sized product applications. Beyond tablets, the adoption of AMOLED panels in laptops is also gaining momentum. Although Apple plans to introduce AMOLED panels to its MacBook series between 2026 and 2027, the company has already encouraged panel manufacturers to expand their investments. This includes scaling RGB AMOLED production lines from Gen 6 to Gen 8.6 or 8.7 to meet future demand.
With this trend firmly established, other brands are positioning themselves early, leveraging existing production lines to carve out market opportunities. By 2025, the scale of AMOLED-equipped laptops is expected to exceed 6 million units, with market penetration projected to reach 3%.
Vision Pro Shifts VR/MR from Entertainment to Productivity Tools; LEDoS Sets Milestones for AR Device Weight and Visual Experience
One of the most pivotal events for VR/MR head-mounted devices in 2024 is Apple’s launch of the Vision Pro. This device successfully repositions VR/MR technology from entertainment and leisure applications to productivity tools, inspiring other manufacturers to introduce new products. The Vision Pro’s display leverages OLEDoS technology, delivering resolutions exceeding 3,000 PPI, making it the leading choice for high-end VR/MR near-eye displays. TrendForce projects VR/MR device shipments to reach 37 million units by 2030.
AR glasses, positioned as assistive devices, will see renewed market focus in 2024 thanks to advancements in AI technology. While Meta’s Orion is not a mass-produced device, it features an LEDoS display and SiC waveguides, offering a field of view (FOV) of up to 70 degrees and weighing less than 100 grams—establishing a new benchmark for lightweight design.
In addition to LEDoS, current near-eye display technologies applicable to AR glasses include OLEDoS, LCoS, and LBS (Laser Beam Scanning). This diversity enriches the development of AR displays, providing hardware designers with greater flexibility. TrendForce forecasts that AR device shipments will reach 25.5 million units by 2030.
Satellite Industry Trends for 2025: Miniaturized CubeSats and Low-Cost Mass Production Drive Global Communication and IoT Revolution
With the guidance of 3GPP Release 17 for satellite application scenarios, the number of CubeSats within low Earth orbit constellations is experiencing exponential growth. Emerging satellite companies are leveraging low-cost production of miniaturized CubeSats and large-scale deployment of satellite constellations to deliver global low-latency satellite communication coverage.
Looking ahead to 2025, the trend toward satellite miniaturization is enabling small to mid-sized satellite operators to adopt modular satellite platforms and commercial off-the-shelf (COTS) components for large-scale CubeSat production, significantly reducing production costs. These startups are deploying CubeSat constellations to support applications such as space situational awareness (SSA), which involves monitoring and clearing space debris.
Additionally, satellite IoT applications are rapidly advancing, with use cases including monitoring IoT devices in remote areas, such as agricultural sensors, further expanding the reach and utility of satellite technology.
2025: A New Era for Autonomous Driving with Modular End-to-End Model Production and Accelerated Commercialization of Level 4 Robotaxis
Autonomous driving, a key application of edge AI, is entering a new phase as Tesla's push for end-to-end models gains momentum. This trend has prompted intensified industry efforts to enhance AI technology and computational power. By 2025, other automakers are expected to begin mass production of such models, primarily adopting modular end-to-end architectures, which offer advantages in interpretability and debugging.
End-to-end models are data-driven and heavily reliant on diverse datasets. Generative AI, with its openness and creativity, is being used to simulate varied and rare scenarios, assisting in model training and addressing long-tail data challenges.
Advancements in AI are also accelerating developments in the commercial realm. Level 4 autonomous Robotaxis are poised to benefit from increasingly supportive regulatory environments, enabling faster scenario replication and commercialization.
However, both electrification and autonomous driving technologies face growing challenges from geopolitical factors, which could complicate technological and commercial expansion efforts.
2025: EVs and AI Data Centers Drive Innovations in Battery and Energy Storage Technologies
Growth in the electric vehicle (EV) market is slowing, particularly for battery electric vehicles (BEVs), whose growth rate is projected to drop to 13% in 2025. Range anxiety remains a significant barrier to BEV adoption, driving industry-wide efforts to address this challenge.
In battery technology, CATL has introduced lithium iron phosphate (LFP) batteries with a 4C charging rate, enabling a 10-minute charge to deliver a 600-kilometer range. The company plans to expand its market deployment in 2025. Additionally, semi-solid-state batteries entered mass production in 2024, with accelerated adoption in vehicles anticipated in 2025. Fully solid-state batteries are expected to enter mass production post-2027.
On the charging infrastructure front, megawatt-level charging equipment, launched in 2024 and designed for commercial trucks and passenger vehicles, is expected to propel advancements in high-power charging technology. These innovations aim to alleviate range anxiety while driving demand for efficient charging solutions and extended driving ranges.
At the same time, with the rapid advancement of charging technology, major automakers are focusing on improving the overall performance and user experience of electric vehicles (EVs) to adapt to market changes and maintain competitiveness. In 2024, newly implemented smart connectivity and autonomous driving technologies are becoming widely integrated into EVs. These innovations not only significantly enhance energy efficiency but also elevate intelligence and safety to new levels.
In addition, the accelerated deployment of AI data centers in 2024 is driving a surge in demand for advanced energy storage systems. With continuous technological advancements and declining costs, global energy storage installation demand is expected to reach 92 GW/240 GWh in 2025, representing annual growth rates of 25% and 33%, respectively.
The rapid development of AI technology is leading to a substantial increase in electricity demand. Energy storage systems play a critical role in ensuring reliable power supply for data centers, especially during renewable energy fluctuations or power outages. As the data center industry and its related supply chain continue to expand rapidly, the scale of data center construction is expected to sustain steady growth, creating vast opportunities for advanced energy storage systems.
Comentarios