The release of the "Top Ten AI Trends for 2026" by China Media Group has injected robust industrial momentum and forward-looking insights into the upcoming 2026 Guangzhou Healthcare Industry Expo (Guangzhou Medical Expo).

During the expo, a dedicated AI Healthcare and Digital Health Exhibition Zone will be set up. From August 21 to 23, 2026, healthcare professionals worldwide are cordially invited to join this grand event. Let us take technology as our tool and innovation as our sail to pioneer a new era of intelligent healthcare together!

China Media Group (CMG), in collaboration with institutions including the China Academy of Electronics and Information Technology (under the Ministry of Industry and Information Technology), Zhongguancun Science City Administration Committee, Wuhan East Lake High-tech Development Zone Administration Committee, University of Science and Technology of China, Huazhong University of Science and Technology, Institute of Artificial Intelligence of Hefei Comprehensive National Science Center, Hefei Institute of Artificial Intelligence and Big Data, and Science Popularization China, has released the Top 10 AI Trends for 2026.

1. Globalization of AI Governance

Inclusive and shared AI has become a core issue on the global development agenda.In November 2025, President Xi Jinping proposed at the APEC meeting: "China has initiated the establishment of a World Artificial Intelligence Cooperation Organization. We hope to actively provide AI public goods to the international community through cooperation on development strategies, governance rules, and technical standards." Technological and infrastructure development aims to bridge the digital divide and deepen international cooperation on AI development and governance, which is of great significance for promoting the healthy development of AI, boosting global economic growth, and addressing global challenges such as climate change, public health, and educational equity.

2. Large-scale Intelligent Computing Power

The supply of key industrial elements will be further enhanced.Domestic AI chips will be applied on a large scale in specific scenarios, and the computing power infrastructure supporting the development of large models will become increasingly sophisticated. Domestic computing power chips are developing rapidly; new architectures such as Application-Specific Integrated Circuits (ASICs) and in-memory computing will drive technological breakthroughs, and a coordinated software-hardware ecosystem will gradually take shape. 10,000-card-level clusters will become the mainstream carrier for large model training; ultra-large-scale cluster technology will make breakthroughs, and high-speed interconnection and green low-carbon technologies will advance simultaneously. The "Eastern Data, Western Computing" project will promote the coordinated scheduling of computing power resources nationwide, greatly improving the inclusiveness of computing power.

3. Mainstreaming of Applications

AI agents will fully integrate into scenarios.In 2026, the paradigm of AI application development will shift from pursuing general capabilities to addressing industry pain points in vertical sectors. As technologies such as the agent technology stack and interaction protocols mature, enterprise-level agents will be deployed on a large scale in core business links (e.g., R&D, customer service, office automation), gradually gaining the ability to handle complete business closed loops—marking the accelerated transformation of AI innovation from the laboratory to real productivity. At the policy level, the Implementation Opinions on the "AI + Manufacturing" Special Action proposes that by 2027, 1,000 high-level industrial agents will be launched, and a batch of enabling application service providers "proficient in both AI and industries" will be built, aiming to cultivate industry agents and AI-native enterprises. The State Council's Opinions on Further Implementing the "AI +" Action proposes that by 2030, the penetration rate of agent applications will exceed 90%.

4. Practical Multimodality

AI core technologies are evolving from "specialized tools" to "general intelligent partners."In 2025, domestic large models such as DeepSeek achieved "high performance, low cost" breakthroughs, driving a significant reduction in the threshold and cost of AI technology applications and opening new paths for global large model development. Currently, global computing power upgrades support long-context processing of millions of Tokens; by integrating multi-source data (text, images, audio, video, 3D point clouds, etc.), human-computer interaction is evolving toward "what you see is what you get" multimodal interaction. Meanwhile, "world models" with reasoning and planning capabilities are becoming a competitive focus; by simulating action consequences, they drive AI to evolve from perceptual intelligence to decision-making intelligence.

5. Popularization of Native AI Terminal Hardware

Next-generation smart terminals integrate with immersive consumer scenarios.In 2025, the consumer electronics industry showed a significant differentiation trend. On one hand, hardware parameter iterations are increasingly approaching physical and cost limits; technologies such as foldable screens, image sensors, and fast charging are becoming homogeneous during popularization, plunging the market into red-ocean competition. On the other hand, AI phones and various AI hardware continued to grow, standing out in shipments, user attention, and ecosystem building, becoming a key driver of industry growth.Terminal hardware will shift from simple "tool adaptation" to "native AI design." Next-generation AI phones, PCs, and XR devices will deeply integrate with multimodal large models, spawning new consumer scenarios of virtual-real coexistence and bringing qualitative leaps in personalized education, health management, and entertainment experiences.

6. Embodied Intelligence of AI

The integration of "Physical AI" and "embodied intelligence" drives deep interaction between robots and reality.In 2025, embodied intelligent robots achieved more technological breakthroughs: ditching remote controls, equipped with the world's first general visual perception system for humanoid robots, they can run autonomously on sports tracks; moving from prototypes to mass production, they entered real scenarios (e.g., inspection, service halls, factories, elderly care/medical care) and secured 100-million-yuan orders. In 2025, China's embodied intelligence market scale is expected to reach 5.295 billion yuan, accounting for about 27% of the global market; from the body to the "brain," embodied intelligent robots are iterating rapidly. The deep integration of "Physical AI" and "embodied intelligence" will drive smart robots from structured environments to more complex open scenarios. In 2026, smart robots are expected to launch iconic products in manufacturing, warehousing, and home services, entering large-scale trial use and achieving deep interaction with the real world.

7. Further Segmentation and Deepening of Professional Fields

"AI for Science" (AI4S) will produce disruptive results in basic disciplines.Scientific intelligence drives the evolution of next-generation AI, fully empowering the transformation of scientific research paradigms. AI large models deeply integrate with scientific computing, starting to independently propose hypotheses, design experiments, and verify results. They are accelerating the "0-to-1" process in fields such as life sciences (e.g., antibody design, new drug molecules), materials science, and astrophysics. By strengthening the collaborative innovation of the three core elements (algorithms, computing power, data), China is building an intelligent scientific research toolchain in cutting-edge fields (e.g., protein structure prediction, quantum simulation, materials genome), driving scientific research efficiency from linear growth to exponential leap.

8. Cross-integration of Cutting-edge Fields

Brain-inspired intelligence and interdisciplinary disciplines are accelerating innovation.Brain science explores the essence of cognition, consciousness, and intelligence—it is the "ultimate frontier" for humans to understand nature. Brain-inspired technologies, inspired by brain science, develop intelligent sciences and technologies such as brain-inspired algorithms, devices, and robots. Brain science drives the development of cutting-edge branches (e.g., bioimaging, data science); brain-inspired technologies optimize AI algorithms and empower applications (e.g., autonomous driving, smart healthcare). The deep integration of brain science and AI will drive hardware and algorithm breakthroughs in disruptive technologies such as spiking neural networks and neuromorphic computing.

9. Highlighted Energy Issues

Green AI gains attention.The massive energy consumption of AI data centers will account for a significant portion of global incremental electricity demand, raising concerns about energy supply and the environment. Regions that can provide large-scale, low-cost, reliable, and clean electricity will have a structural advantage in attracting AI-related investments.At the "AI and Green Low-carbon Development" forum of the 2025 World Artificial Intelligence Conference, China's approach to solving the "AI energy paradox" was proposed. Power Construction Corporation of China released the "Energy-Carbon Intelligent Computing Hub"—aiming to build a "digital foundation" and "central system" for the future green intelligent economy. It integrates previously independent "energy flows, carbon flows, and data flows" for integrated collaborative management and global optimization, allowing energy and business experts to issue instructions in their own language and collaborate with AI to complete complex green low-carbon optimization tasks. This marks a key step for the industry from decentralized applications to systematic top-level design.By developing more efficient model architectures, using clean energy computing power centers, and exploring new energy supply models (e.g., Small Modular Reactors, SMRs), the industry will strive to balance computing power growth and carbon emission control.

10. Intensified Security and Adversarial Challenges

Security and governance will become important guarantees for AI development.As data poisoning, adversarial attacks, and Deepfakes become real threats, security protection will become an inherent demand in AI model development. At the main forum of the 2025 National Cyber Security Publicity Week, the AI Security Governance Framework (Version 2.0) was officially released, marking the shift of AI governance from principle formulation to a new stage of systematization, dynamism, and standardization. In response to risks brought by the rapid development of generative AI (e.g., data abuse, algorithmic discrimination, model out-of-control), the new framework strengthens risk classification, adds derivative security dimensions, promotes full-process prevention and control and ethical pre-positioning, and realizes synergy between technology, ethics, and social governance. Through hierarchical and classified supervision and institutional connection, it provides compliance paths for enterprises, supports the healthy and orderly development of the industry, and demonstrates China's governance wisdom of "balancing development and security" in global AI governance.Governance rules and technical tools for AI ethics, privacy, and security will be improved simultaneously. While encouraging technological innovation, they will strengthen security defenses to ensure the healthy and orderly development of the AI industry.

Source: CCTV News