CMG announces top 10 AI trends for 2026

The genie is out of the bottle, and its name is artificial intelligence (AI).

In China and around the world, AI is becoming integral to human activity, from industry to everyday interactions, making its governance and inclusive access critically important.

To help navigate this future, China Media Group (CMG), in collaboration with multiple think tanks and universities, is releasing a report on the top 10 trends in AI for the year ahead. 

1. Globalization of AI Governance

AI for inclusive and shared benefits has become the central issue on the global development agenda, CMG's trends report says.

In November 2025 at the 32nd APEC Economic Leaders' Meeting, Chinese President Xi Jinping said, "China has proposed the establishment of a World Artificial Intelligence Cooperation Organization to provide the international community with public goods on AI." 

Strengthening international cooperation on AI governance is seen as vital to supporting global economic growth and tackling challenges including climate change and public health, the report notes.

2. The scaling of intelligent computing power

Highlighted in the trends analysis is the central role of strengthening key industrial input supply. Chip technologies, in particular, are advancing rapidly, driving a significant expansion in computing power. Domestic AI chips are set to achieve large-scale deployment in specific application scenarios. 

Clusters with tens of thousands of GPUs have become the mainstay for training large models. China's "Eastern Data Western Computing" project has significantly improved access to computing power.

3. Mainstream adoption of AI applications

According to the CMG report, AI agents are set to be deployed across a wide range of industries. In 2026, the application of intelligent agents is expected to shift from general-purpose tasks toward addressing specific issues. 

On January 8, the Chinese government released an action plan aiming to achieve a secure and reliable supply of key core AI technologies. The plan called for launches of 1,000 high-level industrial AI agents by 2027.

4. Deployment of multi-modal interaction

Core AI technologies are evolving from specialized tools into intelligent partners, the report finds.

Domestic large language models such as DeepSeek delivered breakthroughs in high performance at lower cost in 2025, sharply reducing the barriers and expenses associated with AI deployment. Upgrades in computing capacity are now supporting increasingly sophisticated interactions based on multi-modal data, including text, images, audio, video and 3D point clouds.

5. Proliferation of native AI devices

Next-generation smart terminals are converging with immersive consumption experiences, the trends report shows. In 2025, AI smartphones and a range of AI-powered devices continued to post strong growth. 

Terminal hardware is shifting from AI-adapted tools toward AI-native design. A new generation of AI smartphones, PCs and XR (Extended Reality) devices is expected to be deeply integrated with multi-modal large models, redefining education, health management and entertainment experiences.

6. Convergence of AI and embodied intelligence

The report sketches out how the convergence of "physical AI" and embodied intelligence creates robots that learn through deeper interaction with the real world and adapt to complex environments, thus becoming autonomous and capable of collaborating with humans. Robots are ready to take the next step from prototype development to mass production, finding applications in inspection, service halls, factories, elderly care and healthcare. 

China's embodied intelligence market is expected to reach around 5.3 billion yuan (approximately $759 million) in 2025, accounting for about 27 percent of the global total. In 2026, intelligent robots for large-scale production in areas including manufacturing, warehousing and home services are on track.

7. Specialization within scientific domains 

"AI for Science" is delivering disruptive breakthroughs in fundamental research, the report says. As AI models are more deeply integrated with scientific computing, they are increasingly capable of generating hypotheses, designing experiments and validating results, speeding "from zero to one" breakthroughs across materials science, astrophysics and life sciences, including antibody design and novel drug molecules.

8. Convergence across frontier fields

Brain-inspired intelligence is converging with other frontier disciplines, the report notes, pointing out that advances in brain science are driving progress in fields such as biological imaging and data science.

Meanwhile, brain-inspired technologies are refining AI algorithms for applications in autonomous driving and intelligent healthcare. Deeper integration between brain science and AI is likely to yield breakthroughs in spiking neural networks and neuromorphic computing.

9. Heightened focus on energy issues

The report places emphasis on the emerging field of "Green AI." The rapid growth of AI data centers is expected to significantly lift global electricity demand, heightening concerns over energy supply and environmental impact, while favoring regions with access to low-cost, reliable and clean power. 

With the development of more efficient model architectures and the use of clean-energy-powered computing centers, the industry seeks to balance the rapid growth of computing power with carbon emissions control.

10. Escalation of safety and adversarial challenges

In a move reflecting fast-paced breakthroughs in AI, China released its upgraded AI Safety Governance Framework 2.0 on September 15. The framework promoted the formation of a safe, trustworthy and controllable AI development ecosystem, establishing a collaborative governance model that spans borders, fields and industries.

Safety is more critical for the AI development, the report notes, adding that governance rules and technical tools addressing AI ethics, privacy and security are expected to advance rapidly to ensure the orderly and sustainable development of the AI industry.