Concept picture of AI Photo: VCG

From patent licensing to a Chinese AI model joining a US cloud platform, recent developments are highlighting practical openings for China-US artificial intelligence cooperation, experts told the Global Times, identifying commercial partnerships and shared risk management as areas where progress can begin despite continuing differences.

On October 5, Chinese tech giant Huawei and US chipmaker Qualcomm announced a multi-year patent license agreement covering cross-licenses in 5G, computing, AI and networking, alongside Qualcomm's purchase of certain Huawei US patents. The transaction will close after necessary regulatory approvals, according to their announcement.

On the following day, Amazon Web Services (AWS), the cloud business of US technology company Amazon, announced that Chinese startup Zhipu AI's GLM 5.3 was available on Amazon Bedrock. Zhipu told the Global Times that AWS shares revenue with it based on model usage, and that it has implemented similar arrangements with several other overseas cloud providers.

The developments follow a policy opening. Among eight deliverables and understandings published on September 26, China and the US agreed to establish an AI dialogue on risks and benefits, with the next exchange scheduled for November, and a bilateral communication channel for AI incidents, according to China's Foreign Ministry. 

Chen Jing, vice president and research fellow at the Society for the Study of Technological and Strategic Trends, told the Global Times that commercial interests give cooperation "resilience beyond political cycles," making business partnerships a realistic starting point between China and the US despite geopolitical constraints. Meanwhile, Liu Gang, chief economist at the Chinese Institute of New Generation AI Development Strategies, said the two countries should draw on their respective strengths while managing risks.

Complementary strengths

Zhipu said the deal gives it another source of revenue on top of its own API and enterprise services. Putting its models on cloud platforms lets it reach those platforms' business customers and tap into their infrastructure, with revenue growth depending on paying customers and usage, according to a statement.

AWS said eligible enterprise customers can use GLM 5.3 through managed interfaces without operating their own inference infrastructure. The platform provides deployment capabilities alongside the Chinese developer's model, while Zhipu's revenue-sharing arrangement links its returns to customer usage.

Chen pointed to patent licensing and integrating models with overseas cloud platforms as two opportunities companies can realistically pursue. Established deal structures and legal frameworks are already in place for both. As he put it, companies don't have to wait for complicated government-to-government negotiations - they can go through market channels and get real results.

Across the AI value chain, Chen pointed to US strengths in foundation model research, advanced chip design and cloud ecosystems, alongside China's engineering capabilities, training and inference optimization, large-scale deployment and rapid industrial iteration. Liu said exchanges could help both sides learn from their differing development priorities. 

Beyond existing commercial deals, Chen suggested rare-disease diagnosis as a potential area for cooperation. Limited patient numbers for individual conditions create a need for international collaboration, he said, identifying US model and medical research capabilities and China's clinical cases and engineering capacity as complementary resources.

He proposed exploring federated learning and privacy-preserving computing to support collaborative training while keeping underlying data within national borders. Algorithms and model architectures could be combined with clinical applications and validation environments to develop screening and diagnostic assistance tools, he said.

Clear data-protection standards, intellectual property and benefit-sharing arrangements, and regulatory communication on AI medical devices would be necessary, Chen stressed. Liu also urged continued dialogue wherever feasible on chips, computing infrastructure and communications, while acknowledging that restrictions make cooperation in these fields more difficult.

Shared risks, common ground

The rapid development of AI is creating shared challenges that could provide common ground for China-US cooperation on technology governance.

Chen warned that AI agents could pose risks ranging from unsupervised financial transactions and network operations to automated cyberattacks, large-scale disinformation and disruptions spreading across interconnected critical infrastructure. As such risks transcend national borders, China and the US share an interest in strengthening dialogue on AI safety and governance, even amid intensifying technological competition.

International work offers a reference point for incident reporting without requiring identical domestic rules. An OECD framework published in February 2025 allows countries to adopt a common reporting approach while tailoring responses to their own policies and legal frameworks.

Chen proposed prioritizing notifications of major incidents involving loss of control over advanced models, serious real-world harm and large-scale cross-border misuse. Reports could cover the incident's nature, scope, preliminary attribution and response measures without disclosing model weights or other core commercial secrets.

Notification exercises, comparisons of risk terminology and classification systems, and joint reviews of incidents could help turn dialogue into verifiable results, he said. Liu called for sustained exchanges among think tanks, academics and businesses, combining broader research with experience from the front lines of deployment.

At the opening of the Global Dialogue on AI Governance in Geneva on July 6, UN Secretary-General Antonio Guterres called for common methods to evaluate and verify risks. He also emphasized sharing knowledge and expanding access to AI capacity-building, particularly for developing countries.

Chen said shared assessment methods and risk cases could help developing countries strengthen governance capacity through platforms such as the UN and G20. Liu stressed that such cooperation should be supported by workable arrangements covering technical standards, application boundaries and safety controls.

China and the US should pursue "cooperation and risk management together," Liu said. By leveraging their technological and industrial strengths, maintaining dialogue and strengthening risk prevention, the two countries can advance bilateral scientific and technological cooperation while making a positive contribution to global AI governance, Liu said.