US AI vs China AI

US AI China AI AI competition AI infrastructure AI chips

US AI vs China AI Brief — 2026-09-18

Posted on September 18, 2026 at 09:13 PM

US AI vs China AI Brief — 2026-09-18

Top Stories

1. Chinese AI Models Generate Only About 10% of OpenAI and Anthropic Revenue

Source: South China Morning Post · September 18, 2026
Summary: Seven major Chinese AI developers generated an estimated $10.7 billion in annual recurring revenue between March and August, according to Rhodium Group. That compares with more than $100 billion combined for OpenAI and Anthropic. ByteDance led the Chinese group at roughly $4 billion ARR, followed by Alibaba at $2.4 billion.
Why It Matters: The data highlights an important asymmetry in the US-China AI competition: Chinese models can achieve substantial adoption and technical visibility without yet matching US frontier labs in monetization. The commercial gap remains a critical variable alongside model capability and compute.
URL: https://www.scmp.com/tech/big-tech/article/3367991/top-chinese-ai-models-make-10-openai-anthropic-revenue-despite-high-valuations-report

2. Huawei Expands Its Agentic Cloud Strategy Globally

Source: Huawei · September 18, 2026
Summary: Huawei Cloud announced the global rollout of its AI Cluster Service and an Agentic Model-as-a-Service platform at HUAWEI CONNECT 2026. Huawei said its AgentArts enterprise agent platform already serves more than 100 enterprises, while its Industry AI Foundry has accumulated more than 1,000 industry assets and supports more than 1,000 deployed projects.
Why It Matters: China’s AI strategy is increasingly moving beyond foundation models toward vertically integrated infrastructure, cloud platforms and enterprise deployment. The development strengthens Huawei’s role as an alternative AI stack spanning compute, models, cloud and applications.
URL: https://www.huawei.com/en/news/2026/9/hc-agentic-infra-industry-ai

3. US-China AI Safety Debate Moves Toward Independent Evaluators

Source: Axios · September 18, 2026
Summary: US AI companies, policymakers and safety organizations are debating how advanced models should be independently evaluated while federal AI regulation remains limited. The discussion includes possible third-party evaluators, voluntary incident reporting and even proposals for US and Chinese laboratories to review each other’s models. Treasury Secretary Scott Bessent is also expected to discuss AI safety with Chinese officials.
Why It Matters: AI competition is increasingly extending from model performance into the ability to measure and govern frontier capabilities. The difficulty is that the same geopolitical rivalry driving the race also makes shared evaluation standards and cross-border safety cooperation harder to establish.
URL: https://www.axios.com/2026/09/18/ai-safety-evaluators-metr-white-house-trump

4. US-China AI Rivalry Becomes a Central Issue Ahead of Xi-Trump Summit

Source: Financial Times · September 18, 2026
Summary: The upcoming Trump-Xi summit is taking place amid sharply different approaches to AI development and governance. US policymakers continue to emphasize maintaining technological leadership, while China is positioning AI development as a strategic industrial priority; disputes over model distillation, regulation and technology restrictions remain unresolved.
Why It Matters: AI is becoming intertwined with the broader US-China relationship rather than remaining a standalone technology issue. Decisions around chips, model access, safety standards and international deployment could influence the structure of the global AI ecosystem for years.
URL: https://www.ft.com/content/a528adc7-3a8c-4a00-826f-1fdde748efb5

5. China’s CXMT Targets NAND as AI Server Demand Tightens Memory Supply

Source: Reuters · September 18, 2026
Summary: China’s leading DRAM producer CXMT is preparing an R&D production line for NAND flash memory, according to sources cited by Reuters. AI-server demand has contributed to a global memory shortage, while Chinese memory companies continue expanding under national and local government support. The move would put CXMT into more direct competition with Samsung, SK Hynix, Micron and China’s YMTC.
Why It Matters: AI competition increasingly depends on the entire semiconductor stack rather than GPUs alone. China’s push into memory could reduce another dependency in its AI infrastructure while intensifying competition for components needed by large-scale AI systems.
URL: https://www.reuters.com/world/asia-pacific/chinas-cxmt-eyes-flash-memory-push-amid-global-shortage-firm-take-samsung-ymtc-2026-09-18/

6. Chinese Embodied-AI Startup Targets a ChatGPT-Like Breakthrough for Robots

Source: Reuters · September 18, 2026
Summary: Beijing-based Spirit AI says it expects a major advance in embodied AI by mid-2027, enabling robots to understand natural-language instructions and execute sequences of physical actions. The company has raised more than $670 million and has robots operating on production lines at CATL and JD.com. Its training approach emphasizes large quantities of real-world human-motion data rather than relying primarily on simulation.
Why It Matters: The US-China AI competition is expanding from language models into physical AI and robotics. Progress in embodied intelligence could become strategically important for manufacturing, logistics and industrial automation, where China’s large industrial base provides a substantial deployment environment.
URL: https://www.reuters.com/world/asia-pacific/founder-chinese-startup-spirit-ai-says-robot-brains-set-2027-breakthrough-2026-09-18/

Strategic Takeaways

The competition is becoming a full-stack contest. Models remain central, but compute, memory, networking, cloud infrastructure and robotics are increasingly strategic layers.
US commercial leadership remains significant. The latest revenue comparison shows a large monetization gap between major Chinese AI models and leading US frontier labs.
China is emphasizing infrastructure sovereignty and deployment. Huawei’s latest announcements illustrate an approach built around domestic compute, cloud, enterprise agents and ecosystem development.
Open and closed models are becoming geopolitical instruments. Model access, distillation, open-weight distribution and third-party routing are increasingly intertwined with national technology policy.
AI governance is becoming part of the competitive landscape. US-China discussions are moving toward questions of evaluation, incident reporting and potentially shared safety standards, even as strategic mistrust remains high.

More in US AI vs China AI