Is it a declaration of war? On July 31, at the opening ceremony of the World Artificial Intelligence Conference at Shanghai, Chinese President Xi Jinping positioned China as the leader of a new AI global order, challenging the US. Xi was reiterating the Chinese intent to not just shape, but govern this space. Because, while the US retains decisive advantage in frontier model performance, hi-end compute, and private capital, China already leads in research, patent volume, and industrial deployment.
The Chinese pitch was well thought through. Their articulation of this battle-of-giants hinges around expensive, less accessible, closed-source models (grappling with breach issues) versus the general public availability of their own open-source models. Xi Jinping’s says, “AI development should not be a solo performance by a single country, but a symphony of international cooperation.” As expected, China led the creation of a new AI governance body last month.
It’s also about China challenging the Pax Silica coalition — which was constituted to counter their stranglehold on critical minerals.
The battleground: The US compute ecosystem is far ahead by way of physical infrastructure, and American firms dominate advanced chip design. Their policies have oscillated between banning advanced chips to restricting usage — forcing China to prioritise efficiency, smarter algorithms, novel design choices, and alternative architectures.
The US sought to increase their lead with the US CHIPS and Science Act, 2022, that committed $52 billion to domestic fabrication. China’s national chip fund of $47.5 billion followed in 2024.
China is consciously directing more investments towards state-backed firms like Huawei and Cambricon to reduce outside dependence, underscoring a fascinating contrast between private, market-led US innovation and Chinese state and local government-led execution. An important data point is the vast gap in private AI investments. Stanford’s 2026 AI Index Report puts this at $286 billion for the US, 23 times that of China in 2025. That picture is already changing, with IPOs in Hong Kong powering 40 new listings with $110 billion this JFM quarter.
What complicates the scenario is that global supply chains are deeply entwined. Chip designers, including in the US, rely on Taiwan and South Korea in terms of manufacturing, and Chinese chips are widely embedded, given their extensive footprint in general infrastructure gear.
The world’s most impactful researchers have been trained in US institutions. Yet, China leads comfortably vis-à-vis sheer research volume — ascribable in part to the increasing reverse-flow of Chinese researchers to their home country. Further, the Stanford ’26 Report says that AI scholars moving to the US has dropped precipitously — 89% since 2017. Consequently, China’s 38,000 generative AI inventions are now about six times that of the US.
The Chinese government’s emphasis on maintaining core manufacturing dominance is paying off. Tellingly, China has nearly nine times the US volume of industrial robot installations.
China has quietly and deliberately overinvested in its electricity infrastructure over decades, while AI’s insatiable power demand is outpacing the grid’s stretched development cycles in the US. Goldman Sachs suggests that this could stymie near-term US AI growth.
However, the US leads in AI adoption to accelerate services, whether it be in healthcare or mobility, which reflects the superiority of the country in connecting algorithms to capital-intensive physical networks — like roads, vehicles, warehouses, airports, maps, sensors, and the agencies involved. This has resulted in higher productivity, greater personalisation, and a reinforcing improvement loop.
How do their AI models stack up? On July 31, Chinese AI startup Moonshot released Kimi K3. It has 2.8 trillion parameters, making it the world’s largest open-weight AI system, which is fast catching up with the most advanced US systems. That announcement shook the markets (as always happens when Chinese models threaten US tech leadership). The Dow Jones Industrial Average went down 40 points, the Nasdaq Composite went 1.7% lower, and the S&P 500 dropped 0.8%.
Moonshot was founded in 2023 by Yang Zhelin, 34, who graduated from Tsinghua University and earned his PhD from Carnegie Mellon, before returning to China.
The US government is reportedly investigating if Moonshot accessed advanced chips illegally to train their models — while commentators in China say that this points towards the reality that AI models are influenced by much more than access to advanced chips. And that prior Chinese advancements in semiconductors and EVs have faced the same negative reactions over IP.
Some of the US actions may be to slow the adoption of Chinese AI within the US.
Parameters, weights, and the open-close debate: In the context of AI, “parameters” are settings that determine response to prompts. A large parametric count is reflective of huge compute investments. “Open weight” implies broader disclosure, which allows modifications of parameters. “Open-source” systems publicly share the code, allowing modifications to the model.
US companies have generally not followed the path of open models, in contrast to the Chinese, reinforcing the debatable view that Beijing supports wider AI diffusion and need-based customisation at lower costs.
The in-betweeners: The battle of AI hegemons is creating a new geopolitical layer. India seeks strategic autonomy and aims to become the AI power of the Global South. Europe want to build sovereignty and preclude overdependence on the US or China. Japan wants to turn its industrial strength into AI advantage, leveraging semiconductors and robotics. Singapore, which has led the way in terms of infrastructure, governance, and adoption, aspires to become Asia’s trusted AI hub. Israel wants to translate its defence and cyber advantage to dominate high-value AI applications.
Now most countries aren’t dwelling upon their own success, but on what happens if someone else wins. Deep down, nobody wants a hegemon to decide their future.
The author is the Founder of ThinkStreet.
Disclaimer: The views expressed are the author’s own and do not reflect the official policy or position of Financial Express.