The "AI race": The US and China SHARE and co authored over 46,000 research papers on A.I.

This is not a "race" this is a collaboration.

This is the image being put forward by Elon Musk and the media.

ELON MUSK: "For the next few years America is likely to win the race in AI. Then it will be a function of who controls the AI chip fabrication. If more of the factories are owned by China then China will win. Right now all the chip fabs are in Taiwan. 100%. If China invades Taiwan in the near term the world will be cut off from advanced AI chips. I think it's essential for national security that we begin manufacturing our own chips in the US."

But this article tells a very different story.

https://www.thenationalnews.com/future/technology/2025/01/28/china-us-ai-race-deepseek/

Despite the talk of a deepening rivalry between the US and China in artificial intelligence, our chart paints a different picture. Far from being locked in a tech war, the two nations are closely linked in AI research, with collaboration driving much of the innovation that underpins the major leaps in AI development today.

Data from the Emerging Technology Observatory at Georgetown University reveals how intertwined these two tech powerhouses are. In the past decade, researchers in China and the US have co-written more than 46,000 papers on AI – more than any other pair of nations. The UK and Australia rank a distant second and third, respectively, on China's list of research partners.

By working together, researchers from China and the US have made progress in fields from machine learning to natural language processing, creating a shared foundation for the next wave of AI applications. Much of this has been achieved through partnerships between US universities such as Stanford and the Massachusetts Institute of Technology (MIT) and Chinese institutions including Tsinghua University.

However, this spirit of collaboration contrasts sharply with the narrative playing out in markets and among tech companies. The recent rise of Chinese AI start-up DeepSeek is a case in point.

DeepSeek made headlines this week with the launch of its R1 model, designed to rival OpenAI's GPT-4. The real shock, however, was not the model's capability, but its cost – DeepSeek says it trained R1 in just two months at a cost of $6 million. By comparison, Meta reportedly spent $60 million to develop its Llama model, while OpenAI is said to have poured more than $600 million into the training of GPT-4.

The extraordinary efficiency of DeepSeek has rattled markets and sparked concerns about whether the US is falling behind in the global AI race. Investors and policymakers have called for accelerated innovation to maintain America's edge.

Yet the data on research collaboration tells a more complex story. While there is undeniable competition in the commercial space, research partnerships between the US and China continue to thrive. The relationship between the US and China in AI is better described as a balancing act – a mix of rivalry and co-operation.