If AI really could kill us all, why give any ground to China?

AI Competition Between the US and China Is Becoming a Test of Power and Restraint

Sandego.net – The contest to shape artificial intelligence is moving beyond the technology sector and into the center of national strategy. China has laid out an expansive plan to become the world’s leading AI power before the decade ends, while the United States is trying to preserve an early advantage that has become far less certain.

The stakes reach well beyond commercial success. Governments and technology companies are weighing AI’s potential to transform economies, military capabilities, cybersecurity and public information. At the same time, leading figures in the American industry have warned that systems more capable than their creators can safely control could pose an extreme danger to humanity.

That unresolved risk creates an unusual political dilemma. Washington wants to slow potentially dangerous development through safeguards, yet it also fears that restraint could hand a strategic opening to Beijing. China, meanwhile, sees AI as vital to national strength and to the Communist Party’s continued authority.

A narrowing technological gap

The United States once held a clearer lead in advanced AI systems, aided by major private investment, powerful semiconductor companies and extensive data-center capacity. American policy also sought to preserve that position by limiting China’s access to the most sophisticated AI chips.

Those restrictions created serious difficulties for Chinese developers, but they also encouraged a different kind of progress: making models and software more efficient with less powerful hardware. David Bellman, product lead at data intelligence company SynMax, described the response this way:

“They couldn’t catch up with Americans because they didn’t have the most advanced chips, so they made their software better and more power efficient.”

That effort produced striking results. In February 2025, DeepSeek, a Chinese AI model, was only 0.4% behind OpenAI’s leading model in benchmark testing. The difference represented a dramatic advance from earlier Chinese systems.

Since then, top Chinese and American models have remained closely matched. Stanford University’s 2026 AI Index report concluded that the performance divide between the two countries’ models is “effectively closed.”

Measuring leadership is more complicated than comparing a single benchmark, however. Andrew Chin, an AI law professor at the University of North Carolina, has pointed to continuing American advantages in frontier-model development, private capital and data-center infrastructure. China’s progress means the race is tight, but the United States still possesses important strengths across the wider AI ecosystem.

China’s state-driven strategy

China has shown no indication that it intends to concede ground. Its 2025 New Generation Artificial Intelligence Development Plan outlined a path toward AI leadership built around coordination among government agencies and the country’s party-directed technology industry.

The strategy treats artificial intelligence as part of a broader industrial push. China aims to use the technology to strengthen its economy and accelerate work in fields including batteries, robotics and automation. Its political structure can make it easier to direct major resources toward national priorities, especially when leaders view the goal as essential to the country’s future influence.

Researchers at the University of Virginia National Security Data and Policy Institute highlighted the benefits of that coordinated approach in a white paper released after the policy was introduced:

“Central planning has allowed China to rapidly develop its sovereign technology and expand its global influence.”

Yet the same system can impose limits. The University of Virginia researchers noted that politically directed AI development can encourage favoritism and bias in model outputs. China’s regulatory focus is also shaped by concerns that differ from those driving much of the safety debate in Silicon Valley.

Control is central to Beijing’s concerns

For Chinese leaders, the threat posed by AI is closely tied to political stability and the Communist Party’s control over information. Regulations have largely emphasized what AI systems may tell users, placing ideological restrictions and political oversight at the center of the framework.

That concern was made explicit in an article published Sunday by China’s State Security Minister Chen Yixin in the state-run China Cyberspace magazine. Chen argued that China must retain dominance over AI and maintain control of the technology, warning that a failure to do so could weaken the party’s authority over the population.

He also raised the possibility that the United States could use AI for cybersecurity or espionage advantages, or deploy it in an information campaign against the party. The warning illustrates how AI policy in China is intertwined with national security, propaganda and domestic governance.

A race with no settled safety rules

The wider argument over AI safety remains unsettled. Experts disagree on how likely it is that highly advanced AI could escape human control, and they also dispute what capabilities would be required for such a scenario. The uncertainty has not removed the issue from policy debates; it has made it harder to decide how aggressively governments should intervene.

The contest is often compared with the Cold War nuclear rivalry, when two adversaries accumulated devastating power while also creating agreements intended to reduce the risk of catastrophe. AI is not identical to nuclear weapons: it is developed by private firms as well as states, spreads through software, and can be applied across many civilian activities. Still, the comparison captures the pressure facing leaders who believe a technology may be both indispensable and dangerous.

American AI companies have publicly discussed voluntary limits and self-regulation, even as political leaders argue that the United States cannot afford to fall behind China. President Donald Trump’s approach reflects that competitive pressure: slowing AI development may appear risky if it is seen as helping a rival close the gap.

The central question is therefore not simply which nation builds the most capable model first. It is whether either country can pursue technological leadership while establishing credible limits on systems whose future power remains difficult to define. The answer could influence economic leadership, global security and the rules that govern one of the most consequential technologies of the century.