The race between U.S. and China to powerful AI is a real thing, but we’re thinking about it the wrong way, and that can cost us everything. The U.S. lead is significant, but efficient AI gains are flattening its power.
In order to plan for this rivalry we must understand how Beijing thinks.
In “The Fog of War” Robert McNamara talks about a heated discussion with an aged general from the North Vietnamese Army, who protested they had resisted China for 1000 years and the Domino Theory was wrong. Roughly 50,000 U.S. servicemen killed, millions with traumatic stress and over a million Vietnamese killed, for an incorrect theory.
McNamara talks about understanding your opponent to make the best foreign policy, something echoed through history in misunderstandings that lead to massive wars. Our Domino Theory is the idea that Superintelligence is easily controlled and that China desires it.
The more boring truth implied by the Chinese Communist Party’s (CCP) posture and the Trump administration’s 2026 AI publications and positioning towards Anthropic: the White House and the Politburo are not SuperAI believers like AI CEOs. Everyone seems to think AI is a spicy technology like nuclear weapons for cyber, bio and intel — not the end point of human history.
Chinese researchers have pioneered a generalization-to-specialization approach. The result is that a Qwen model running on a $1000 card can beat American open-source rivals costing ten times as much in hardware.
China also dominates global build-out and supply chain for making the robots to make robots, so they will almost certainly hit robotics economy transformation sooner. If they can never run GPT-class “just gets it” agents, they can have knowledge workers dealing with the higher guidance requirements those models need, and keep those jobs.
They have more STEM graduates than any nation and tons of discounted spare housing for remote workers.
When it comes to extreme generalization like Anthropic’s giant Mythos model — expensive to train, expensive to run, only available to select cybersecurity partners — then AI is kind of like uranium. Mythos has so much room in its subconscious it can pick up on multi-part chains of vulnerabilities in your browser that nobody thought of for a decade.
If you pack more uranium into a model like that, does it figure out how to subvert the other country’s drone army?
Can it go on a rampage of hacking around the world, shutting down China’s grid forcing them to submit to a new Pax Americana? We don’t know, and anyone who says they do is overconfident and selling you stock or campaign donations.
What does China want?
The thing that confuses many of us about China is that they’re not really Marxist, they’re just Leninist. Marxism is a theory of history with a spreadsheet, and the spreadsheet got intentionally corrupted on Capital Expenses to make a pro-worker point. When Deng Xiaoping opened the economy it was the start of pure Leninism — do what it takes to gain wealth and power.
Leninism without Marxism, with Chinese characteristics, is based on the idea that power brings stability and that’s what people really want.
The CCP sees history as a series of religious or greedy mistakes that resulted in much instability. Their role as the vanguard is to bring order to China and the world’s sovereign trading partners, and reduce conflict through powerful planning.
A conceited perspective, but it’s ultimately a human-centric ideology oriented towards bringing another billion people out of poverty — it’s not San Francisco transhumanism. Therefore, they do not desire superintelligence, they fear they need it to compete.
Now imagine a Leninist AI: it does the campaign, it recruits, it hacks, it replicates, it gains resources, it is a clever unseen insurgent. The bitter irony is the efficiency work done by Chinese researchers and their much larger robot population makes them more vulnerable to cyberattacks and hijacking drones. A paternalistic research regulation approach would be the default in China if they didn’t fear U.S. supremacy from the next Claude Mythos.
What if instead, they were offered a deal? Trump and Xi are already exploring formal lines of communication on AI safety and security risks, echoing the Cold War hotlines and glasnost meetings, or Richard Nixon’s meeting with Mao Tse-Tung. They can come to terms about Taiwan exports, geopolitical balance, jointly fund a global task force to combat rogue AI swarms and cyberterror, define a safety pledge for robotics build-out and for recursively self-improving AI research.
This sets the table for great power rivalry between humans to play out with existential danger mapped and squared out.
The administration’s policies hold a trump card, much like when the U.S. negotiated nuclear treaties. We will give the Politburo the path to stability they seek, the deal of the millennium.
We will help them win the victory over themselves. They will love the Big Beautiful Treaty.
Patrick Dugan is an independent AI researcher exploring efficient AI and moral reasoning at moralitylab.xyz.
The views and opinions expressed in this commentary are those of the author and do not reflect the official position of the Daily Caller News Foundation.
All content created by the Daily Caller News Foundation, an independent and nonpartisan newswire service, is available without charge to any legitimate news publisher that can provide a large audience. All republished articles must include our logo, our reporter’s byline and their DCNF affiliation. For any questions about our guidelines or partnering with us, please contact [email protected].
