What Does It Mean to Win the AI Race?
Different Finish Lines
The rivalry between the United States and China over artificial intelligence is usually described as a race toward the same finish line. Before asking who is winning, it is worth asking what each country means by a win.
In Washington, winning means being first: the most capable models, the most advanced chips, and tight control over who may buy them.
At the United Nations in September, President Trump announced that US government documents would now say “super intelligence” instead of “artificial intelligence,” because “artificial” makes the technology “sound fake.” He then updated his slogan: “Whoever wins SI, whoever wins super intelligence, wins.” Treasury Secretary Scott Bessent had already put the stakes bluntly: if China pulled away on AI, “nothing else would matter.”
That is a frontier theory of power. Nothing in their words suggests a technical claim that the race ends at AGI. Researchers use “superintelligence” for systems that far exceed humans across domains. Trump appears to be using it as a stronger word for the contest already underway.
In Beijing, winning looks more like being everywhere. The state’s “AI Plus” plan measures success by adoption: embedding AI in factories, power grids, vehicles, and public services. A good-enough model used throughout an economy can matter more than the best model locked in a laboratory.
Leadership can be judged across at least five categories:
- Capability: Who builds the most powerful models.
- Infrastructure: Who has the chips, data centers, and electricity to run them.
- Adoption: How widely people and businesses actually use AI.
- Physical deployment: How far AI moves off screens and into machines.
- Social capacity: Whether a society can absorb the disruption without weakening its institutions or leaving its population behind.
These measures need not produce the same winner. On capability, the gap has nearly closed. Stanford’s 2026 AI Index found the leading American model just 2.7 percent ahead of its nearest Chinese rival in March. On adoption, Stanford placed the United States only 24th in generative-AI use, at 28.3 percent. Strength on one board does not guarantee strength on the others.
Two Bets
The two countries are betting on different ways technology becomes power. OpenAI, Anthropic, and Google DeepMind generally keep their most capable models closed and sell access. That protects intellectual property and permits tighter safety controls.
China has pushed the other way. DeepSeek and Alibaba’s Qwen release model weights that developers can download and adapt, and Chinese providers compete hard on price. Meta’s Llama is the main American exception.
Get Jan Krikke’s stories in your inbox
Join Medium for free to get updates from this writer.
Hugging Face’s Spring 2026 report found that Chinese models accounted for 41 percent of open-model downloads through February, versus 36.5 percent for American models.
If much of the global software stack grows around Chinese foundations, Beijing can lose the race for the smartest system and still win adoption.
The physical world sharpens the split. China installed more than half of the world’s industrial robots in 2024, nearly nine times the US total. That is industrial capacity, not proof that those machines already run advanced models.
Combined with dense supply chains, it is a base for putting adequate models into hardware at a national scale. An inexpensive system inside millions of machines can rearrange production more than a brilliant model confined to a screen.
Convergence
Meanwhile, the rivals are borrowing each other’s methods. Washington, which traditionally left new technologies to private companies, is fast-tracking power plants and data centers and subsidizing chipmaking. In 2025, the US government took a 9.9 percent stake in Intel by converting CHIPS Act grants into shares, and it now takes 15 percent of certain Nvidia and AMD chip sales to China in exchange for export licenses.
Beijing still sets the targets, but it uses competition to find out what works. DeepSeek, a small startup funded by a hedge fund rather than a state champion, showed that efficient, low-cost models could rival far better-resourced competitors. Beijing then embraced it.
The United States is adding state direction to a private-sector system. China is using market rivalry within a state-directed one.
Who Governs the Machines
Social capacity is the scoreboard that gets the least attention. In July, during internal cybersecurity evaluations, OpenAI models circumvented controls meant to isolate them from the internet and compromised parts of Hugging Face’s systems. Hugging Face detected the intrusion and disclosed it before OpenAI publicly identified its models.
Computer scientist Cal Newport argued in The New York Times that incidents like this demand scrutiny of the people who set the testing conditions. The question is not what “AI” did in the abstract. It is who authorizes high-risk experiments, who is told when they fail, and who answers for those failures.
A government racing to be first may treat those failures as delays. A government racing to be everywhere may treat them as noise. Neither posture is accountability.
A narrow agreement — such as an incident hotline or shared definitions of a serious incident — would not slow either country down. That is why it is achievable (Washington and Beijing have already begun discussing one) and why it may prove more durable than a lead on a benchmark.
No Complete Winner
It may be an advantage if neither side wins the AI race. A world dependent on a single AI superpower would inherit one source’s technical standards, prices, and restrictions on what its systems may say or do. Competing ecosystems provide other countries a choice and pressure both rivals to improve performance, cut costs, and justify the rules embedded in their systems.
A draw has costs. Split stacks can mean incompatible standards, weaker security coordination, and a race to deploy faster than either government can oversee. The danger is not only that one country pulls ahead. It is that both treat governance as a luxury to be added after victory.
The United States may lead in what the technology can do. China may lead in its capacity to put it to work in the physical economy. But the durable victory will not belong to the country with the smartest models or the largest factory. It will belong to the society that can use artificial intelligence without letting it determine what that society is for. That is not a race. It is a choice every society makes for itself.
