As the US Runs the AI Race, China Plays a Different Game

If AI is the new space race, how will we know when a country has won?

After all, an August article in Asian Economic Policy Review argues that China and the United States are not even racing toward the same planet as they pursue fundamentally different AI strategies. The US is fixated on a “frontier-first” approach, prioritizing increasingly powerful foundation models and infrastructure in hopes of capturing global dominance, while China is focused on embedding the technology deeply throughout its economy.

The distinction matters for businesses and policymakers because the country that gains the most from AI may not develop the most powerful model, but could instead be the one that translates AI most effectively into productive economic activity, explains Harvard Business School Professor Meg Rithmire.

“China’s Diffusion-Forward AI Strategy: The ‘AI Race’ in Political Economic Context,” which Rithmire wrote with Harvard Kennedy School postdoctoral fellow Hao Chen, details how China’s state-directed approach to AI investing aims to improve industrial efficiency by integrating AI into manufacturing and robotics. Drawing on Chinese policy documents and registration data for AI services, as well as an analysis of humanoid robotics firm UBTECH, the researchers document the government’s expanding investment role, with nearly one in four Chinese AI services crediting some degree of state involvement in 2025, up from less than 5% in 2023.

In a conversation edited for length and clarity, Rithmire, the James E. Robison Professor of Business Administration, explores the race between China and the US to develop increasingly capable systems, including artificial general intelligence (AGI), which can outperform the “median human” in economically valuable work, as well as the broader competition over the technology stack that could shape the global economy.

Why is the current tech moment often compared to the space race?

"One interpretation is that whoever reaches AGI first is going to have enormous capabilities vis-a-vis the rest of the world. The dominant thinking right now in Washington and Beijing is about who has better capabilities of escalation dominance—meaning you can hurt me, but I can always hurt you more. I can dominate the escalation of our competition, and I then no longer fear your retaliation."

Why does it matter how advanced AI gets?

"The thinking around AGI is that there are some amazing capabilities that could allow one country to destroy the other. These are allegedly such powerful models that, if some adversary had that technology, they could use it to infiltrate systems, hack passwords, and shut down critical infrastructure. There’s a military defense logic to this thinking: If China gets these capabilities before the US, they’re going to use them to destroy other societies, so we should get them first because we won’t use them to destroy everyone else.

The way US companies and a lot of policymakers in DC have thought about it is: The biggest risk is China getting to AGI before the US does. And that’s the reason we have to focus on the quality of large language models. But what AGI really is, nobody knows. Sometimes I have this interesting debate with technology people, which is—will we even know when we have AGI?"

What else matters besides superintelligence?

"There’s another version of the AI race, which is really about whose technology stack the world is going to rely on for the next several generations. In a world where the US wins the AI race, most of the world is dependent on the US for the hardware stack, the chips, and US LLMs to build the application-layer products that are then sold into the rest of the world.

That's more of an economic power—locking the world into American technology such that American companies benefit, American markets benefit, and the world softly depends on the US for the AI stacks that are changing the way people work, receive services, buy things, and live."

How do the US and Chinese approaches to AI differ?

"The US is pouring massive amounts of capital into the tech companies that are developing LLMs, with the idea that they need massive investments to train on more data to get better.

China thinks of AI as a factor of production. Chinese models are super low-cost, and they’re OK in intelligence. Most of the world doesn’t need the most advanced model of all time to run their HR department or optimize their factory floor. They’re going to go with the model that’s 14% of the cost."

How stable are these investments?

"When we think about the way that Chinese industrial policies worked in the past, they target a sector, so much money goes into it, everyone in China starts competing fiercely to be the winners, they produce more than China will consume, and then companies around the world can’t compete with Chinese prices.

China’s got an employment problem already. It doesn’t have enough jobs for people because they’ve automated manufacturing too quickly. There are some social effects in China, just like there are some social effects here.

The US has extended so much private credit into the LLM part of the tech stack. [The question is,] are they going to make any money?"

What can the US learn from China?

"The US should look at how AI is being deployed, rather than just the LLMs themselves.

China has a huge manufacturing sector. The autonomous vehicle sector is much further ahead. If you go to a Chinese campus, there are tons of humanoid robots lined up ready to deliver food to dorm rooms."

How do US policies affect what’s happening?

"The US strategy has been to block Chinese companies and ban many technologies. The thinking is that our companies will fill the void and they’ll be better.

I don't think they're better anymore. We have to reckon with the reality that Chinese companies are offering technology products that are very low cost and high quality to most of the world. Is there a middle road between banning all Chinese technology products and opening the floodgates? It takes coherent government intervention and regulation."

Who is going to win this race?

"No one is going to win or lose. Just like the space race, someone’s going to get somewhere first—get to the moon first, deploy a satellite first, but it’s not as if someone definitively won the space race and ended the ambitions of the other country.

Competition with China and with Chinese firms on the frontier of technology is not going away. Everyone will be confronted with Chinese and American participation in these different parts of the tech stack.

If you’re in the US and you think, ‘no problem, I’m insulated from it,’ that’s not the right conclusion. The rest of the world is not blocking itself off, so having some understanding of how the dynamics work in China and what the next wave of competition is going to be is necessary."

Photo credit: Meg Rithmire