The Hidden Front in the US–China AI Race
Udumbara
The United States and China are locked in a race for supremacy in artificial intelligence (AI), the driving technology behind the next industrial revolution.
With sweeping impacts across national security, economics, and civil liberties, the U.S.–China AI competition may decide whether we ultimately live in a democratic or authoritarian world.
Industry experts are sounding the alarm.
Levesque warned that this is no longer just a government challenge, but also “an American problem,” and called for a whole-of-nation response to China’s AI surge.
That sense of urgency permeated the expo, with panelists underscoring the U.S.–China AI race while discussing various topics covering defense, biotechnology, semiconductors, and human rights.
One defense contractor displayed a pair of jars labeled “Pentagon” and “Beijing,” inviting attendees to vote with ping-pong balls on who held the superior AI arsenal. The result was a tie.
DeepSeek’s breakthrough in January was considered a “Sputnik moment” for many experts in the AI field.
The Chinese company’s R1, a ChatGPT-like AI model, achieved performance comparable to that of American platforms by utilizing slower chips. Months later, China hosted the world’s first hybrid half-marathon, featuring both human and humanoid robot runners in Beijing, as well as a humanoid robot kickboxing competition in Hangzhou.
While the general public may evaluate AI leadership based on new and flashy technology, experts point to a different set of keys to winning the AI race: AI global infrastructure and adaptive government policies.
“The thing you want to measure is global market share,” James Lewis, an expert and distinguished fellow at the think tank Center for European Policy Analysis, told The Epoch Times. “Is the U.S. building the global infrastructure for artificial intelligence, or are we letting someone else do it?”
AI infrastructure encompasses the internet, satellites, undersea cables, and many other components. Lewis said the core infrastructure consists of graphics processing units (GPUs), or advanced chips, and data centers, which store massive data and use them to train AI models.
Last month, President Donald Trump visited countries in the Gulf region and secured AI investments with Saudi Arabia and the United Arab Emirates, worth $600 billion and $200 billion, respectively. The United States will allow Saudi Arabia to import Nvidia’s advanced chips for the data centers to be built there.
Lewis welcomed the change and said the deals with Gulf countries would put the United States ahead in the AI race.
In his view, selling high-end chips doesn’t risk the United States losing AI leadership, but selling equipment to make those chips does. Therefore, he said, restricting other countries’ access to U.S. AI chips and models will simply drive them to use China’s instead.
Gregory C. Allen, senior adviser of the Wadhwani AI Center at the Center for Strategic and International Studies (CSIS), also highlighted the strategic value of massive data centers that the United States will build in the Middle East.
He told The Epoch Times that while current facilities operate with up to about 200,000 AI chips, future developments aim to scale up to the millions—a leap that could drive the next major scientific advances.
U.S. government policies also play a vital role in the AI race.
Anne Neuberger, a former national security official in the Biden administration, told The Epoch Times that it’s crucial “to think about how we adapt our policies in the most critical fields for our economy and our security to use AI quickly and in a way that’s game-changing.”
.
A vendor organizes an ad-hoc survey about "who has the superior AI arsenal" and asks visitors to pick "the Pentagon" or "Beijing" at the AI+ Expo hosted by Special Competitive Studies Project in Washington on June 3, 2025. Terri Wu/The Epoch Times
.
Years in the Making
Although AI has only recently gained widespread adoption among the general public, China began its push for AI supremacy more than a decade ago.It’s unclear how much of China’s AI infrastructure—specifically, its data centers and models—has been adopted in other countries. However, the regime has positioned its AI infrastructure as a cost-effective solution for the Global South, including the Middle East, particularly in BRI countries.
More Than Just Technology
The AI race not only determines whose chips and models will be used, but also involves values, data privacy, and access to information.Sen. Pete Ricketts (R-Neb.) said at the AI+ Expo that a Chinese victory would mean the regime is in control of Americans’ data. In addition, he pointed to China’s record of censorship and the risk that this approach could become globally normalized.
“If they’re in charge of the stack, they’re going to censor their enemies—that means us,” he said.
“So you’re not going to be able to get the right information or true information about what’s going on in the world or what real answers are,” he added, alluding to the potential erasure of history, such as the Tiananmen Square Massacre in June 1989.
The censorship in Chinese AI platforms, according to Ricketts, is why the importance of America’s victory in the AI race goes beyond foreign relations; it matters to people in his home state of Nebraska.
There is more at stake than data privacy and free access to information, according to Neuberger.
She warned at the expo that global adoption of China’s facial recognition models could facilitate “Chinese pursuit of U.S. human assets around the world.”
.
So Who’s Leading?
Artificial intelligence is no longer a future concept. It’s here.If U.S. AI products are not the cheaper option, they need to be the one with better technology if they are to compete with China’s products.
The quality of U.S. products is better, said Lewis, who rated the United States as leading in 65 percent of all AI areas and China leading in 35 percent.
Despite export controls, breakthroughs like China’s DeepSeek AI could quickly narrow the gap, he added.
A common approach is advanced packaging, which involves stacking slower chips together to create a package that functions as a faster unit.
The United States leads in terms of fundamental research and application, while China excels in model performance benchmarks and publication of AI research papers, according to Ethan Tu, founder of Taiwan AI Labs and a veteran practitioner in the AI field.
“China is good at solving a known issue,” Tu told The Epoch Times. “They are good at it like a student is good at taking a test.”
It’s unclear whether DeepSeek used advanced packaging of the slower Nvidia chips that complied with U.S. export controls. However, according to Allen and Tu, DeepSeek likely trained its model with OpenAI data by breaching OpenAI’s terms of service.
The Chinese platform used a technique called distilling—training the DeepSeek-R1 model on a smaller scope of capabilities—to achieve comparable results to OpenAI in answering the performance benchmarking questions.
According to Tu, China’s good test results are not the same as taking the lead in technology; the United States still excels at defining new problems and providing answers.
However, there’s one area where China is leading, according to a deep-learning expert at one of the AI labs at the Department of Energy, who spoke to The Epoch Times on condition of anonymity because he’s not authorized to comment publicly.
That area is computer vision technology, which is used in surveillance.
CSIS’s Allen revealed a chilling detail, claiming that ASML, a Dutch company that manufactures the most advanced chip-making equipment, is China’s top target for cyberespionage.
“The CEO of ASML says they have to double their cybersecurity budget literally every year,” Allen said at the AI+ Expo. “They would increase it faster, but that’s just the pace at which they can increase their hiring. This is the No. 1 espionage target.”
Allen told The Epoch Times that China will continue to try to steal U.S. technology. That includes espionage attempts on the most advanced U.S. labs.
The Road Ahead
Last month, the U.S. Commerce Department reminded the world of its serious commitment to leading the global AI infrastructure. It issued a guidance stating that using Huawei’s AI chips “anywhere in the world” violates U.S. export laws.The global ecosystem is beginning to split into two distinct blocs: one dominated by Western companies, the other by Chinese firms.
According to AI expert Tu, this divergence is inevitable; trust in every part of the AI supply chain is essential, and a single compromised component can undermine the entire system.
“Imagine if the burgeoning AI system gets built on, let’s use it as an example, Chinese principles, which include surveillance and other things, you wouldn’t like it at all,” he said. “Nor would I.”
.
