AI data centers in America are key to winning the China AI arms race

Reindustrialization is alive in America. Yet the recent backlash on artificial intelligence (AI) data centers reveals we have not learned from our offshoring mistakes: creating dependencies that will not be easily reversed.

Whether or not AI data centers are built in America, AI is here to stay. Either we choose to shape the trajectory of AI technology by embracing the AI data center buildout or cede this advantage to other nations that are eager to lead the AI race. China, for example, announced nearly $300 billion in data center investment through 2030, not including private sector spending. India is offering a 20-year tax holiday for foreign cloud operators and the Republic of Korea is offering regulatory relief. Meanwhile, many Americans are calling for moratoriums.

The national security consequences of who hosts AI data centers go beyond competition with China. If we do not build AI data centers in the United States, we risk everything that is inside the data centers: the data, compute, and intellectual property (IP) that drive innovation, protect privacy, and enable sovereignty. We will also forfeit our ability to set the security standards for this critical infrastructure.

LIZ PEEK: CHINA IS CHEERING AS THE LEFT BLOCKS AMERICA’S AI INFRASTRUCTURE

If AI data centers are the engines of AI innovation, data is the fuel. A significant amount of data is required to make AI useful, all of which must pass through AI data centers. Data, therefore, must be handled carefully — especially for privacy reasons — and viewed as a national asset.

Adversaries will view data routed through AI data centers as an opportunity to advance their own AI ambitions. Data may be collected for espionage purposes or to accelerate adversaries’ own AI models, as we have seen China do time and again in cyberspace and through adversarial distillation. Adversaries may also conduct adversarial attacks during model training, which could, without our knowledge, degrade and manipulate data that determines the performance of AI models. Either way, insecure or compromised data impacts everyone — from large businesses to individuals — and exposes potentially sensitive information. Then there is compute. America currently hosts 75% of global high performance compute, making it a true strategic advantage.

CHINA-LINKED HACKERS INFILTRATED US GOVERNMENT NETWORKS BEFORE FBI TAKEDOWN

Strong export controls on advanced compute enable us to widen America’s AI lead against adversaries and work with allies and partners to shape the trajectory of AI. Our commitment to increased domestic semiconductor production through implementation of the CHIPS and Science Act is reinforcing our lead with new U.S. capacity. But where do chips go once we have them? Many of them go to AI data centers.

By sending AI data centers overseas, the chips will follow — and potentially the businesses that supply them. Re-offshoring will have a detrimental impact on U.S. resiliency and the U.S. economy, which has prepared for the reindustrialization of semiconductor manufacturing for the past four years.

CHINA IS BUILDING AN AI WAR MACHINE. WASHINGTON MUST WAKE UP BEFORE IT’S TOO LATE

Today, China’s semiconductors are a few generations behind the leading-edge due to U.S. export controls on high-end chips and equipment. But if chips are increasingly housed overseas where we cannot enforce security standards, we may increase the ease of remote access instead of closing this loophole, further enabling China to accelerate its own AI development.

Relying on overseas AI data centers will also kneecap us in the event of a global crisis because other nations–not the United States–will decide where to surge compute. There is no guarantee that America will receive any of that capacity, but what we do receive will be slow because it will not be on our shores.

Compute capacity is already severely constrained and will only become more limited as demand grows for AI tools. This will impact our military, which already faces a potential compute bottleneck, and our innovators powering new breakthroughs, including researchers, students, and startups. Having AI data centers in America ensures we can comfortably prioritize access during a crisis while continuing to foster our innovation economy and technical workforce.

Tying together data and compute is AI IP, which includes the AI models and model weights that have enabled America to stay at the frontier. IP is already a well-known target for the Chinese Communist Party (CCP), and this extends to AI. China has already engaged in IP theft through adversarial distillation to compete with proprietary, American frontier models. Sending AI data centers overseas is like handing IP over to the CCP, which they will continue to use for their own market dominance and, presumably, their military advantage. Luckily, America currently hosts the most AI data centers. However, given about 60% of announced projects for this year have yet to begin construction, this lead cannot be taken for granted. As companies and policymakers seize this moment to reframe the narrative about resource use and benefit to local communities, three immediate actions can build public confidence and ensure America’s security interests.

CHINESE BOT FARM CAUGHT RED-HANDED TRYING TO INFLUENCE HEATED AI DATA CENTER DEBATE

First, America should leverage federal land to start rapid AI data center buildout. In July 2025, President Trump signed an Executive Order to open up federal land and streamline permitting for commercial AI data centers. By using this authority, America can ensure AI is viewed as a common good instead of a local burden. It also puts the responsibility on the U.S. government, rather than local communities, to work with data center developers to scale AI using modest resources.

Second, the Commerce Department should leverage the AI Exports Program to promote secure, global AI data center buildout with allies and partners. Allies and partners need AI capacity as much as the United States, and we should help enable their success. Standards for cyber and physical security of AI data centers are vital to securing and shaping AI use, but are still in development. The AI Exports Program, which aims to share the American tech stack with allies and partners, can be the enforcement mechanism to standardize safe and secure AI data center operations overseas. Through this program, we can mitigate the impact of insider threats and natural disasters; failures of interdependencies, such as water and energy; and compromise of data, compute, and IP. By setting these expectations through an established program, companies can confidently build and other countries will know how to leverage the American tech stack for their own advantage, reinforcing it as the secure alternative to China’s tech stack.

CLICK HERE FOR MORE FOX NEWS OPINION

Third, we must save room for AI data center improvements. Although we do not want to stall AI data center buildout, which erodes their value, it is worth asking if the pace of this buildout is necessary to stay ahead in AI. Not every data center proposed is expected to be built, but estimates project that we could reach $7 trillion in global AI data center investment by 2030. The domino effect of more compute is greater opportunity to innovate better methods to power AI that may not require a massive data center, or to improve AI models so that they do not need as much power. We must give innovators time to make these breakthroughs, and us the flexibility to adapt.

CLICK HERE TO DOWNLOAD THE FOX NEWS APP

America is poised to lead or lose the AI industrial revolution — a physical representation of its security interests. It must choose to lead by embracing AI data center buildout in the United States.