Go figure: President Trump Claims We Must Beat China in the AI Race, Then Sells Them Our Advantage.

The Hypocrisy That Defies Common Sense

It does not make sense. President Trump has repeatedly claimed that America must beat China in the AI race, yet he is allowing the sale of our key advantage, NVIDIA chips, to the very rival he says we must defeat. He uses the beat China argument to fast-track new AI data centers, and conservative talk show hosts such as Glenn Beck echo his statements as a valid reason to build even more of them. Common sense says you do not sell your advantage to the enemy.

Trump’s Own Words and Emergency Declarations

“We have to be at the top. Otherwise, China is going to take it over,” Trump said during an interview July 2 with CNBC. [1] [2]

“We need to double the energy we currently have in the United States … for AI to really be as big as we want to have it” to compete with China and other countries.” [3]

Trump has repeatedly framed the rapid permitting of data centers and cell towers, while ignoring local protests, along with co-located power plants for data centers, overriding states’ rights to regulate data centers, and broad deregulation as essential to maintaining U.S. leadership over China. For example, in remarks on AI infrastructure:

“China is a competitor… We want it to be in this country… I’m going to help a lot through emergency declarations, because we have an emergency. We have to get this stuff built.” [4]

Stop Arming the Enemy: End NVIDIA Chip Sales to China Now

If we truly do not want China to surpass us in artificial intelligence, we must stop selling them NVIDIA chips. These powerful processors are the engines that power advanced AI systems, giving whoever controls them a huge edge in technology, military strength, and economic power. By allowing sales, even with some limits, we are handing China the very tools they need to train smarter systems and close the gap on us. China already pours massive resources into AI and has shown it will use any advantage against American interests. Keeping our chips at home protects jobs, national security, and our lead in the future, instead of helping our biggest rival build up at our expense. I repeat, common sense says do not arm the competition if you want to win the race.

I will give a real-world example from the world of competitive speed skating that I grew up in. Back in the mid-1990s, the Dutch developed the clap skate (or “klapschaats”), a hinged-blade design that gave their skaters a clear seconds-long competitive edge by allowing longer, more powerful strokes on the ice. When the technology proved revolutionary, Dutch manufacturer Viking refused to sell the skates to Americans in order to preserve that advantage for their own athletes ahead of the 1998 Nagano Olympics. American skaters and officials lodged complaints with the International Skating Union, highlighting the unfair restriction of critical technology and were finally able to acquire the clap skates two weeks before the Olympics. Just as the Dutch withheld this performance-enhancing equipment from their American rivals to protect their lead, the United States must not sell advanced NVIDIA chips to China. Handing our primary strategic competitor the very tools needed to close the AI gap is not smart competition. It is self-sabotage. National security and technological superiority demand that we keep our decisive advantages at home.

Congress Also Failed to Protect Us

In a powerful Hudson Institute article titled “China’s Military Machine Shouldn’t Run on American Chips,” the authors highlight a critical missed opportunity. The proposed GAIN AI Act would have given American buyers priority access to advanced chips, ensuring U.S. companies stay ahead in the global AI race. As the piece rightly states: “There is nothing more America First than prioritizing U.S. companies over the Chinese Communist Party (CCP). Some corporations, however, are putting Beijing’s interests first.” [5]

Congress had a chance to protect America in the high-stakes AI race with a bill called the GAIN AI Act. This  legislation would have treated advanced AI chips like sensitive military equipment: it would have given Congress 30 days to review big sales to China and required the government to make sure exports didn’t hurt American companies or our national security by short-changing U.S. buyers when supplies are tight. Unfortunately, the bill was killed before it could pass. NVIDIA and other business interests lobbied hard against it, and the White House reportedly pushed back too, putting corporate profits and short-term sales ahead of long-term American security. As a result, we’re left without strong guardrails while China gains access to the powerful chips that fuel AI and modern warfare.

While the GAIN AI Act would have offered some meaningful protection, I believe we must go further and impose an outright ban on the sale of these advanced chips to the Communist Chinese Party.

According to an Atlantic Council article titled “Why Exporting Advanced Chips to China Endangers U.S. AI Leadership:”

• Allowing Nvidia to sell H200 series chips to Chinese customers appears to be a compromise move by the Trump administration to help Nvidia’s global market share without giving China access to cutting-edge Blackwell chips.
• But the move will allow China to eat into the US advantage in global computing, undermining the administration’s own goals of maintaining US global leadership in AI.

President Trump shook up the global tech race in a Truth Social post when he announced his approval for Nvidia to sell its H200 (“Hopper”) series chips to “approved customers” in China, with the United States receiving a 25 percent cut of the revenues. Trump said that Chinese President Xi Jinping “responded positively” to the decision. The real implications

The United States and China are locked in an existential race for AI supremacy. Until now, the United States’ one true advantage has been access to cutting-edge compute. [6]

The Trump Administration Is Already Shipping Advanced H200 Chips to China

The United States is indeed selling NVIDIA chips to China right now. In July 2026 a top Commerce Department official told Congress that shipments of the powerful H200 artificial intelligence chips had begun reaching customers in China after the Trump administration approved licenses for about 10 Chinese companies. These H200 chips rank among NVIDIA’s most advanced for training and running AI models. Reports indicate approvals cover major firms including units tied to ZTE, Alibaba, ByteDance, and Tencent, with potential for hundreds of thousands of chips overall.

President Trump supported the move as part of broader negotiations, sometimes linking it to revenue sharing for the U.S. government. Critics from both parties and major think tanks warn that this giveaway erodes America’s critical AI lead. China could use these advanced chips to power military AI systems, even more mass surveillance, and the training of cutting-edge models. By weakening safeguards and turning export controls into a bargaining chip, China can amass enormous total computing power and close the gap with the United States. [7]

The Revenue Excuse Crumbles Under Scrutiny

Proponents claim that allowing limited sales of certain NVIDIA chips actually benefits America more than an outright ban would. They argue that a total export cutoff to China would cripple U.S. companies like NVIDIA, destroying American jobs and eliminating billions in revenue that could otherwise fund further innovation here at home.

Really? Is this truly about protecting U.S. innovation, or is it mainly about NVIDIA raking in massive profits by selling advanced technology to the CCP? And are policymakers genuinely worried about losing jobs in the U.S. when the same AI systems, powered by these chips, are aggressively deployed to replace American workers across industries?

Let’s Finally Put to Rest the Repeated Lie that We are Behind China in AI Data Centers

Before I go any further, the oft repeated mantra that we must build more data centers to keep up with China is not only misleading but an outright falsehood.

Recent developments underscore the urgency of protecting America’s lead in the AI race.  China’s Moonshot AI just released the Kimi K3 model, which some analysts claim challenges top U.S. frontier systems like GPT-4o, Claude 3.5/Opus, Gemini 1.5, Grok-3, Llama 4, etc

The AI race is about who develops the most powerful artificial intelligence first. The winner gains enormous advantages in the economy, jobs, medicine, scientific discovery, and national defense. Superintelligence refers to the next level: AI systems far faster in solving problems than the best human experts combined. These could solve incredibly complex problems almost overnight, such as curing diseases, optimizing energy systems, and advancing technology. However, they also carry risks if developed carelessly or controlled by adversaries like the Chinese Communist Party.

NVIDIA chips are the powerful engines that make advanced AI possible. They are super-fast math processors specially designed to handle the enormous calculations needed to train and run AI models. The best NVIDIA chips give a huge advantage because they can do this work much faster than anything else currently available. The more powerful NVIDIA chips a country or company controls, the faster and smarter their AI becomes. China badly needs these top chips to keep improving its AI systems. For example, the NVIDIA chips we are currently selling them (such as the H200 series) are significantly more advanced and efficient than China’s own domestic AI chips. This allows China to develop and train stronger models like the Kimi-K3 faster than they could using only their homegrown technology. While China releases models like Kimi-K3 and makes them seem open and generous, every conversation users have with the model feeds data back to Chinese companies and ultimately the Communist Party. At the same time, the model is carefully censored to present only the information and viewpoints Beijing approves. This is not neutral technology. It is a strategic tool for gathering global intelligence while shaping narratives. Chinese AI models must follow strict government censorship guidelines. They are trained to avoid sensitive political topics (Tiananmen Square, criticism of Xi Jinping, Taiwan independence, etc.). That is why the continued export of advanced NVIDIA chips to China is so dangerous.

We do not need to flood the country with thousands more data centers to win this race. Victory depends far more on securing and prioritizing advanced chips for American use than on simply building additional facilities. While the United States maintains a commanding lead in physical data center infrastructure and total computing power, China is advancing rapidly in AI model development (the software brains). This reality makes the continued export of advanced NVIDIA chips to China even more shortsighted and dangerous. We cannot claim to be in an existential race for AI supremacy while simultaneously handing our primary strategic competitor the very tools they need to close or erase the gap.

Let’s do the math on our existing dominance in data centers:

The United States already has over 5,000 data centers (estimates range from ~4,184 to 5,427 as of early-to-mid 2026), while China has roughly 350–450.

China has 1.42 billion people  USA has 347 million people (China has ~4× more people)

USA has 5,426 data centers  China has only 449

Per person:

• 1 data center for every 64,000 Americans
• 1 data center for every 3.2 million Chinese

Simple way to say it: The average American is 50 times more likely to have a data center “nearby” than the average Chinese.

America has more facilities than the next 14–26 countries combined, depending on the dataset referred to. We are not behind. We dominate global data center capacity and power.

United States: 5,427 data centers (higher-end estimate from Cloudscene/Statista-linked sources as of late 2025/early 2026)

Running tally of the next countries (cumulative sum building toward the US total):

1. United Kingdom: 555 (cumulative: 555)
2. Germany: 523 (cumulative: 1,078)
3. France: 394 (cumulative: 1,472)
4. China: 369 (cumulative: 1,841)
5. India: 296 (cumulative: 2,137)
6. Canada: 288 (cumulative: 2,425)
7. Australia: 284 (cumulative: 2,709)
8. Japan: 257 (cumulative: 2,966)
9. Italy: 252 (cumulative: 3,218)
10. Brazil: 212 (cumulative: 3,430)
11. Spain: 209 (cumulative: 3,639)
12. Indonesia: 198 (cumulative: 3,837)
13. Netherlands: 186 (cumulative: 4,023)
14. Russia: 183 (cumulative: 4,206)
15. Ireland: 139 (cumulative: 4,345)
16. Sweden: 117 (cumulative: 4,462)
17. Switzerland: 120 (cumulative: 4,582)
18. Mexico: 173 (cumulative: 4,755)
19. Poland: 144 (cumulative: 4,899)
20. Turkey: ~100 (cumulative: 4,999)
21. South Korea: ~100 (cumulative: 5,099)
22. Denmark: ~80 (cumulative: 5,179)
23. Hong Kong: ~80 (cumulative: 5,259)
24. Singapore: 99 (cumulative: 5,358)
25. Belgium: ~80 (cumulative: 5,438)
26. Austria: ~70 (cumulative: 5,508) ← Surpasses US total here

Source for these statistics: Statista / Cloudscene Data Center Counts :full global database via Cloudscene/Data Center Map).

Critics argue that the US may be ahead in small data centers and overall numbers, but China has more large data centers with more computing power. That is also a complete fabrication.

Mega-Scale Powerhouses over 100 Megawatts: America’s 580 Giants far exceed China’s 100–150

The U.S. runs about 580 huge powerhouses over 100 megawatts that train the smartest AI models. These facilities have a whopping 53.7 gigawatts of total power, grabbing 51-54% of the world’s share. That’s roughly four times more sites than China’s 100-150 facilities and twice the raw power at 31.9 gigawatts (just 16-26% globally) This spread-out U.S. network makes AI training much more reliable, because if one center goes down or loses power, the work instantly shifts to others and nothing crashes.[1] [2]

Medium and Edge Facilities: America’s Massive Backup Network

 Medium-Scale Data Centers 10-100 megawatts: US-over 1,000 vs China-200 to 300

For general AI and cloud work the U.S. leads with over 1,000 facilities as part of its massive hyperscale network, packing more than 20 gigawatts. That is about five times the number of sites and way more power than China’s 200-300 setups with around 7 gigawatts.
This means American AI companies have many backup data centers ready to take over instantly if one gets overloaded or goes offline, so their services almost never slow down or crash. China has far fewer of these backup options, so when one of their big centers has a problem, a lot of AI work can get delayed or stuck. [3] [4]

Small and edge data centers under 10 megawatts: US-3,500 vs China-100-200

For fast, local AI (like the smart features in your phone, maps, or voice assistants), the U.S. has more than 3,500 small “neighborhood” data centers spread across the country, delivering at least 5 gigawatts of power in total.

China has only about 100–200 of these small local centers with very little power between them, because almost all of its 449 data centers are huge centralized facilities far from most users. [5]

The United States maintains a definitive edge is on aggregate computing power.

As of mid-2025, the US share of global AI computing power reached 74 percent, with China at only 14 percent. Aggregate computing power is critical for training new frontier models, supporting the widespread use of AI and new applications of the technology, and exploring new architectures and pathways toward more powerful systems. [8]

We Have Shattered AI Data Center Computing Power Since 2014

In my recent exposé, “US AI Data Centers: Enough Capacity to Watch Every Person on the Planet,” [9] I uncovered former NSA whistleblower information about the capacity of the NSA Utah “Bumblehive” data center completed in 2014:

“They (NSA/Bumblehive) would have plenty of space with five zettabytes to store at least something on the order of 100 years worth of the worldwide communications, phones and emails and stuff like that.”

I go on to explain how the AI data centers built since then are massive compared to the relatively small Bumblehive. Just two of the many examples are:

Amazon Web Services campus in New Carlisle, Indiana: This data center is already operating with 2.2 gigawatts of power. That is about 34 times more power than the old Bumblehive, which used only 65 megawatts.

The QTS data center network owned by Blackstone: This group of centers is already running at 3 gigawatts of power across its sites. That is more than 46 times the power of Bumblehive.

These already built centers are impressive, but even bigger ones are now under construction or in the planning stages. Two examples are:

Meta’s Hyperion campus in Richland Parish, Louisiana: This project is being built much larger than first planned. It will use up to 5 gigawatts of power. That is about 77 times more power than the old Bumblehive, which used only 65 megawatts. It shows a huge leap in both electricity use and AI computing power.

Delta Gigasite in Millard County, Utah: This one is planned for more than 10 gigawatts. That is over 150 times the power of Bumblehive. It will cover thousands of acres with tens of millions of square feet of buildings.

The Verdict: Stop the Hypocrisy, Halt Chip Sales, and Put America First

These projects, along with many others, show how far we have already exceeded the scale and capacity that once seemed astonishing with the 2014 Bumblehive era.

The truth is clear: the United States already dominates the world in AI data centers and computing power, yet the push for even more continues under the false claim that we are behind China. This agenda is built on misleading information. At the same time, the Trump administration is selling advanced NVIDIA chips to Chinese companies, directly handing our greatest technological advantage to our biggest rival. President Trump warns that America must stay ahead, but his actions tell a different story. You cannot claim to fight for U.S. leadership while arming the competition with the very tools they need to overtake us. It is time to stop the hypocrisy, end the chip sales to China, protect our national security, and put America first. We are already winning. Let us stop helping our adversary win at our expense.

 In Conclusion

Protecting our chip advantage and maintaining disciplined control over exports is the real key to victory, not unchecked data center sprawl that burdens American communities with higher energy costs and other impacts while enriching our strategic rival.

 

[1] https://www.stardem.com/news/data_centers/trump-ai-is-bigger-than-the-internet-u-s-needs-to-double-power-for-data/article_925d63ca-57b1-477e-b2de-a9b24f119b55.html

[2] https://www.cnbc.com/2026/07/02/cnbc-transcript-president-donald-trump-speaks-with-cnbcs-joe-kernen-today-.html

[3] https://insideclimatenews.org/news/06072026/trump-wants-to-fast-track-ai

[4] https://insideclimatenews.org/news/06072026/trump-wants-to-fast-track-ai/

[5]  https://www.hudson.org/foreign-policy/chinas-military-machine-shouldnt-run-american-chips-michael-sobolik

[6] atlanticcouncil.org/dispatches/why-exporting-advanced-chips-to-china-endangers-us-ai-leadership/

[7] https://www.bloomberg.com/news/articles/2026-07-14/small-amount-of-nvidia-ai-chips-shipped-to-china-with-us-license.

[8] https://epoch.ai/data-insights/ai-supercomputers-performance-share-by-country

[9] https://glorytogodstudio.com/commentary/us-ai-data-centers-enough-capacity-to-watch-every-person-on-the-planet/