The U.S. Treasury secretary has just delivered one of the bluntest warnings yet about America's race with China over artificial intelligence.
Scott Bessent says the United States cannot afford to pause.
Not for a year.
Not for political convenience.
Not even because communities are increasingly angry about giant data centers being built near their homes.
His message is stark: if China pulls ahead in artificial intelligence, the consequences could overwhelm America's advantages in other areas.
“There is no day after tomorrow if China wins at this,” Bessent said during a Breitbart News event in Washington on Tuesday. He went even further, saying that if China were to pull ahead of the United States in AI, “nothing else would matter.”
Those comments reveal just how high the U.S. government now considers the AI race to be.
For Washington, artificial intelligence is no longer merely a technology story.
It is a national-security issue.
It is an economic issue.
And increasingly, it is being treated as an infrastructure race.
That last point is at the heart of Bessent's argument.
The United States cannot maintain its AI advantage without enormous data centers.
Those facilities require land, electricity, water, fiber connections and specialized hardware.
They are also becoming increasingly controversial.
Residents in parts of the United States have pushed back against large data-center developments, citing concerns about electricity demand, water consumption, noise and the local cost of infrastructure.
Bessent acknowledged that backlash.
But instead of calling for a slowdown, he criticized the technology industry for failing to explain its economic case effectively.
He gave data-center companies, hyperscalers and major AI laboratories what he described as a “D-minus” for community outreach, arguing that the industry has done a poor job communicating with residents.
That distinction is important.
Bessent is not saying the public opposition is meaningless.
He is saying the industry has mishandled it.
The message is effectively: build the data centers, but get better at explaining why they are necessary.
That puts American technology companies in a difficult position.
They need to build infrastructure at unprecedented speed.
But they also need local approval.
And those two objectives increasingly collide.
The scale of the investment explains why.
U.S. technology companies committed to more than $850 billion in data-center leases during the first quarter of 2026, according to reporting cited in connection with Bessent's comments.
That number reveals how enormous the infrastructure expansion has become.
It is no longer just a handful of Silicon Valley companies building servers.
It is an industrial buildout involving utilities, construction companies, semiconductor manufacturers, real-estate developers and energy providers.
The AI economy is creating its own physical infrastructure sector.
And Bessent believes the United States must keep expanding it.
The reason is China.
Washington is increasingly worried that Chinese companies are rapidly improving their AI capabilities.
China's advantages are different from America's.
The United States has deep pools of venture capital, a powerful semiconductor-design ecosystem and some of the world's most advanced AI laboratories.
China has enormous manufacturing capacity, an aggressive industrial-policy framework and the ability to mobilize resources around strategic technologies.
That creates a competition that extends beyond software.
It includes chips.
Data centers.
Power generation.
Networking.
Advanced manufacturing.
And access to capital.
The country that builds the most capable ecosystem may eventually gain advantages in productivity, defense and economic competitiveness.
That is why Bessent rejected calls for a temporary AI pause.
His argument is geopolitical.
Even if the United States slows down, China will not necessarily do the same.
In his view, that makes a unilateral pause potentially dangerous.
The logic is simple.
AI development compounds.
A better model can help create better software.
Better software can improve research.
Research can create new chips.
New chips can improve the next generation of models.
The cycle can become self-reinforcing.
The longer one country maintains a lead, the harder it may become for a rival to catch up.
That is why the infrastructure debate has become so politically charged.
A community may see a giant data center as a local development project.
Washington sees it as part of a global strategic competition.
Both perspectives can be true.
A data center can impose real costs on a community while also creating national economic benefits.
That creates a policy problem rather than an easy ideological answer.
The industry's responsibility, according to Bessent, is to acknowledge those local costs.
He said companies need to explain the real economic benefits of data centers instead of assuming local communities will accept them automatically.
That is especially important because some communities are not simply worried about construction disruption.
They are concerned about electricity prices.
Large AI data centers consume massive amounts of power.
In regions already struggling with grid capacity, new demand can force utilities to invest in additional generation and transmission.
The financial burden can eventually reach consumers.
That makes local resistance understandable.
At the same time, delaying infrastructure can have national consequences.
AI companies may move projects to other countries.
Investors may redirect capital.
Cloud capacity may grow more slowly.
And the United States could potentially lose part of its technological advantage.
Bessent's argument is therefore a race against time.
The challenge is not merely to build.
It is to build faster than China while maintaining enough public support to keep projects moving.
There is another geopolitical issue in his remarks.
Bessent suggested that some organized opposition to American data centers may be influenced by Chinese propaganda or “agitprop.”
That is a serious allegation and one that would require evidence on a case-by-case basis.
But it reveals how deeply the AI competition has entered the national-security conversation.
In Washington's view, even a domestic debate over where to build a data center can have geopolitical implications.
That makes the politics around AI unusual.
Most major technologies eventually become consumer debates.
AI has become a strategic contest before the infrastructure phase is even complete.
The consequences could be enormous.
If America wins, U.S. companies could dominate another generation of computing while creating productivity gains across industries.
If China wins, Beijing could gain influence over the next technological platform that shapes defense, manufacturing, medicine and communications.
That is the stakes Bessent is describing.
It also explains why the administration has been strongly supporting AI infrastructure expansion.
President Donald Trump has similarly warned that public resistance to data centers could undermine America's competitive position and hand China an advantage.
The broader strategy is becoming clear.
Build quickly.
Secure electricity.
Expand semiconductor capacity.
Attract capital.
Keep AI laboratories operating at full speed.
And prevent local opposition from becoming a national bottleneck.
But there is an important reality Washington cannot ignore.
AI infrastructure has physical limits.
Power grids take time to expand.
Semiconductor factories require years of planning.
Water resources are finite.
Communities cannot simply be ordered to accept every project.
The most successful strategy may therefore be the one that combines speed with better public engagement.
That is essentially Bessent's message.
He does not want America to choose between local communities and AI.
He wants the technology industry to convince those communities that the benefits are worth the costs.
Whether that argument succeeds will matter enormously.
The AI race is now being fought on two fronts.
One is inside laboratories, where researchers compete to build smarter models.
The other is outside them, where governments and companies compete to build the physical infrastructure those models require.
Bessent's warning makes clear which side of that equation worries Washington most.
America cannot afford to slow down.
Because in the race for AI supremacy, the Treasury secretary believes there may be no second chance.
The next few years could determine not just which country builds the best AI.
They could determine who controls the economic and strategic infrastructure of the digital age.