Something important is changing in the American debate over artificial intelligence and the enormous physical infrastructure required to support it.

The Washington Times is reporting directly on an uncomfortable fact that Americans need to understand: our technological adversaries are not competing with the United States only by developing better technologies, factories, energy systems and artificial intelligence. There is now evidence that foreign actors are also attempting to influence American public opinion about the infrastructure upon which our technological advantages depend.

That makes the September 2, 2026 Washington Times article reporting Commerce Secretary Howard Lutnick’s comments particularly important.

Washington Times article:
https://www.washingtontimes.com/news/2026/sep/2/commerce-secretary-howard-lutnick-says-data-center-protesters-falling/ 

Lutnick’s larger argument is that opposition to American data-center construction is being exploited by foreign interests attempting to weaken America’s position in the global AI competition.

That allegation should not be interpreted to mean that Americans who question a data-center project are agents of a foreign government. They are not. Communities have every right to demand answers about electricity, water, noise, pollution, infrastructure costs, land use and economic benefits.

The important point is different.

There is documented evidence that foreign actors have recognized these legitimate disagreements as an opportunity to influence American public opinion and potentially slow the construction of infrastructure supporting one of America’s most important technological advantages.

The Foreign Influence Operation Is Real

In June 2026, OpenAI disclosed that it had disrupted ChatGPT accounts likely originating in China that were being used to support apparent covert influence operations targeting American debates about AI and technology policy.

One operation was appropriately named “Data Center Bandwagon.”

The operators generated social-media comments and images arguing that American AI data-center construction was increasing electricity prices for ordinary families. The material was then distributed through accounts presenting themselves as Americans.

OpenAI emphasized something equally important: it found no evidence that the operation had achieved meaningful influence beyond its own activity. Nor does the existence of the operation mean that legitimate American concerns about electricity prices or data centers originated in China.

The significance is that a likely PRC-origin influence operation deliberately identified an existing American disagreement surrounding strategically important technological infrastructure and attempted to amplify it.

That is exactly how an effective influence operation can work.

It does not necessarily invent the disagreement.

It identifies an existing disagreement, introduces or amplifies emotionally powerful narratives, conceals the actual source of the messages and attempts to make those arguments appear to originate organically from within the targeted population.

Americans therefore need to distinguish between legitimate criticism of a particular data-center project and deliberate amplification intended to persuade the United States to constrain its own technological infrastructure.

The Data-Center Debate Is Missing an Important Fact: The Technology Is Changing Rapidly

There is another major problem with much of the public discussion.

It frequently treats the data center as though it were a technologically static object.

It isn’t.

A data center designed fifteen years ago, a hyperscale cloud facility designed five years ago and a high-density AI facility being engineered today can represent substantially different generations of technology.

The industry has learned from problems involving water consumption, electrical efficiency, cooling, noise, backup generation, grid interconnection and infrastructure-cost allocation.

More importantly, enormous economic incentives now exist to solve these problems.

Electricity is a major operating cost. Water availability can constrain siting. Excessive noise creates community opposition. Slow utility interconnections can delay facilities for years. Inefficient cooling wastes energy. Shifting infrastructure costs onto other customers creates regulatory and political opposition.

Consequently, solving these problems isn’t simply environmental responsibility.

It is increasingly an engineering and economic requirement for building competitive AI infrastructure.

Water Consumption Is a Good Example of the Change

Older discussions frequently assume that a data center necessarily requires enormous quantities of continuously consumed cooling water.

Some facilities certainly do.

Traditional evaporative cooling can consume substantial quantities of water because heat is ultimately rejected through evaporation.

But that is not the only architecture available.

The Department of Energy reports that many modern facilities use closed-loop cooling systems that recirculate water rather than continually consuming and discharging it. DOE is also supporting advanced cooling technologies intended to reduce both water and energy requirements.

The change becomes even more significant with high-density AI computing.

Traditional air cooling becomes increasingly difficult as rack power densities increase. Direct-to-chip liquid cooling, immersion technologies, coolant distribution units, closed loops and advanced heat-rejection technologies are therefore receiving enormous engineering attention.

DOE’s COOLERCHIPS 1.5 program illustrates where the technology is going. The program is developing and validating advanced cooling systems capable of handling AI heat loads approaching one megawatt per rack, including technologies intended to operate with no water consumption for cooling.

That is not evidence that every new data center consumes no water.

It demonstrates something more important:

the technological trajectory is toward dramatically more capable cooling with substantially improved energy and water efficiency.

Lawrence Berkeley National Laboratory research reinforces the importance of design choices. Its analysis found enormous variation in workload-level water consumption depending upon server efficiency, utilization, cooling architecture, electricity source, climate, infrastructure efficiency and equipment refresh cycles.

Consequently, statements such as “data centers use enormous amounts of water” are becoming increasingly inadequate.

The technically meaningful questions are:

Which data center? Which cooling architecture? Which generation of equipment? Which climate? Which electricity source? Which workload? And what is its measured water-use effectiveness?

Electricity Is Undergoing the Same Transformation

The same evolution is occurring with electricity.

The first generation of enormous AI facilities frequently approached the electrical grid essentially as exceptionally large customers.

That model is becoming increasingly difficult.

AI campuses can require hundreds of megawatts, and proposed facilities are moving toward gigawatt-scale electrical requirements. Waiting years for utilities to build generation and transmission capacity can become a direct constraint on AI deployment.

That creates a powerful economic incentive to bring energy generation closer to the computing load.

The result is growing interest in behind-the-meter generation, microgrids, battery storage, natural-gas generation, fuel cells and eventually advanced nuclear generation, depending upon the particular project.

DOE projects already demonstrate this transition.

In April 2026, DOE described a Data Center Flexibility as a Grid Enhancing Technology project involving microgrids at major data-center sites in Virginia and South Carolina. The project includes increased onsite generation and battery energy storage.

Even more dramatically, in July the National Nuclear Security Administration selected Amentum to enter negotiations concerning development of a proposed one-gigawatt AI data center paired with dedicated onsite energy generation at the Savannah River Site in South Carolina.

That begins to look less like a conventional commercial building connected to the electric grid and more like an integrated computing-and-energy industrial complex.

Generating Your Own Electricity Does Not Eliminate Responsibility

Dedicated generation can solve one problem while creating others.

A data center operating hundreds of megawatts of natural-gas generation, for example, has effectively incorporated a substantial power-generating facility into the project.

That means communities should evaluate it accordingly.

Air emissions must be permitted and monitored.

Turbines, reciprocating engines, cooling equipment and backup generators must be acoustically controlled.

Noise should be measured at property boundaries and sensitive receptors.

Fuel infrastructure must meet appropriate safety requirements.

Emissions-control equipment must operate as required.

Emergency-generation systems need appropriate operating limitations.

The objective should therefore not be to pretend that onsite generation has no environmental consequences.

The objective should be to identify those consequences, engineer controls for them, measure performance and enforce the operating requirements.

That is how American technological development should work.

Data Centers Should Increasingly Pay Their Own Way

Another legitimate concern has been whether ordinary electricity customers could end up paying for transmission, substations, generation or other infrastructure constructed primarily to serve enormous new data-center loads.

That problem is also being addressed.

In June 2026, the Federal Energy Regulatory Commission took major action involving all six regional transmission organizations under its jurisdiction. FERC directed them to justify or reform rules governing how data centers and other very large electricity users connect to the grid.

One of the central objectives is protecting ordinary ratepayers.

FERC Commissioner David Rosner specifically described Cost Recovery Agreements intended to make large loads responsible for the infrastructure costs incurred to serve them—even when a proposed large customer subsequently fails to materialize.

That is exactly the direction public policy should move.

If a 500-megawatt or one-gigawatt computing campus requires specialized infrastructure, the economic model should increasingly ensure that the project bears the appropriate incremental costs rather than quietly transferring them to residential customers.

The solution to yesterday’s bad infrastructure economics is therefore not necessarily stopping tomorrow’s data center.

The solution is fixing the economics.

The Same Principle Applies to Noise

Noise provides another example.

Large numbers of cooling fans, chillers, pumps, transformers, generators and turbines can create significant acoustic problems when facilities are badly designed or located too close to homes.

Again, the answer should not be denial.

The answer is engineering.

Modern facilities can employ equipment selection, acoustic enclosures, silencers, barriers, building orientation, setbacks, vibration isolation and operating controls. Communities can require preconstruction acoustic modeling, property-line standards, independent measurements and continuing monitoring.

If a facility incorporates its own generating station, the generating equipment should be treated as industrial equipment and appropriately isolated and mitigated.

The principle is straightforward:

identify the externality, engineer it down, measure it and correct it if the engineering does not perform as promised.

This Is Why Comparing Tomorrow’s Data Centers With Yesterday’s Mistakes Is Misleading

America has made mistakes during previous waves of infrastructure construction.

We should learn from them.

But learning from mistakes is fundamentally different from using those mistakes to prevent technological development.

The next generation of AI infrastructure is increasingly being designed around challenges the previous generation exposed:

less water consumption;

more efficient cooling;

greater computational output per unit of energy;

higher-density computing;

better power management;

onsite generation and microgrids where appropriate;

battery storage and load flexibility;

better grid-interconnection rules;

greater responsibility for infrastructure costs;

and stronger requirements for noise, emissions and environmental performance.

The transition is incomplete. Not every proposed project incorporates every best practice. Some poorly designed projects will undoubtedly still be proposed.

But the direction of development matters enormously.

Data-center technology is getting better rapidly.

And AI Itself Will Accelerate the Improvement

There is an important positive feedback mechanism here.

AI data centers provide the computing infrastructure required to develop increasingly capable artificial intelligence.

But increasingly capable AI can then help engineers design better data centers.

AI can assist in optimizing cooling systems, electrical distribution, workload scheduling, server utilization, power conversion, predictive maintenance, battery operation, microgrid dispatch, building design and even the development of new cooling fluids, semiconductor materials and energy technologies.

The infrastructure helps build better AI.

Better AI helps build better infrastructure.

That creates a potentially powerful cycle of technological advancement.

The Larger Competition Is Much Bigger Than Data Centers

This is why the Washington Times story should be understood in a much larger context.

The United States remains an extraordinary technological power and currently possesses some of the world’s most important advantages in frontier artificial intelligence, advanced computing, software, cloud infrastructure, semiconductor design, scientific research, entrepreneurship and capital formation.

But America does not lead every important technological category.

China has established formidable positions in advanced manufacturing, batteries, solar manufacturing, electric vehicles, critical-material processing and several other industrial technologies. It is rapidly expanding nuclear-energy capacity and competing aggressively in robotics, aerospace, telecommunications, biotechnology, quantum technologies and artificial intelligence.

The appropriate American response is neither complacency nor panic.

It is innovation.

AI Is Not Merely Another Technology Sector

Artificial intelligence is particularly important because it can amplify progress throughout the technological system.

AI can help scientists search chemical and biological spaces that humans could never examine manually.

It can help materials scientists discover new compounds.

It can help engineers optimize aircraft, automobiles, factories, power plants, electrical grids and semiconductor designs.

It can accelerate pharmaceutical research.

It can improve robotics.

It can optimize manufacturing.

It can assist with nuclear engineering, battery chemistry, logistics, transportation and infrastructure.

AI therefore increasingly participates throughout the innovation cycle:

discover → understand → invent → engineer → test → manufacture → commercialize → improve.

This is why America’s AI infrastructure matters far beyond the AI industry.

America Must Compete Across the Entire Technology Spectrum

America currently possesses extraordinary technological advantages, but some competitors have surpassed us in selected technologies and industrial capabilities.

Our objective should therefore be broader than simply “winning AI.”

America should work aggressively to establish or regain leadership across information technology, communications, advanced materials, energy, biotechnology, medicine, manufacturing, robotics, transportation, aerospace and the technologies that connect them.

Where we lead, we must continue innovating.

Where we have fallen behind, we must understand why and regain competitiveness.

Where legitimate infrastructure problems exist, we must solve them rather than pretending they do not exist.

And where yesterday’s technologies created legitimate community problems, we should demand that tomorrow’s technologies be better.

That is precisely what is happening in significant parts of the data-center industry today.

The Information War Is Therefore Part of the Technology War

That brings us back to the central importance of the Washington Times story.

A technological competitor does not necessarily have to defeat an American technology in the laboratory if America can be persuaded not to deploy it.

It does not necessarily have to destroy American computing infrastructure if political paralysis prevents us from constructing it.

And foreign influence operations do not have to invent false controversies from nothing.

They can exploit genuine American disagreements and amplify the most frightening interpretations.

That is why Americans should approach claims about AI infrastructure with the same discipline that engineers apply to technological problems:

What are the facts?

What technology is actually being proposed?

What generation of technology is it?

What are its measured water, energy, noise and emissions characteristics?

Who pays for the infrastructure?

What engineering controls are required?

How will performance be independently measured?

And who benefits strategically if America simply decides not to build it?

America does not have to choose between technological leadership and responsible development.

We can demand both.

Build It Better — and Keep America Ahead

The strongest answer to legitimate criticism of AI infrastructure is therefore not, “Don’t worry about it.”

It is:

We heard the problem. We measured it. We learned from it. We engineered a better solution. And the next generation will be better still.

That is one of the fundamental characteristics of technological progress.

The United States currently possesses some of the world’s most consequential technological advantages. But maintaining them will require abundant energy, advanced manufacturing, semiconductor capacity, computing infrastructure, scientific research and the freedom and capital to turn discoveries into commercial technologies.

Artificial intelligence can amplify every part of that process.

It can help Americans research faster, discover faster, invent faster, engineer better, manufacture more efficiently and commercialize new technologies faster.

Our competitors understand the strategic importance of that capability.

Americans need to understand it too.

The answer to yesterday’s mistakes is not to stop building tomorrow.

The answer is to build tomorrow better.

And increasingly, that is exactly what America is doing.

References

1. Howell Jr., Tom. “Commerce Secretary Howard Lutnick Says Data Center Protesters Falling Prey to Foreign Influence.” The Washington Times, September 2, 2026.
   https://www.washingtontimes.com/news/2026/sep/2/commerce-secretary-howard-lutnick-says-data-center-protesters-falling/
   The article reports Secretary Lutnick’s argument concerning foreign attempts to influence American opposition to data-center construction.
2. OpenAI. “PRC-Linked Influence Operations Are Targeting AI Debates in the US.” June 10, 2026.
   https://openai.com/index/prc-linked-influence-operations-ai-debates/
   OpenAI documents likely PRC-origin accounts used in apparent covert influence operations, including the “Data Center Bandwagon” campaign targeting American concerns about AI infrastructure and electricity prices.
3. OpenAI. “Data Center Bandwagon Campaign: US-Targeted Influence Activity.” June 2026.
   https://openai.com/index/disrupting-malicious-uses-of-ai-data-center-bandwagon/
   This case study provides additional technical and investigative information concerning the likely PRC-origin influence operation aimed specifically at the American data-center debate.
4. U.S. Department of Energy, Federal Energy Management Program. “Cooling Water Efficiency Opportunities for Federal Data Centers.”
   https://www.energy.gov/cmei/femp/cooling-water-efficiency-opportunities-federal-data-centers
   DOE describes advanced cooling and water-efficiency approaches, including direct liquid cooling, recirculating systems and engineering techniques capable of reducing both energy and water requirements.
5. U.S. Department of Energy, ARPA-E. “COOLERCHIPS 1.5 — Cooling Operations Optimized for Leaps in Energy, Reliability, and Carbon Hyperefficiency for Information Processing Systems.” August 26, 2026.
   https://www.energy.gov/nepa/articles/cx-271071-cooling-operations-optimized-leaps-energy-reliability-and-carbon
   DOE’s current program is developing advanced cooling technologies for extremely high-density AI computing, including systems intended to handle approximately one megawatt per rack while moving toward water-free cooling.
6. Lei, Nuoa, Jun Lu, Arman Shehabi, and Eric R. Masanet. “The Water Use of Data Center Workloads: A Review and Assessment of Key Determinants.” Resources, Conservation and Recycling, Vol. 219, 2025, 108310. Lawrence Berkeley National Laboratory.
   https://eta-publications.lbl.gov/publications/water-use-data-center-workloads
   The research demonstrates that workload-level water use can vary by more than four orders of magnitude and identifies server efficiency, cooling technology, electrical generation, utilization, climate and equipment refresh cycles as important determinants.
7. Federal Energy Regulatory Commission. “FERC Launches Aggressive Targeted Action to Speed Large Load Integration.” June 18, 2026.
   https://www.ferc.gov/news-events/news/ferc-launches-aggressive-targeted-action-speed-large-load-integration
   FERC’s action addresses the integration of data centers and other very large electrical loads while explicitly seeking protections for ordinary electricity customers.
8. Rosner, David. “Commissioner Rosner’s Remarks on the Large Load Show Cause Orders.” Federal Energy Regulatory Commission, June 18, 2026.
   https://www.ferc.gov/news-events/news/commissioner-rosners-remarks-large-load-show-cause-orders-e-7-e-12-june-18-2026
   Rosner explains FERC’s proposed Cost Recovery Agreements designed to prevent infrastructure costs created by large loads from being unfairly shifted to residential customers.
9. U.S. Department of Energy. “Data Center Flexibility as a Grid Enhancing Technology.” April 22, 2026.
   https://www.energy.gov/nepa/articles/cx-035792-data-center-flexibility-grid-enhancing-technology
   The DOE project demonstrates the emerging integration of data centers with onsite generation, microgrids and battery energy-storage systems.
10. National Nuclear Security Administration, U.S. Department of Energy. “NNSA Selects Amentum for AI Data Center and Energy Project at Savannah River Site.” July 20, 2026.
    https://www.energy.gov/nnsa/articles/nnsa-selects-amentum-ai-data-center-and-energy-project-savannah-river-site
    The proposed project would pair a one-gigawatt AI data center with dedicated onsite energy generation, illustrating the movement toward integrated computing-and-energy infrastructure.