AI Data Centers Face a Double Threat: Extreme Weather and Manipulated Public Debate

The artificial intelligence boom depends on physical infrastructure that is becoming harder to protect. Data centers need huge amounts of electricity, constant cooling, reliable water supplies, strong grid connections and uninterrupted communications. Heatwaves, droughts, floods, storms and wildfires can threaten every part of that system.

Now there is another layer.

OpenAI says suspected China-linked influence operators used ChatGPT to create content aimed at shaping American debate about data centers, electricity costs and AI infrastructure. The operation reportedly pushed messages about blackouts, rising power bills and environmental damage.

Some of those themes reflect real public concerns.

That is what makes the story important.

A misleading campaign does not mean every criticism of AI data centers is false. It means a legitimate debate can be exploited, amplified and distorted by foreign actors trying to weaken trust in American technology and institutions.

The physical risk is real.

The information risk is real too.

Data Centers Are Being Built in High-Risk Locations

New climate-risk analysis has found that a large share of global data center capacity is located in areas exposed to severe weather and long-term environmental stress.

Around 79% of global capacity reportedly sits in markets facing elevated exposure to acute hazards such as floods, damaging winds, storms or wildfires.

More than half is also located in places facing chronic risks such as extreme heat or drought.

That creates a major contradiction.

AI companies are building some of the most expensive and power-hungry infrastructure in modern history. Yet much of it is being placed in locations where climate conditions may make operations more difficult, costly and unreliable.

The industry is expanding fast because demand is high.

Site selection is often driven by:

• Available land
• Access to electricity
• Tax incentives
• Fibre connections
• Proximity to customers
• Local permitting rules
• Construction costs

Climate resilience can become one factor among many.

That may work in the short term.

It becomes dangerous over the life of a facility.

Heatwaves Put Pressure on Cooling and Power

Data centers produce large amounts of heat.

Every server, processor and graphics chip converts electricity into computing power and waste heat. That heat must be removed continuously. If cooling fails, equipment can slow down, shut off or suffer damage.

During a heatwave, the problem gets worse.

Outside air is hotter. Cooling systems work harder. Electricity demand rises because homes and businesses are also using air conditioning. Power plants and transmission lines can lose efficiency in extreme temperatures.

The data center then needs more electricity at the exact moment the grid is under the most pressure.

That is the collision.

AI workloads do not stop because the weather is hot.

Data centers are designed to operate all day, every day. Their customers expect cloud services, financial systems, communication tools and AI products to remain available.

A few hours of disruption can carry large financial and operational costs.

Cooling Is Becoming a Strategic Technology

Cooling used to be treated mainly as a building-management issue.

That is changing.

Modern AI chips can consume far more power than traditional server processors. High-density racks produce more heat in smaller spaces. Standard air cooling may struggle to keep up.

Companies are shifting toward liquid cooling, including systems that move coolant directly to chips.

These designs can improve efficiency and reduce dependence on traditional cooling towers. Some newer systems recycle liquid in closed loops and can operate at higher temperatures, cutting the need for energy-intensive chillers.

That is progress.

But it does not eliminate the wider problem.

Data centers still need electricity. Many existing facilities continue to use water. Retrofitting older buildings can be expensive. Cooling systems also depend on pumps, power supplies and mechanical equipment that can fail during severe weather.

The most advanced design is still vulnerable if the grid goes down.

Water Stress Creates Local Conflict

Water has become one of the biggest flashpoints around data centers.

Some facilities use evaporative cooling because it can consume less electricity than fully mechanical cooling. But evaporative systems can use large volumes of water.

That becomes controversial in drought-prone areas.

Residents may see a new data center receiving access to water while households, farmers and local businesses are being told to conserve.

The global total may look manageable compared with agriculture or heavy industry. The local effect can still be serious.

Water risk is about location.

A facility using a large amount of water in a wet region may create little pressure. The same demand in an arid region can become politically explosive.

That is why companies are pushing water recycling, closed-loop cooling and alternative designs.

It is also why communities are asking harder questions before approving new projects.

Severe Weather Is Already Driving Insurance Losses

Climate risk is no longer a distant modelling exercise for insurers.

Severe weather is reportedly responsible for a significant share of losses connected with data center construction projects in the United States.

Storms can damage unfinished buildings, transformers, power lines, cooling systems and construction equipment.

Flooding can delay projects for months.

Wildfires can cut transmission routes or force evacuations.

High winds can damage roofs and external systems.

The cost does not end when construction is complete.

Operators may face higher insurance premiums, stricter building requirements and more expensive backup systems. Investors may demand better climate-risk analysis before financing a project.

A facility that looks cheap to build can become expensive to protect.

Climate Risk Can Become a Financial Risk

The AI boom has triggered enormous spending on data centers.

But the value of those facilities depends on long-term reliability.

If a site faces frequent heat stress, water shortages or storm damage, operating costs rise. If insurers reduce coverage, owners carry more risk. If communities impose new restrictions, future expansion becomes harder.

Climate conditions can affect:

• Construction costs
• Insurance premiums
• Cooling expenses
• Electricity prices
• Water access
• Maintenance
• Downtime
• Asset value
• Financing terms

That makes climate resilience a business issue, not a public-relations exercise.

The smartest companies will treat weather data and climate projections as part of core investment planning.

The weakest will rely on old assumptions and hope conditions remain stable.

OpenAI Says Foreign Actors Targeted the Debate

While companies face genuine infrastructure challenges, OpenAI says suspected China-linked networks tried to manipulate the public discussion around them.

The company reported banning two clusters of ChatGPT accounts that appeared to support covert influence operations.

One effort focused on American data center construction. Operators allegedly used AI tools to draft social media posts and other content claiming that data centers were raising electricity prices, threatening blackouts and damaging communities.

OpenAI described the activity as limited in scale and impact. The accounts did not appear to generate significant public traction.

That is an important detail.

This was not evidence that the entire debate had been taken over.

It was evidence that foreign actors were testing messages inside an existing debate.

That difference matters.

Why Real Concerns Are Useful to Influence Operations

The most effective propaganda usually does not begin with a completely invented issue.

It starts with something real.

Electricity demand from data centers is rising.
Communities are worried about water use.
Some residents fear higher utility bills.
Grid operators are concerned about new large loads.
Climate risk is increasing.

An influence operation can take those concerns and remove context.

It can exaggerate uncertainty. It can make every project look dangerous. It can present worst-case outcomes as certain. It can create the appearance of widespread public anger using coordinated accounts.

The goal is not always to make people believe one specific lie.

The goal may be to create confusion, distrust and delay.

That can be enough to slow infrastructure development.

Criticism Is Not Automatically Disinformation

This point needs to be clear.

OpenAI’s report should not be used to dismiss local opposition to data centers.

Residents have legitimate reasons to ask:

• Who will pay for new power infrastructure?
• Will household electricity prices rise?
• How much water will the facility use?
• How many permanent jobs will it create?
• What happens during a heatwave?
• Will backup generators increase pollution?
• Who is responsible if the grid becomes less reliable?
• What tax incentives is the company receiving?

Those are normal democratic questions.

Calling every critic a foreign influence agent would be dishonest and counterproductive.

The right response is transparency.

Companies and governments should publish clear data, explain costs, disclose water demand, model grid effects and show how risks will be managed.

Facts are the strongest defence against manipulation.

AI Infrastructure Needs Better Communication

The data center industry has often made its own problem worse.

Projects are sometimes announced with limited detail. Developers may use shell companies during site acquisition. Communities hear about deals after important decisions have already been made.

That creates suspicion.

When officials say a project will bring jobs and growth but do not explain water use, electricity demand or public subsidies, residents assume something is being hidden.

Foreign influence campaigns can exploit that gap.

Better communication would include:

• Public energy-use estimates
• Water-consumption plans
• Cooling technology details
• Grid-upgrade costs
• Emergency operating plans
• Local tax agreements
• Job estimates
• Independent environmental reviews
• Climate-risk assessments

Transparency does not guarantee approval.

It makes the debate harder to manipulate.

The Industry Needs Flexible Operations

Data centers may also need to change how they use electricity.

Some computing tasks can be moved to different times or locations. Training workloads that are not urgent may be delayed during periods of extreme grid stress. Backup generation can be replaced or supplemented with batteries, cleaner fuels and on-site energy.

Operators can also build in regions with stronger grids and better climate conditions.

Flexibility could help data centers become part of the solution instead of simply adding demand.

But that requires coordination with utilities and grid operators.

It also requires companies to accept that unlimited, uninterrupted power demand may not always be realistic during emergencies.

The Key Facts

• A large share of global data center capacity is exposed to severe weather or long-term climate stress
• Heatwaves raise cooling demand and can strain the power grid
• Drought and water scarcity are becoming major local concerns
• Modern AI chips create higher heat loads than traditional computing equipment
• Liquid cooling and closed-loop systems can reduce some water and energy needs
• Extreme weather is already creating insurance and construction losses
• OpenAI says it disrupted suspected China-linked influence operations targeting U.S. AI and data center debates
• The campaigns reportedly used real concerns about power bills, blackouts and environmental effects
• OpenAI said the operations had limited reach
• Legitimate criticism of data centers should not be confused with covert manipulation

The Bottom Line

AI data centers sit at the centre of two fast-growing risks.

The first is physical.

Extreme heat, drought, storms and grid stress can threaten the facilities that power the AI economy.

The second is informational.

Foreign actors can exploit public concern to create distrust, confusion and political resistance.

The answer is not to deny the climate problem.

The answer is not to dismiss every critic as propaganda.

The answer is to build stronger infrastructure, publish better data and make decisions in public.

AI needs electricity, cooling and trust. Losing any one of them can stop the system.