Are Data Centres Becoming a Front in the AI Information War?

In this short thought piece, Stuart Hall explores the debate around Data Centre water usage, geopolitical influences and the importance of separating legitimate concerns from deliberate amplification.

Read the full LinkedIn piece below.

A recent debate about data-centre water use sent me down an unexpected rabbit hole. I expected to find arguments about cooling systems, WUE and environmental impact. Instead, I found evidence that the debate may have acquired a geopolitical dimension.

In June, OpenAI disclosed an influence campaign it called “Data Center Bandwagon”. It identified accounts it assessed as likely originating in China, posing as Americans and generating social-media content critical of US AI and data-centre infrastructure. One of the narratives being amplified was that data centres were increasing electricity costs for ordinary households. OpenAI found no evidence that the campaign achieved significant reach. But the interesting question is: Why target data centres at all?

Compute is now strategic infrastructure

AI competition is not just about models and GPUs. It is also about land, power, networks, cooling, planning consent and the ability to build data centres quickly enough to deploy compute. If a competitor cannot build the infrastructure needed to operate AI at scale, its technological ambitions eventually hit a physical limit. That creates an obvious strategic incentive. You do not necessarily need to prevent another country buying GPUs if you can make it slower, harder or politically more difficult for it to build the facilities that house them.

The complication is that the concerns are real

This is where we need to be careful. There are legitimate concerns about data-centre development. Some facilities consume significant quantities of water. Some place substantial demands on electricity networks. Communities can reasonably question noise, land use, grid reinforcement, water availability and local economic benefit. None of that is propaganda.

And that may be precisely why these topics are useful for influence campaigns. Effective misinformation rarely has to invent something entirely false. It can start with a genuine issue, remove the qualification and amplify the most alarming interpretation. A worst-case design becomes representative. A local problem becomes universal. A technical “water footprint” becomes “water consumed by data centres”. The underlying fact may still be real. The impression created can be very different.

Water is a good example

Some data centres use evaporative cooling and consume significant volumes of water. Others use very little. Some designs can operate with essentially no ongoing water consumption for cooling. Yet discussion often collapses all of these designs into a single statement: “Data centres consume huge amounts of water.” The reality is much more dependent on cooling architecture, climate, water source and operational strategy. That does not make the environmental concern invalid. It means we should ask better questions.

A foreign megaphone

Threat-intelligence researchers at Alethea have also reported foreign actors amplifying genuine opposition to US data-centre development. Their conclusion was not that China, Russia or Iran created the backlash. It was that foreign actors were layering themselves on top of an existing domestic debate. That distinction matters. An influence campaign does not need to convince people of something new. It may simply need to make an existing concern appear larger, angrier or more certain than it really is. Generative AI makes that increasingly cheap.

The industry should not overreact

There is an equally dangerous conclusion for the data-centre industry: “Anyone opposing a data centre is spreading misinformation.” That would be both wrong and counterproductive. Some projects deserve criticism. Some locations are unsuitable. Some developers need to explain their environmental impact much better. If anything, the existence of information operations makes transparency more important.

Publish the expected PUE. Publish the expected WUE. Explain the cooling design. Explain where the power and water come from. Explain the local benefit. Correct inaccurate claims with evidence rather than rhetoric.

Perhaps the debate has changed

Data centres are becoming strategically important infrastructure. That means they will attract environmental scrutiny, political opposition, lobbying and, apparently, geopolitical influence operations. The challenge is separating legitimate criticism from deliberate amplification without dismissing either. Perhaps the question is no longer simply: “Are data centres sustainable?”

It may increasingly be: “Who is shaping the public debate about the infrastructure needed for AI, and why?” That is a question I suspect we will hear much more about.

Join the conversation here: Stuart Hall – LinkedIn Post

Sources: OpenAI, June 2026 Disrupting malicious uses of AI: “Data Center Bandwagon”; OpenAI, June 2026 Threat Report; Alethea, July 2026 How State Actors and AI Slop Are Amplifying the Data Center Revolt.