The summer of 2026 will be remembered for datacentres becoming a mainstream issue. Unfortunately, for all the wrong reasons.
A topic long-confined to IT and infrastructure teams is suddenly being hotly discussed by everyone from acclaimed authors, to drink brands, and the US president. There is even evidence that datacentres may become the defining issue of the approaching US midterms.
While those of us in the industry know that datacentres have quietly underpinned technological innovation for decades, AI has pulled them into a leading role and the limelight has been unflattering.
There is no getting around the fact that public sentiment on datacentres today is resoundingly negative, with studies showing that around half of Brits now oppose datacentre construction in their local area. In America, that figure is 75%, making it a rare issue that unites opinion across party-political lines.
The public’s opposition to datacentres is representative of a wider anxiety around the environmental impact of increased computing demand, driven by the rise of agentic AI. Their physical presence makes them an obvious focal point for outrage but it is the technology running within them that is driving unsustainable consumption.
It is up to companies to address the sustainability of their AI roll-outs now before the public backlash extends beyond datacentres to the Large Language Models (LLMs) running within them. Failure to do so could put their investment in AI deployments – and all of the efficiency gains they are yet to realise – at risk.
Sustainability issues drive public backlash
Studies in the US and UK consistently expose sustainability issues as the driving cause of the datacentre backlash:
A YouGov survey showed environmental impact as the main cause of opposition in the UK (cited by 39% of respondents), followed by pressure on energy supply / infrastructure, and water usage concerns.
A University of Massachusetts (UMass) poll showed “Environmental Harm and Depletion of Natural Resources” was by far the largest bracket of opposition in the US.
Fox News also found that 75% cited concerns such as energy use, water bills and traffic as the main reason they opposed a datacentre in their area.
While there is undeniably work to be done by the hyperscalers to improve sustainability practices and prove environmental responsibility, businesses also need to acknowledge their role in bringing public opinion to a boiling point.
Datacentres are meeting demand driven by companies that have ramped up computing consumption with little-to-no concern for sustainability. Nearly a third of all cloud usage is wasted on “zombie” resources, which demonstrates the efficiencies that could be gained, the capacity that could be used, and the waste that could be reduced to conserve resources and win back public trust.
This waste is now exacerbated by a rush to deploy agentic AI at scale with the same lack of care. According to data from the British Standards Institute, 62% of businesses say they expect to increase investment in AI this year but only 27% consider their AI footprint in sustainability strategy and reporting. Most companies have no sustainability plan, no commitments, or even a top level number for the impact of their AI usage.
Companies need to learn the lessons from cloud computing and demonstrate a commitment to adopting AI in an informed and responsible way from the start.
Action companies can take
There are three actions that all businesses that use datacentres can take to play their part in addressing legitimate public concerns around environmental impact:
Incentivise computing efficiency: A large part of the 29% of zombie resources are due to a culture where consumption is incentivised and there are few consequences for generating waste. For proof that this attitude has carried into the AI era you need look no further than tokenmaxxing – where exhaustive AI consumption was falsely equated with innovation. Companies including Uber, Amazon, and Meta were among the first to feel the enormous expense of this practice but all companies need to go further to incentivise employees to use AI in a cost- and resource-effective way to reduce environmental and financial waste.
Demonstrate sustainable decision making: Companies have the agency to make choices when it comes to datacentres and AI to reduce their environmental footprint. AI providers and models have levels of different environmental impact and organisations can make smart decisions about using the right LLM for the right task, in order to minimise the carbon, energy and water impact of their AI. Intentionality in decision making will go a long way to winning back public trust.
Increased public transparency: It has been argued by some commentators that the environmental impact of datacentres is overblown. However, this is not the fault of the public – it is the consequence of a lack of transparency. Obfuscation around water consumption, energy sources, and carbon output, means that it is very easy for public opinion to be swayed by, for example, reports of rising local temperatures, noise pollution, and use of forever chemicals. As companies increase their use of AI, it is within their own interests to be open about the environmental impact of their deployments and methods they are taking to offset adverse effects.
All three of these actions hinge on the availability and use of sustainability data to provide trusted figures that companies’ IT teams can use to identify the greatest areas of environmental impact, take positive actions, and demonstrate progress to the public.
Datacentres are in the spotlight but AI could be next
While much of the public’s ire has been directed towards the most visible target of datacentres, in reality it is the businesses that run services within them that need to change their practices to reduce waste, improve efficiencies, and ensure datacentre usage grows sustainably.
This has not been a priority for businesses to date. There have been few incentives for companies to be more efficient in their cloud, AI, and datacentre usage and it wasn’t in hyperscalers’ interests to encourage companies to use fewer resources.
But, in the era of AI, there is a lot at stake if companies don’t learn the mistakes from the past and tackle sustainability openly and head on. AI has far more organisational impact than cloud computing as it extends past IT teams and is more reliant on wider employee buy-in. Early studies have shown that environmental responsibility is a major issue for staff, especially younger generations. More than half (53%) of Gen Z employees say that they are concerned about the environmental impact of AI and they are up to four times more likely to list it as an issue than baby boomers.
This isn’t just about awkward questions for executives. It could derail deployment of the “big AI roll-out”. If employees are dissatisfied with answers around ethical concerns about AI usage, leaving licenses unused, and companies facing a lot of cost with very little gain.
Companies should see the datacentre backlash as a bellwether of what is to come if they don’t adopt AI in an informed and responsible way. There is an opportunity now for organisations to create the framework for their AI usage to grow sustainably, where every kilowatt of power leads to maximum output and minimum environmental cost. Those that learn from the lessons of the past and take sustainably seriously will reap the benefits of greater adoption rates, happier staff, and higher ROI for their AI roll-outs.
John Ridd is co-founder and CEO of Greenpixie.
