Amazon, Microsoft, and Google asked to disclose data center water and energy use

More than a dozen investors have called on Amazon, Microsoft, and Google to disclose how much water and electricity their data centers consume in the United States. The impetus is the rapid growth of AI infrastructure: the market wants to understand what the AI boom costs not just in money, but in resources.

What investors are demanding

The request came on the eve of annual shareholder meetings, when major shareholders typically increase pressure on management. Investors want to see not general promises about sustainable development, but specific metrics: how much energy goes into running the facilities, how much water is needed for cooling, how the load changes as computing power expands for AI services. The focus specifically on American data centers is no accident. The US hosts a significant portion of Amazon, Microsoft, and Alphabet's infrastructure, along with disputes surrounding the connection of new capacity to power grids, water consumption, and impact on local communities. For shareholders, this is no longer an image issue, but a matter of operational risks and future costs.

In essence, investors are pushing for a reporting format that can be used for comparison and control: year-on-year dynamics, breakdown by resource type, plans to reduce load on networks and water systems. Without this, it's difficult to understand whether AI infrastructure is becoming more efficient or whether the growth in services is simply outpacing any progress in energy conservation. For the market, this is also a way to check how confidently companies can scale AI without conflicts with local infrastructure.

Why the issue became acute

Interest in the topic has surged against the backdrop of the race for compute for generative AI. The more actively companies deploy accelerator clusters and build new server capacity, the higher the demand for electricity and cooling systems. Even if business continues to publish aggregated ESG reports, it's often hard to understand what portion of consumption is tied specifically to AI load and how rapidly these figures are growing.

• energy consumption of new AI clusters
• water expenditure on cooling
• dependence on local networks and rates
• risk of restrictions from regulators and cities

The market views such data not just through an environmental lens. If companies need to rapidly modernize cooling, buy more electricity at high prices, or relocate construction due to water shortages, this will directly hit margins and timelines for launching new AI services. That's why investors need transparency that allows them to assess the sustainability of strategy, not just a glossy presentation.

Where data is lacking

Currently, technology giants do publish sustainable development reports, but the detail is often limited. Usually it's about aggregate consumption, the share of renewable energy, or long-term climate goals. This isn't enough if a shareholder wants to understand how AI infrastructure expansion is changing the resource profile of a specific business and which facilities create the greatest load.

That's precisely why the request to Amazon, Microsoft, and Google looks like an attempt to shift the conversation from general principles to measurable metrics. The more precisely companies disclose data on water and energy, the easier it will be to compare them with each other, assess infrastructure efficiency, and understand where AI scaling is sustainable and where it hits physical limits.

The problem is also that the industry still has no unified standard for such reporting. One company might show total consumption, another—intensity per unit of compute, a third—limit itself to regional estimates. Formally the data will be disclosed, but investors will still struggle to compare businesses and see who actually controls the resource efficiency of AI facilities. That's why shareholder pressure could over time turn into demands for stricter, more comparable reporting.

What it means

For big tech, the era when you could talk about AI growth without detailed discussion of resources is ending. If shareholder pressure intensifies, data center reporting could become as important a part of AI strategy as new models, chips, and partnerships.