- Investors are demanding greater detail on ESG reporting ahead of this year's AGMs
- North American data centers consumed 1 trillion liters of water in 2025
- Current reporting and targets are too vague, they argue
More than a dozen investors are said to be mounting the pressure on some of the biggest names in tech to disclose detailed data center water and energy usage ahead of annual shareholder meetings.
New Reuters reporting claims that central to the demands are that companies including Amazon, Microsoft and Google offer site-level transparency instead of high-level metrics to provide a better insight into the environmental effects of their data centers.
This comes amid growing concern over AI's impact on local communities and the environment, with North American data centers alone said to have consumed around one trillion liters of water in 2025.
Investors want more detailed data center sustainability reporting
With fears that data centers can strain local water supplies and power grids, many large projects have already been cancelled, delayed or relocated due to local opposition, thus investors are demanding not only more detailed impact metrics but long-term strategies.
At the moment, there are a number of gaps in reporting, with Google failing to include third-party data centers in its reporting and Amazon reporting water usage relative to power, rather than total consumption. Microsoft also lacks site-level water usage data.
As for each of the three companies, their broader long-term strategies target positive outcomes, but specific progress is more cryptic in ESG reporting. Most are targeting 2030 deadlines for being water positive or carbon neutral, while many have hit 100% renewable energy matching to varying degrees already.
As for where investor demands leave data center operators in the meantime, it's clear that there is an increasing expectation for companies to prove AI growth won't come at the expense of local communities and the environment.
Moving forward, standardized environmental reporting could be introduced for more granular details, ultimately incentivizing hyperscalers to become even more efficient.
