After years of preaching climate responsibility, Big Tech is now facing some uncomfortable questions about what they're actually disclosing regarding AI's environmental footprint. New reports from Google, Amazon, and Microsoft show emissions and water use continuing to climb as AI infrastructure expands—and the gap between what they reveal and what's actually happening is becoming harder to ignore.

The Transparency Gap

Let's be real: comparing these companies' disclosures right now is like trying to match screenshots across different operating systems. There's no standard reporting format, no regulatory requirement forcing them to detail most AI-related environmental metrics, and plenty of wiggle room for spin. United Nations Secretary-General António Guterres called on tech companies this week to publicly disclose the "full footprint" of their data centers—including carbon, water, and land use—as part of the U.N.'s new AI Environmental Transparency Initiative.

Ranking the Reports

Alex de Vries-Gao, a researcher at VU Amsterdam university who tracks digital sustainability through his Digiconomist platform, reviewed the latest disclosures and ranked them on transparency alone. Meta comes out on top, followed by what he describes as a "toss-up" between Google and Microsoft for second place. Amazon lands dead last. The reasons are telling: Amazon reports the strongest water-efficiency figure but provides the fewest metrics overall—including omitting basic electricity consumption data that would help outside researchers piece together the full picture.

The Water Question

Here's where it gets interesting—and where most coverage glosses over the technical details. Only Meta discloses the water associated with electricity generation, not just direct data center consumption. This matters enormously because according to de Vries-Gao's calculations based on Meta's 2025 disclosures, that indirect water use was roughly 24 times larger than what the company consumed at its actual facilities. Generating electricity from fossil fuels and nuclear power requires massive amounts of water for cooling, and tech companies are increasingly swapping water-intensive cooling for more energy-hungry alternatives as they scale up AI workloads.

Data Center Scale

Amazon, Google, Microsoft, and Meta account for roughly two-thirds of the data-center power capacity in a top-15 ranking by financial firm Jefferies. What these four do shapes the entire market. Boris Gamazaychikov, who co-founded Sustainable AI Group—a research and advisory firm helping companies address AI's environmental impacts—noted there's "a bit of a reluctance to share a lot of things in this competitive dynamic" given that these are public companies competing fiercely for AI dominance.

What's Next

Microsoft made a notable move by newly disclosing specific water and power metrics at individual data center locations in its latest report, which de Vries-Gao called "a significant improvement in transparency." Google doesn't publish a company-wide water-efficiency metric, arguing that a global average masks its local risk-based approach—but using Google's own data, de Vries-Gao estimates the company would rank last on this measure if it did disclose. Gamazaychikov expects customer pressure from consumer brands and other enterprise clients will likely drive more transparency before any regulatory mandate materializes.

Key Takeaways

  • Meta leads in overall transparency; Amazon ranks last despite having strong individual metrics
  • Indirect water use from electricity generation is the biggest blind spot—Meta's is ~24x its direct consumption
  • No regulatory standard exists yet for AI environmental disclosures, and none is coming soon
  • Microsoft recently improved its reporting by disclosing location-specific water and power data

The Bottom Line

The AI boom isn't just a compute race—it's an infrastructure arms race with real physical consequences. Until there's actual teeth behind disclosure requirements or market pressure forces standardization, we're essentially taking Big Tech's word on the environmental cost of the systems they're building to replace human workers. That should sit wrong with anyone paying attention.