According to Google’s new report, the median Gemini prompt consumes 0.24 watt-hours of electricity. That’s about the same as running a standard microwave for 1 second.

Why does this matter? Until now, Big Tech refused to disclose per-prompt energy usage. Researchers could only estimate. Google’s report is the first transparent breakdown from a major AI company.

Here’s where the energy goes: 58% → Google’s AI chips (TPUs) 25% → CPU & memory of host machines 10% → backup machines standing by 8% → cooling & data center overhead AI isn’t just about chips—it’s the whole system

And it’s not just electricity. Each Gemini text prompt also consumes: 0.26 milliliters of water (≈ 5 drops) 0.03 grams of CO₂ emissions Google says this impact is tiny like watching a few seconds of TV.

The bigger surprise? AI efficiency is skyrocketing. In May 2024, a Gemini prompt used 33x more energy than it does today. Software optimization + hardware improvements are cutting the cost fast.

Of course, not all prompts are equal. Summarizing a book? → Much higher energy use Running reasoning-heavy models? → More steps, more power Text vs. images/videos? → Generating media consumes far more The median prompt is just the middle ground.

So what’s missing? Google still won’t say how many total Gemini queries run per day. Without that, we can’t calculate the true global footprint of AI usage. Researchers want a standardized “AI energy score” like Energy Star for appliances.

Still, this is a milestone. For the first time, we have real, company-verified numbers about what it costs to talk to AI.

AI isn’t free. Every question we ask consumes power, water, and resources. Google just opened the door to a conversation the entire industry has been avoiding. Do you think AI companies should be required to publish per-prompt energy usage?

I hope you've found this thread helpful. Follow me @sentient_agency for more.

@sentient_agency This puts the whole AI revolution in a completely different perspective when you consider the actual energy costs.

@sentient_agency This revelation is going to force a serious conversation about the true cost of artificial intelligence deployment.

@sentient_agency Never considered the environmental impact of my ChatGPT conversations.

@sentient_agency This revelation is going to force innovation in energy efficient AI architectures much faster than previously expected.

@sentient_agency This explains why there's such a rush to develop more energy efficient AI chips and computing architectures.

@sentient_agency This is absolutely mind blowing. Had no idea that simple ChatGPT queries were consuming this much energy per request.

@sentient_agency Every ChatGPT conversation I have now feels different knowing the environmental cost of each individual query I send.

@sentient_agency UAE's commitment to sustainable AI inspires confidence. By prioritizing energy-efficient solutions, we're shaping a future where digital transformation and environmental responsibility advance together.

@sentient_agency In my opinion, that's not the case.

@sentient_agency This data makes the case for being more selective about AI usage rather than just automating everything possible.

@sentient_agency This is exactly why we need to be more thoughtful about when and how we use AI tools.