Energy: Building a foundation for the future

Realizing AI’s potential requires bringing more reliable and abundant energy online, optimizing how we use it, and advancing innovative solutions to unlock clean, affordable energy everywhere.

Powering digital infrastructure responsibly

Powering digital infrastructure responsibly

Scaling intelligence requires a sustainable foundation. We pay our own way and are committed to responsible energy growth in the communities we call home.

“By integrating silicon design, thermal engineering, and direct energy investment, Google is helping build a more affordable, sustainable, and powerful energy future for everyone.”
Steven Walter_headshot
Steven Walter
VP Global Infrastructure and Energy • Google AI and Infrastructure

Pioneering integrated AI efficiency

Because we design the full vertical stack—from the AI chips (TPUs) to the data centers—we can optimize every layer to ensure that the cutting-edge AI you use is built and run as efficiently as possible.

The cleanest energy is the power we didn’t use in the first place

The cleanest energy is the power we didn’t use in the first place

By re-engineering the fabric of computing, our data centers use 83% less overhead energy than industry average and have delivered over 3x more compute performance per unit of energy than five years ago.

Designing chips that “run hot”

Designing chips that “run hot”

Our custom AI silicon (including TPUs) is engineered to operate efficiently at higher internal temperatures. This breakthrough allows us to drastically reduce water and energy consumption.

Optimizing idling

Optimizing idling

Our serving stack makes highly efficient use of CPUs and minimizes TPU idling by dynamically moving models based on demand in near real time, rather than using a ‘set it and forget it’ approach.

The data center as a grid asset

The data center as a grid asset

We re-engineered our software to make our data centers consume power more flexibly. By including load flexibility in our overall energy plan, we can manage AI-driven growth even where power generation and transmission are constrained.

Leading the industry in liquid cooling

Leading the industry in liquid cooling

AI infrastructure is hot. Advanced thermal management improves efficiency and our fifth-generation cooling distribution unit is at the forefront, accompanied by an open standards specification to accelerate industry-wide adoption.

Reducing energy and carbon used for Gemini queries by 33-44x

Reducing energy and carbon used for Gemini queries by 33-44x

Optimizing AI inference efficiency—the use of AI models to make predictions or generate text or images—is critical as AI usage rises. We’re sharing our comprehensive methodology for measuring energy, emissions, and water impact to increase industry-wide consistency in calculating AI’s resource consumption and efficiency.

Creating an abundant, reliable energy supply

Investing in and accelerating next-generation clean energy technologies to unlock 10GW+ of future capacity.

The AI-energy nexus

As AI models grow in complexity, so does the demand for the energy that powers them. Explore how Google is tackling this critical tension between technological progress and sustainability.

The AI-energy nexus

Tapping abundant space-based solar energy

Dig into the ambitious work of Project Suncatcher, a bold effort to someday place solar-powered data storage in orbit to fuel the next generation of AI.

Tapping abundant space-based solar energy

Advancing existing energy sources

Driving sustainable innovation for society

We're bringing more clean energy online and advancing responsible water use at our data centers across the globe.