The Latitude stage: How AI changes our digital energy footprint

14 snips

Aug 20, 2025

Vijay Gadepally, a senior scientist at MIT Lincoln Laboratory and CTO of Radium Cloud, shares insights on how artificial intelligence is evolving from simple tools to always-on companions, akin to the movie 'Her'. He explores the rising energy demands of AI, explaining that even basic tasks can consume significant power. Gadepally discusses how better software and thoughtful design can lead to energy savings and lower emissions, as well as the need for innovations in data center design to ensure sustainability amid growing AI usage.

Ask episode

AI Snips

Chapters

Transcript

Episode notes

INSIGHT

Real-Time Generation Drives Energy Use

Large language models generate responses via billions of real-time calculations rather than simple lookups.
Millions of such interactions scale to power-plant-level electricity demand when used constantly.

INSIGHT

Agentic Tasks Match Household Appliances

Agentic reasoning tasks can be extremely compute intensive, often using many GPUs for a single instance.
Vijay estimates a single agentic task can consume roughly three kilowatt-hours, comparable to running a dishwasher.

ANECDOTE

Kids' Story App Revealed Hidden Costs

Vijay built a children's story app and found each generated story used as much energy as the entire iPad.
The cloud-side compute for that simple interaction created a hidden digital footprint he couldn't see locally.

Get the Snipd Podcast app to discover more snips from this episode

Get the app

We're witnessing a profound shift from discrete AI tools to always-on AI companions — systems that provide constant feedback, conversation, and support. Sound familiar? It's the 2013 movie "Her" becoming reality.

In this episode of Open Circuit, we have a conversation with MIT’s Vijay Gadepally from our Transition-AI conference about how the spread of artificial intelligence is reshaping our digital energy footprint.

As a senior scientist at MIT’s Lincoln Laboratory and CTO of cloud computing company Radium Cloud, Gadepally has an inside view on the energy intensity of reasoning models, AI agents, and chatbots. He details how simple tasks are now becoming energy-intensive computing events.

Gadepally also explains how operations per watt have improved dramatically, why better software can dramatically reduce emissions, and what it will take for computing innovations to keep pace with our growing appetite for AI.

Registration is now open for the Power Resilience Forum in Houston, Texas next January. Power Resilience Forum 2026 is the premier event on grid resiliency, bringing together leaders from across the power sector to address the new realities of planning and operating the grid in an era of extreme weather and wildfires.

Open Circuit is brought to you by Natural Power. Natural Power specializes in renewable energy consulting and engineering, supporting wind, solar, and battery storage projects from concept through financing. Discover how we're creating a world powered by renewable energy at naturalpower.com.

Open Circuit is brought to you by Sungrow, the trusted provider of PV inverters and battery storage. With over 605 GW installed worldwide and a BloombergNEF ranking of “most bankable” in power conversion and energy storage, Sungrow provides solar tech you can count on. Learn more at sungrowpower.com.

Credits: Co-hosted by Stephen Lacey, Jigar Shah, and Katherine Hamilton. Produced and edited by Stephen Lacey. Original music and engineering by Sean Marquand.