Big professional update: Iโm thrilled to share that Iโve joined @Google as Head of Market Innovation on the Advanced Energy team, and today is my first day.
The power sector is undergoing a transformation as rapid load growth, emerging technologies, and ambitious decarbonization goals all converge. In this role, Iโll be focused on identifying and advancing innovations to better enable electricity markets to accommodate AI-driven demand and clean energy technologies.
As I transition to the Bay Area, I'm excited to be joining a world-class team working at the frontier of energy innovation, and to collaborate with partners across the energy ecosystem. I'm also grateful that I can continue my PhD research and stay close to the research community.
I wouldn't be here without the support of incredible mentors and colleagues at @DukeU and beyond โ thank you for everything.
Letโs get to work!
AI efficiency is important. Today, Google is sharing a technical paper detailing our comprehensive methodology for measuring the environmental impact of Gemini inference. We estimate that the median Gemini Apps text prompt uses 0.24 watt-hours of energy (equivalent to watching an average TV for ~nine seconds), and consumes 0.26 milliliters of water (about five drops) โ figures that are substantially lower than many public estimates.
At the same time, our AI systems are becoming more efficient through research innovations and software and hardware efficiency improvements. From May 2024 to May 2025, the energy footprint of the median Gemini Apps text prompt dropped by 33x, and the total carbon footprint dropped by 44x, through a combination of model efficiency improvements, machine utilization improvements and additional clean energy procurement, all while delivering higher quality responses.
See the blog or technical paper for more about our methodology and ongoing efforts.
Blog:
https://t.co/CoMm5gV9SR
Link to detailed paper: https://t.co/UBi9rd6gEC
This is my quant... @ddpaolella talking about the new @Breakthrough GRIDS initiative to help speed up grid modeling and get study times down to weeks not months or years. Thanks @gurobi for having us in Berlin!
@nickvanosdol You'd get a much cheaper $/ton (maybe $100-150 avg) -- up to, say, a GT.
Constrained innovation proving what techs might actually scale (across both DAC + point source).
Flue stack retrofits not necessarily slam dunks, need long lived assets.
Net positive for humanity, probly
@JessePeltan Agree, generally. But all relevant tech are within ~1 order of mag of 1.0, so the math is done kinda automatically.
Also, thinking in the $ M/MW can be useful to remind us that line items like a substation or admin building can cost many millions - scaling matters
@joelhedwards@JessePeltan Fantastic, thanks, Joel. What is the data source there?
I've heard generically 0.5% to 2% per year resource decline, that's wide range (4:1 ratio) but is it reasonable incl for newer techs?
Burning fuels like oil, gas, and coal feels comfortable and familiar. We've been doing it in our homes for centuries. Human impact gradual.
Hazards from things like batteries and nuclear feel much more exotic. Interaction much more rare.
Education needed for long time to come.
Moss Landing, the largest battery in the US caught fire yesterday.
A journalist just asked me, "Are batteries safe?"
I told him the fossil fuel plants batteries replace kill 1,000s of people in the US / year.
Batteries aren't perfect. But they're way better than coal/gas.
@duncancampbell 93% for SMRs in early 2030s is ambitious, as the kinks get worked out. but, they are modular, and failure modes can be relatively isolated. there's probably a sweet spot of reactor MWe and # units for this type of application. there are diseconomies of scale the smaller you get