Gianluca Guidi
Postdoc, Harvard
I’m a postdoc at Harvard working with Francesca Dominici, after a PhD in AI for Social Sciences.
I investigate the hidden costs of AI and what they mean for society and policy.
Any Colour You Like what I work on
-
Energy, carbon and water
The electricity and carbon of U.S. hyperscale data centers, and where they use water, on site and at the power plant.
-
Health
The air pollution from the power plants that run AI, the lives lost when coal plants stay open to feed it, and a grid planned around health.
-
People and justice
Who pays for AI, and what it means for the communities living next to bitcoin mines or a proposed gas plant.
Us and Them interests
- AI’s energy, carbon and water use
- Who pays for AI
- Environmental justice
- Health equity
- Air pollution and public health
- Social justice
- Energy and climate policy
- Network science
In short: what the great gig in the sky really costs, and who ends up paying for it.
Paperback Writer papers · 10
- Assessing the carbon emissions and energy consumption of US hyperscale data centresNature Sustainability2026
- Who Pays for Artificial Intelligence? Energy, Climate, and HealthAmerican Journal of Public Health2026
- The environmental burden of the United States’ bitcoin mining boomNature Communications2025
- The Hidden Water Geography of U.S. Hyperscale Data Centers in the AI EraarXiv2026
- Environmental Burden of United States Data Centers in the AI EraarXiv · 100+ citations2024
- Delayed Coal Plant Retirements in the Era of Data Centers Carry a Measurable Mortality CostmedRxiv2026
- Public Health and Economic Burden of a Proposed 3,500-MW Natural Gas Plant in VirginiaSSRN · the plant was withdrawn2026
- Tracing two decades of carbon emissions using a network approachScientific Reports2024
- Towards a Health-Based Power Grid Optimization in the AI EraarXiv2024
- The Global Carbon Trade Networkworking paper—
Here, There and Everywhere press · 12
- The tech industry is investing in mega gas plants to power their data centersThe Washington Post2026
- Trump’s tax bill seeks to prevent AI regulations. Experts fear a heavy toll on the planetThe Guardian2025
- We did the math on AI’s energy footprintMIT Technology Review2025
- AI’s emissions are about to skyrocket even furtherMIT Technology Review2024
- How AI Could Help Reduce Climate Change And Air PollutionForbes2025
- The Dual-Geography of AI’s Water FootprintHarvard Business School AI Institute2026
- Study maps water footprint of AI data centersHarvard Salata Institute2026
- Bitcoin’s Toxic FootprintPublic Health Post2025
- This rural community fought one of the country’s biggest gas-powered data centers, and wonGrist2025
- The Energy Hunger of the AI Boom and its ConsequencesFrankfurter Allgemeine Zeitung, in German2024
- Le Emissioni dell’Intelligenza Artificiale Pronte a una Crescita EsponenzialeAI4Business, in Italian2024
- Researchers sound the alarm on controversial AI plan rejected by lawmakersThe Cooldown
Welcome to the Machine code & data · 4
- Hyperscalers Emissions and Energy Consumption in the USHarvard Dataversedata
- hyperscale-emissionselectricity and carbon of U.S. hyperscale data centerscode
- hyperscale-water-geographytheir water, on site and at the power plantcode
- bitcoin_mining_env_burdenelectricity, PM2.5 and communities near U.S. bitcoin minescode
A Day in the Life elsewhere
The rest of the time goes to music and art, the only parts of my life without error bars.
I play in Project42, a 10+ person band that started with a few Harvard Data Science Initiative scholars. The name is the ultimate random seed, and roughly the latitude of both Boston and Rome. , not quite Live at Pompeii.
I got by with a little help from my PhD committee: Falco J. Bargagli Stoffi, Dino Pedreschi, Giulio Rossetti and Rossana Mastrandrea.
Come Together contact
For press, policy or research, write to ggianluca@hsph.harvard.edu. “Help!” is also a valid subject line.