The Environmental Cost of AI Depends Entirely on Build Out, Use and Governance, Tristan Dolphin
Picture: Aerial view of the construction of the AI hub and Data center near Dublin ,Ohio
Whether AI serves more as a powerful environment solution than a contributor to its harm is a variable that depends entirely on its use and governance.
Katie Self PhD, Senior Investment Manager at Pictet Asset Management, compares the situation today to the early days of the worldwide web. “We're in a really challenging phase right now. We don't yet know the reality of what AI could be used for and what it will be used for. Just as we didn't know in 1999 what the bulk of the internet would be used for, we can’t accurately predict what life post-AI will be like either. When Tim Berners-Lee unleashed the internet to the world in the early 90s, he didn't foresee the proliferation of Social Media Managers in the job ads of 2016. We know roughly the same amount about the potential impacts of AI”.
This lack of clear visibility includes uncertainty around the potential environmental impacts of AI – both positive and negative.
The environmental cost of the ‘build’
Self, who manages the Pictet Global Environmental Opportunities Fund, feels that this goes with the territory of a scaling technology. “We are in the ‘build’ phase now, and with that build comes environmental impact. Finding solutions to that challenge lies with energy efficiency. It's about making sure that the buildout happens in as efficient a way as possible and with minimal environmental impact.”
How, though, will it be possible to prepare for the new reality and build the infrastructure before we fully understand the – no doubt huge number of – applications?
One inevitability is the increasing need for data centres, which bring with them the need to combat massive heat and power demands driven by intense computing. As Self explains, “Data centres can be thought of as AI factories. The average facility requires about three times the amount of power as a traditional data centre, giving off nine times the amount of heat because of that computer intensity.
Combatting the problem of heat, Self believes, sits with power efficiency or cooling innovations. The fact is, she says, these are currently produced by “moonshot companies”, pursuing highly audacious transformative ambitions and requiring early investment to achieve them. “Liquid cooling can now reduce heat intensity almost tenfold, if not more with some of the new technologies coming out. Targeted investment is required to help get those early-stage technologies up the curve.”
Optimism for the ‘use’ phase
Self is more hopeful about entering the ‘use’ phase of the AI transition, believing that new technologies will turn AI into a climate asset, leveraging predictive modelling for forecasting, environmental mitigation and enhanced disaster prevention systems.
“We're already seeing AI enhancements in our ability to forecast the weather. It will be fantastic for things like orienting wind turbines ahead of time to make the most out of wind currents and maximise clean energy generation. It will also be critical to forecasting extreme weather events, putting in place early warning systems,” says Self, citing the tragic May 2025 Swiss glacier collapse in Blatten when widespread human disaster was avoided due to the high precision early warning technology. “Thankfully, thanks to meteorological early warning systems, the area was evacuated – even the livestock – and widespread casualties were avoided.”
Ultimately, AI is neither a climate solution nor an environmental liability alone. It has the capacity to be both. While the build phase highlights tangible environmental costs, the emerging use phase suggests that AI could become a critical asset in climate mitigation and adaptation, from improving renewable energy efficiency to preventing or preparing for natural disasters.
Investors that focus on early-stage investment in innovative solutions to combat its environmental impacts, may be participating in some of the more positive stories to emerge from the unfolding AI transition.
Tristan Dolphin is Head of Sustainable Investments at Stonehage Fleming. He was in conversation with Katie Self at the Stonehage Fleming UK Philanthropy Conference in London, May 2026.
Disclaimer: This has been prepared for information only. It is neither an offer to sell, nor a solicitation to buy, any investments or services.
SFIM cannot guarantee accuracy of the information made by third-parties and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based.
Issued by Stonehage Fleming Investment Management Limited (SFIM). Authorised and regulated by the Financial Conduct Authority (FRN.194382) and registered with the Financial Sector Conduct Authority (South Africa) as a Financial Services Provider (FSP No. 46194).