Uploaded July 2026 | Updated September 2026, 2 weeks ago
AI isn’t weightless. Every model sits on data centres, power, water, land, heat, and governance choices.
I’m joined by Sophia Mendelsohn, who leads SAP’s Global Sustainability Platform, for a sharp look at sustainable AI, climate tech, and the real infrastructure demands behind digital transformation. Sophia argues that treating AI as “just software” lets leaders sidestep the hard questions — where the energy comes from, how water is used, who carries the risk, and whether the business case still holds when the externalities are counted.
That matters now because companies are racing into AI while climate, energy, procurement, and policy pressures are tightening around them. Data centres are no longer a side issue. They are becoming a board-level question, a local community question, a Scope 3 question, and in many cases, a social licence question.
What changed my thinking was Sophia’s insistence that sustainability teams cannot stay trapped in PDF production and post-hoc carbon accounting. The better opportunity is earlier and more strategic: bring sustainability into AI procurement, supplier data, financial planning, product carbon footprints, and investment decisions before those systems lock in.
We also get into a particularly useful Scope 3 point: asking suppliers for emissions data does not mean you’ll get reliable answers. AI could help sustainability teams move from waiting for disclosures to calculating baselines, testing assumptions, and making stronger decarbonisation decisions.
No hype. No silver bullets. No pretending AI has no physical footprint.
This is for business leaders, sustainability professionals, policymakers, investors, and technologists trying to cut emissions without breaking the systems they depend on.
If you’re working on sustainable AI, data centres, Scope 3 emissions, or climate policy in the real world, I’d be interested in your perspective.
Podcast website: climateconfidentpodcast.com
Subscribe for more conversations on climate tech, decarbonisation, energy transition, emissions reduction, and net zero strategy.
Timestamps below.
00:00 - AI Needs Data Centres
00:57 - Meet Sophia Mendelsohn
02:17 - Beyond PDFs to Action
03:50 - AI as Physical Infrastructure
06:36 - Business Stakes and ROI
07:46 - From Reporting to Decision Rules
11:50 - Bad Data Bad Decisions
17:36 - Social Licence to Operate
19:38 - Procurement and Early Governance
22:50 - AI for Scope 3 and Baselines
27:09 - Embedding Sustainability in AI
30:05 - Practical Use Cases and Finance
34:39 - CEO Message and Big Picture
38:25 - Where to Start and Wrap Up
AI isn’t weightless. Every model sits on data centres, power, water, land, heat, and governance choices.
I’m joined by Sophia Mendelsohn, who leads SAP’s Global Sustainability Platform, for a sharp look at sustainable AI, climate tech, and the real infrastructure demands behind digital transformation. Sophia argues that treating AI as “just software” lets leaders sidestep the hard questions — where the energy comes from, how water is used, who carries the risk, and whether the business case still holds when the externalities are counted.
That matters now because companies are racing into AI while climate, energy, procurement, and policy pressures are tightening around them. Data centres are no longer a side issue. They are becoming a board-level question, a local community question, a Scope 3 question, and in many cases, a social licence question.
What changed my thinking was Sophia’s insistence that sustainability teams cannot stay trapped in PDF production and post-hoc carbon accounting. The better opportunity is earlier and more strategic: bring sustainability into AI procurement, supplier data, financial planning, product carbon footprints, and investment decisions before those systems lock in.
We also get into a particularly useful Scope 3 point: asking suppliers for emissions data does not mean you’ll get reliable answers. AI could help sustainability teams move from waiting for disclosures to calculating baselines, testing assumptions, and making stronger decarbonisation decisions.
No hype. No silver bullets. No pretending AI has no physical footprint.
This is for business leaders, sustainability professionals, policymakers, investors, and technologists trying to cut emissions without breaking the systems they depend on.
If you’re working on sustainable AI, data centres, Scope 3 emissions, or climate policy in the real world, I’d be interested in your perspective.
Podcast website: climateconfidentpodcast.com
Subscribe for more conversations on climate tech, decarbonisation, energy transition, emissions reduction, and net zero strategy.
Timestamps below.
00:00 - AI Needs Data Centres
00:57 - Meet Sophia Mendelsohn
02:17 - Beyond PDFs to Action
03:50 - AI as Physical Infrastructure
06:36 - Business Stakes and ROI
07:46 - From Reporting to Decision Rules
11:50 - Bad Data Bad Decisions
17:36 - Social Licence to Operate
19:38 - Procurement and Early Governance
22:50 - AI for Scope 3 and Baselines
27:09 - Embedding Sustainability in AI
30:05 - Practical Use Cases and Finance
34:39 - CEO Message and Big Picture
38:25 - Where to Start and Wrap Up
