Descriptions:
Hannah Ritchie, researcher at Our World in Data, joins Scott Galloway on The Prof G Pod to ground the AI energy debate in International Energy Agency numbers. Data centers as a whole consume roughly 1.5% of global electricity, with AI-specific facilities accounting for approximately 0.5% — figures that seem modest but obscure a severe geographic concentration problem: nearly all global AI compute demand is served by a handful of locations, placing acute pressure on local grids even when the aggregate share of world electricity remains small.
The conversation’s most striking data point concerns China’s electricity buildout: China is currently adding the equivalent of Germany’s entire electrical grid every single year, and in the most recent year all of that new generation came from solar and wind. By contrast, electricity demand in Europe has actually fallen over the past 30 years, and US grid growth has been essentially flat — leaving both regions poorly positioned to absorb the infrastructure demands of an AI expansion. Ritchie argues this pace-of-build gap, not energy price differences, is the real bottleneck for Western AI competitiveness.
Ritchie also addresses whether AI could net-positively affect the energy transition by optimizing grid balancing, charging infrastructure, and permitting processes — concluding the honest answer is we do not yet know whether efficiency gains will offset demand increases. On nuclear, she views Microsoft, Google, Amazon, and Meta signing data center nuclear deals as a positive signal, noting that nuclear’s badly damaged public brand is beginning to recover as tech companies provide long-term demand commitments that make new projects financially viable.
📺 Source: The Prof G Pod – Scott Galloway · Published August 24, 2026
🏷️ Format: Interview
