Uploaded March 2026 | Updated September 2026, 2 weeks ago
AI’s power problem may have less to do with chips than turbines.
And that changes how we think about decarbonisation, firm power, and the energy transition.
Brad Hartwig, Co-founder and CEO of Arbor Energy, joined me to unpack a hidden constraint sitting underneath the AI boom, grid reliability, and climate tech deployment: the world cannot build enough turbines fast enough. Arbor is developing modular supercritical CO2 turbines with integrated carbon capture, aimed at delivering firm, scalable power for data centres, utilities, and industrial users.
That matters now because electricity demand is rising from multiple directions at once. AI is accelerating data centre buildout. Electrification is pushing more load onto the system. Utilities are under pressure to keep power reliable while cutting emissions. And yet the machinery needed to provide round-the-clock power is stuck in a brutal supply crunch. In Brad’s telling, more than 50% of data centres are being delayed by power generation constraints, and turbine lead times now stretch into years.
A few things shifted my thinking here. First, this is not just a fuel story. It is a manufacturing story. A turbine shortage, and more specifically a turbine blade shortage, is becoming a real bottleneck for AI infrastructure and possibly for parts of decarbonisation. Second, modularity matters more than many people realise. The case Brad makes for smaller 25MW units, stacked like building blocks, is really a case for flexibility, redundancy, and faster deployment. Third, the conversation around firm clean power still has a habit of becoming ideological when it should be practical. We got into long-duration storage, methane leakage, biomass, sequestration, nuclear, and why no single technology gets to declare itself the winner and go home smug.
This is not silver-bullet territory. It is a grounded discussion about bottlenecks, trade-offs, economics, and what it takes to keep cutting emissions without pretending grids run on slogans.
Useful for climate professionals, energy executives, policymakers, investors, founders, and anyone trying to scale clean power in the real world.
If you’re working on this from the grid, policy, utility, data centre, or climate tech side, I’d be interested in your perspective.
🌍 Climate Confident podcast: climateconfidentpodcast.com
🎧 Follow Climate Confident for weekly episodes on climate tech, clean energy, emissions reduction, and the practical realities of the transition
🔔 Subscribe for more grounded conversations on decarbonisation, energy security, net zero, and climate solutions
Timestamps:
00:00 – Cold open
00:30 – Arbor Energy and the turbine bottleneck
03:51 – Why AI and electrification are straining power systems
08:31 – Why modular turbines change the deployment model
13:43 – How oxy-combustion and carbon capture work
18:23 – Biomass, carbon-negative power, and CO2 storage
28:19 – Why batteries are not enough for firm power
32:18 – Clean baseload, nuclear, and all-of-the-above systems
37:01 – Lightning round: gas, batteries, modularity, bottlenecks
39:37 – Arbor’s timeline to put electrons on the grid
AI’s power problem may have less to do with chips than turbines.
And that changes how we think about decarbonisation, firm power, and the energy transition.
Brad Hartwig, Co-founder and CEO of Arbor Energy, joined me to unpack a hidden constraint sitting underneath the AI boom, grid reliability, and climate tech deployment: the world cannot build enough turbines fast enough. Arbor is developing modular supercritical CO2 turbines with integrated carbon capture, aimed at delivering firm, scalable power for data centres, utilities, and industrial users.
That matters now because electricity demand is rising from multiple directions at once. AI is accelerating data centre buildout. Electrification is pushing more load onto the system. Utilities are under pressure to keep power reliable while cutting emissions. And yet the machinery needed to provide round-the-clock power is stuck in a brutal supply crunch. In Brad’s telling, more than 50% of data centres are being delayed by power generation constraints, and turbine lead times now stretch into years.
A few things shifted my thinking here. First, this is not just a fuel story. It is a manufacturing story. A turbine shortage, and more specifically a turbine blade shortage, is becoming a real bottleneck for AI infrastructure and possibly for parts of decarbonisation. Second, modularity matters more than many people realise. The case Brad makes for smaller 25MW units, stacked like building blocks, is really a case for flexibility, redundancy, and faster deployment. Third, the conversation around firm clean power still has a habit of becoming ideological when it should be practical. We got into long-duration storage, methane leakage, biomass, sequestration, nuclear, and why no single technology gets to declare itself the winner and go home smug.
This is not silver-bullet territory. It is a grounded discussion about bottlenecks, trade-offs, economics, and what it takes to keep cutting emissions without pretending grids run on slogans.
Useful for climate professionals, energy executives, policymakers, investors, founders, and anyone trying to scale clean power in the real world.
If you’re working on this from the grid, policy, utility, data centre, or climate tech side, I’d be interested in your perspective.
🌍 Climate Confident podcast: climateconfidentpodcast.com
🎧 Follow Climate Confident for weekly episodes on climate tech, clean energy, emissions reduction, and the practical realities of the transition
🔔 Subscribe for more grounded conversations on decarbonisation, energy security, net zero, and climate solutions
Timestamps:
00:00 – Cold open
00:30 – Arbor Energy and the turbine bottleneck
03:51 – Why AI and electrification are straining power systems
08:31 – Why modular turbines change the deployment model
13:43 – How oxy-combustion and carbon capture work
18:23 – Biomass, carbon-negative power, and CO2 storage
28:19 – Why batteries are not enough for firm power
32:18 – Clean baseload, nuclear, and all-of-the-above systems
37:01 – Lightning round: gas, batteries, modularity, bottlenecks
39:37 – Arbor’s timeline to put electrons on the grid
