Nov 12, 2026 · 1h 8m
AI’s energy promise depends on hardware, data and control
AI in energy: How can new technology strengthen the resilience of energy systems? | From ADIPEC in Abu Dhabi
As electricity demand rises and geopolitical disruption strains infrastructure, the episode examines what it takes to deploy AI safely across critical energy systems.
- 1Industrial AI needs trusted operational data, deterministic controls and human accountability before it can manage critical processes.
- 2Standardized small modular reactors could improve nuclear economics, but policy certainty and repeatable construction remain essential.
- 3AI-driven data-center growth is tightening transformer supply chains, making capacity expansion and long-term planning strategic priorities.
Don't miss
Bruno Melles explains why the surge in AI-related electricity demand is exposing transformer capacity as a strategic constraint on energy resilience.
The brief
At ADIPEC in Abu Dhabi, energy leaders frame AI as both an opportunity and a constraint: smarter systems can strengthen resilience, but only when reliable physical infrastructure keeps pace.
Per Erik Holsten argues that regional energy strategies will mix gas, nuclear and renewables differently, while standardized small modular reactors could make nuclear more competitive if policy supports repetition.
Chris Rezendes makes the data case: industrial AI depends on provenance, harmonization and real-world sensor data, turning emissions measurement and asset information into operational and financial advantages.
Jim Masso insists that AI in hazardous facilities must sit behind deterministic controls and explainable recommendations, preserving operator accountability rather than handing critical decisions to probabilistic systems.
Bruno Melles closes on the physical bottleneck: data-center demand is tightening transformer supply, so manufacturing expansion, supplier investment and long-term agreements become central to resilience.
What was said on this episode
17 statements · 13 positive · 2 negative · 1 mixed · 1 neutral
- Per Erik Holstenon Global electricity demandPositive4:22
Global electricity demand will rise 3.6% annually over the next decade, doubling from current levels.
“electricity demand there is out there, raising by 3.6% a year. over the next decade, meaning doubling of what it has been.”
Listen at 4:22
- Per Erik Holstenon SMR deploymentsPositive12:26
First operational SMR deployments will begin around 2030–2035.
“We see already now probably the 20, 30, 2 to 35 timeframe where you start to see the first deployments of an operational”
Listen at 12:26
- Chris Rezendeson Real-world industrial asset dataPositive28:41
Real-world asset data is the best input source for industrial AI.
“The best data comes from organic operations. It comes from your well heads. It comes from your compressors, your pumps, your fans, your blowers. It comes from your real world assets and the data that is monitoring those assets. That's the best source of data to feed AI.”
Listen at 28:41
AI agents are about to become a third economic actor alongside humans and machines.
“We are now absolutely on the cusp of adding a third entity or type of actor in our economy, and it's the agent.”
Listen at 31:08
- Chris Rezendeson AI in energy capital allocationPositive33:44
AI will let companies analyze unlimited energy investment and allocation scenarios.
“the application of AI intentionally will give companies the ability to analyze unlimited scenarios for how they might allocate capital to a broader collection of production, transportation, or application or delivery assets.”
Listen at 33:44
- Chris Rezendeson Autonomous industrial operationsPositive34:49
Autonomous operations will radically reduce industrial maintenance truck rolls.
“the future in autonomous operations will radically reduce not only the need for truck rolls”
Listen at 34:49
- Chris Rezendeson Predictive autonomous asset managementPositive35:48
Predictive autonomous asset management will increase uptime and reduce failure and repair times.
“Increase uptime, reduce mean time to failure, mean time to repair, all that.”
Listen at 35:48
- Jim Massoon Industrial AINegative40:26
Industrial AI currently has a credibility problem due to failed initiatives.
“industrial AI has a credibility problem.”
Listen at 40:26
- Jim Massoon Industrial AIPositive41:45
Industrial AI works best when grounded in first-principles engineering and equipment operation.
“It's first principles. It's how equipment actually works and using AI to augment that operation versus just AI alone.”
Listen at 41:45
- Jim Massoon Industrial automationPositive43:14
Industrial automation made plants safer, more reliable, predictable, and financially effective.
“automation changed not only the quality of operation, the effectiveness, the financial outcome of the plant, but simplified the operation so it was safer, more reliable, more predictable.”
Listen at 43:14
Current AI cannot produce outputs uninformed by historical data or prior examples.
“AI is not going to do anything that hasn't been informed by previous history or previous data.”
Listen at 51:16
- Jim Massoon Industrial AIPositive52:29
Industrial AI is rapidly expanding from control-room tools to enterprise capabilities.
“We're finding this is developing pretty quickly from a control room capability to an enterprise capability.”
Listen at 52:29
- Jim Massoon Industrial AI adoptionMixed52:56
Industrial AI early adopters will see significant gains, while others will not.
“you will see these early adopters have significant gains if they pick the right approach, the right structure. You'll see many others who've spent a lot of money not seeing the gains.”
Listen at 52:56
- Jim Massoon Industrial AI early adoptersPositive53:11
Successful early AI adopters will achieve significant financial results and adopt enterprise infrastructure.
“They're going to see significant financial results, and they're going to adopt this enterprise infrastructure that goes all the way back to the control room.”
Listen at 53:11
- Jim Massoon Industrial control-system architectureNeutral54:14
Control-system architecture will change substantially over the next five years.
“I think there's going to be a lot of change in the core architecture of control systems over the next five years that people aren't ready for.”
Listen at 54:14
- Bruno Melleson Solid-state transformersPositive1:05:40
Solid-state transformers will evolve strongly and attract substantial investment.
“we will see a very, very strong evolution. We will see a lot of investment”
Listen at 1:05:40
- Bruno Melleson Transformer supply lead timesPositive1:07:19
Framework agreements and long-term planning will help reduce transformer lead times.
“the frame agreement, the long-term planning are going to be key to ensure that the lead time stay and get reduced on a reasonable level.”
Listen at 1:07:19
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.