Nov 2, 2026 · 34 min
AI data centers force a new definition of energy security
Ep596 Energy security in the age of AI: What do we need for a more resilient grid?
As electricity demand, cyber threats, geopolitical conflict, and supply disruptions converge, grid resilience increasingly depends on digital infrastructure and flexible demand.
- 1Energy security now spans fuel, infrastructure, software, supply chains, and the digital systems that coordinate them.
- 2Integrated historical and real-time data can expose operational shocks earlier and help companies respond across logistics, inventory, and markets.
- 3AI data centers should operate as flexible industrial assets that strengthen grid reliability rather than simply intensify electricity demand.
Don't miss
Caspar Herzberg argues that large AI data centers should become flexible grid assets by managing workloads, cooling, procurement, and local power conditions.
The brief
Caspar Herzberg, CEO of AVEVA, broadens energy security beyond fuel supplies to include the resilience of assets, supply chains, software, and digital infrastructure.
The episode argues that breaking down data silos can reveal disruptions earlier, connecting logistics, inventory, operations, and market signals before shocks spread.
AI data centers are emerging as major energy consumers and critical targets, making redundancy, local capacity, regulation, and workload visibility central to grid security.
The proposed answer is not a single grid model but a mix of distributed and interconnected systems, supported by long-term investment and political cooperation.
The standout idea is that data centers could actively manage cooling, workloads, procurement, and supply to support the wider grid instead of merely drawing from it.
Herzberg’s broader outlook is a more digitized, distributed, and electrified energy system that preserves multiple generation and infrastructure options.
What was said on this episode
5 statements · 5 positive
Data centers can preserve five-nines reliability while gaining operational visibility.
“you can have the five nines, right? But you can still get visibility into the data center's operations itself.”
Listen at 30:08
Operational visibility enables data centers to plan energy demand and return energy to the grid.
“the more you are able to plan for it. And that means also then the more you will know when do you need actually a lot of energy and when can you give back some of the energy that you are producing back to the grid.”
Listen at 30:43
- Caspar Herzbergon Data-center energy managementPositive30:55
Flexible data-center energy management is achievable.
“I think that is absolutely achievable.”
Listen at 30:55
- Caspar Herzbergon Data-center energy-management softwarePositive31:19
Operational energy-management software will become increasingly common in data-center construction.
“it's going to become more and more present in the built out of data centers.”
Listen at 31:19
Data centers will increasingly understand their demand and supply patterns.
“I think the first step is that data centers will understand their demand and own supply patterns much, much better”
Listen at 31:40
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.