The HC Commodities Podcast

Episode: Energy & AI with the IEA’s Thomas Spencer and Siddarth Singh

Host: Paul Chapman, HC Group
Guests: Thomas Spencer and Siddarth Singh, Lead Authors of IEA’s "Energy and AI" Report
Release Date: June 17, 2025

Episode Overview

This episode dives deep into the transformative interplay between artificial intelligence (AI) and the global energy sector, guided by the International Energy Agency’s (IEA) 2025 landmark report "Energy and AI". Host Paul Chapman is joined by the report’s lead authors, Thomas Spencer and Siddarth Singh, who illuminate both the major anxieties—such as fears of data center-driven power crises—and the nuanced efficiencies AI promises to unlock across industries. The discussion spans from data center proliferation and power grid pressures to global consumption forecasts, AI-induced uncertainties, and the often-overlooked efficiency and innovation boons AI brings to energy systems worldwide.

Key Discussion Points & Insights

1. The State of AI: Investments, Market Scale, and Data Center Growth

• Rapid AI Progress & Investment Surge

  • AI capabilities are advancing quickly, with "new announcements almost every day" (Thomas, 02:10).
  • Approx. US$500 billion to be invested in new data centers in 2025 alone.
  • 60% of market capitalization growth since ChatGPT’s launch was from AI-related firms, peaking at US$12 trillion (Thomas, 02:10).

• Geographic Breakdown of Data Centers

  • US leads with 40% of installed data center capacity; China at 25%, Europe at 15% (Thomas, 04:09).
  • Data center electricity use: Global—415 TWh in 2024 (190 TWh US alone), ~1.5% of global electricity (Siddarth, 05:44).

2. Contextualizing Data Center Electricity Demand

• Relative Electricity Use

  • Data centers: 1.5% of global electricity (2024), expected to double by 2030 to 950 TWh.
  • By 2030, data centers could consume more power than present-day Japan (Siddarth, 05:44).
  • Notably, AI-related workloads ("accelerated servers") will quadruple their electricity demand by 2030.

• Hyperscaler Data Centers

  • “A hyperscaler data center today may consume as much electricity as 100,000 households, but the largest under construction could consume as much as 2 million households." (Siddarth, 06:43)

3. Data Centers’ Outsized Impact in Developed Economies

• US Example
  • Over 45% of US electricity demand growth to 2030 is from data centers (Siddarth, 08:56).
  • By 2030, data centers will consume more electricity than all US heavy industries combined.
  • Clustering: Data centers "tend to be very clustered around each other" and close to major cities for low latency (Siddarth, 10:37).

4. Meeting Soaring Data Center Demand: All Hands On Deck

• Peak vs. Annual Demand

  • Electricity grids are designed for peak—not average—demand.
  • US data centers’ share of peak demand: "from 6% now to up to 15% by 2030," rivaling total industrial peak demand (Thomas, 12:00).

• Energy Mix for New Demand

  • 50%+ of incremental electricity for data centers anticipated from renewables; rest from natural gas, and possibly SMRs (Small Modular Reactors) and advanced geothermal long-term (Thomas, 13:35).
  • “We need an all hands on deck situation. We need contributions from all sources. But it’s also a smarter-is-faster situation.” (Thomas, 12:44)

5. Integration Challenges and Uncertainties

• Grid vs. Behind-the-Meter

  • Most tech companies prefer grid connection over “behind-the-meter” (on-site) generation due to cost and security benefits (Thomas, 18:14).
  • Utilities also prefer grid integration (Paul/Thomas, 20:02).

• Forecasting Uncertainties

  • Difficulty measuring/forecasting precise AI-related demand due to lack of data, differences in data center types, variances in reporting and hardware efficiency (Siddarth, 21:42).
  • “What is precisely the data center related electricity consumption, of which what part comes from artificial intelligence? It is very much a black box.” (Siddarth, 21:53)

• Scenario-Based Projections

  • Base Case: Data center electricity rises from 415 TWh (2024) to 945 TWh (2030).
  • Lift-Off Case: 30–50% higher demand if constraints ease and AI adoption outpaces expectations.
  • High Efficiency Case: Advances in efficiency could offset demand growth.
  • Headwinds Case: Delays in energy infrastructure could leave 20% of capacity at risk (Siddarth, 25:49).
  • "The idea is never to present a single view of the world, but to present different alternatives which can help peg our expectations." (Siddarth, 27:21)

6. Data Sovereignty & Security

• Localization Trends
  • National data sovereignty leads to requirements for local data storage, driving more domestic data center build-outs.
  • But: reliability of electricity supply (not just data policies) heavily influences where data centers are sited—outages in emerging economies are a major barrier (Siddarth, 28:26).

7. AI’s Efficiency Potential Across the Energy Chain

• AI as Optimizer and Innovator

  • Power Sector:
    • Advanced weather forecasting for renewables (hourly to long-range accuracy).
    • Real-time monitoring of power lines, using AI to dynamically boost grid capacity.
    • Automating smart consumption response (e.g., EV charging, building HVAC) (Thomas, 31:25).
  • Resilience:
    • Monitoring infrastructure using AI-enabled sensors and drones aims to prevent or swiftly correct faults.

• Industrial and Societal Efficiency Gains

  • Oil & Gas: AI used for seismic analysis, extraction optimization, methane leak detection (Siddarth, 36:46).
  • Transport: Route optimization by AI for freight could save energy equivalent to “120 million cars” (Siddarth, 38:50).
  • Aviation: AI-green route selection could cut “contrail-related warming by 50% with only 0.5% more energy input.” (Siddarth, 39:57)
  • Urban Mobility: AI-optimized traffic lights massively reduce idling and thus wasted fuel (Paul, 40:30).

8. The Macro Balance: Will AI Consume or Save Net Energy?

• Potential for Net Energy Savings… with Big Ifs

  • “The decrease in energy consumption from AI-driven efficiencies could outweigh data center consumption increases—but only if solutions are widely adopted.” (Thomas, 41:50)
  • Critical challenges: skills gaps, lack of digital infrastructure (e.g., only half of commercial-sector HVAC is digitized in advanced economies), and regulatory barriers (data privacy/liability, critical infrastructure risks).
  • Quote: “No energy market operator wants to be on the hook for a blackout… because the AI algorithm did something wrong.” (Thomas, 44:33)

• Rebound Effects & Uncertainty

  • Long-term: Increased economic growth/productivity from AI could result in greater energy demand (rebound effect), especially with trends like autonomous vehicles (Thomas, 45:52).
  • "The big picture… our assessment is that the energy optimization potential of AI exceeds, at least in the near term, the increase in energy demand from AI. But it’s by no means a given that that optimization potential is tapped." (Thomas, 46:51)

9. AI as Catalyst, not Panacea, for Energy Transition

• Workforce Call-to-Arms

  • Demand for tech talent in energy is soaring; a need to highlight the critical role AI can play in the accelerated transition (Paul, 46:52).
  • “AI has turned up right at the perfect time to actually accelerate the energy transition… this is where you can make the biggest difference.” (Paul, 47:44)

• Innovation Engine, Not Magic Bullet

  • AI can radically speed up innovation, e.g. drug discovery (AlphaFold), or screening for new energy materials like perovskites, but “Even with high performing low emissions technologies, we’d still have to do what we’ve been needing to do for some time.” (Thomas, 49:10)
  • “For us, AI is definitely an enabler of the energy transition, but it’s not a silver bullet.” (Thomas, 51:01)

Notable Quotes & Moments

• On Market Scale:
  “About 60% of the market capitalization increase that took place since the launch of ChatGPT was from firms that had a value proposition related to AI… almost US$12 trillion.” — Thomas (02:10)

• On US Data Centers:
  “In the year 2030, more electricity will be consumed for processing data… than by all the heavy industries put together.” — Siddarth (09:37)

• On Power Grid Strain:
  “Within the space of a few years, data centers moving from 6 to about 15% of peak demand in the United States. That’s a huge increase.” — Thomas (12:00)

• On Forecast Uncertainty:
  “What is precisely the data center related electricity consumption, of which what part comes from artificial intelligence? It is very much a black box.” — Siddarth (21:53)

• On AI’s Efficiency Potential:
  “AI-related route optimizations can reduce over 50% of the condensation trails in aviation, at only 0.5% more energy.” — Siddarth (39:57)

• On Macro Impact:
  “The energy optimization potential of AI exceeds, at least in the near term, the increase in energy demand from AI. But it’s by no means a given that that… is tapped.” — Thomas (46:51)

• On Policy Needs:
  “AI is not a silver bullet for the energy transition… but it can be a catalyst, an enabler of the changes we need.” — Thomas (48:13, 51:01)

Timestamps for Major Segments

• AI Investment & Industry Overview — 02:10
• Data Center Geography & Consumption Data — 04:09
• Power Demand Forecasts & Context — 05:44
• Developed World Infrastructure Challenges — 08:56
• Meeting Power Demand (Energy Mix) — 12:00 – 16:39
• Uncertainties & Scenario Forecasting — 21:42 – 27:59
• Data Sovereignty & Outage Constraints — 28:26
• AI for Efficiency: Power, Industry, Mobility — 31:25 – 41:50
• Barriers to Realizing AI Savings — 41:50 – 46:51
• AI as Enabler (Not Silver Bullet) — 48:13 – Epilogue

Conclusion

The episode delivers a rigorous, insightful journey into how AI is both a profound new source of electricity demand—especially via mushrooming data centers—and a complex new toolkit for driving efficiency and innovation across the energy system. While the data center boom presents serious infrastructure and sustainability challenges, AI’s ability to optimize everything from renewables to logistics to industrial operations could ultimately deliver net energy savings, provided regulatory, infrastructure, and skills hurdles are overcome. In the words of the IEA authors, AI is "an enabler of the energy transition, but it’s not a silver bullet." The next decade’s outcome will depend on our collective ability to integrate, regulate, and adapt.

For a detailed look, listeners are encouraged to read the full IEA report “Energy and AI.”