Field notes The Energy Transition for the Rest of Us

Threads

AI compute

29 notes · Catalyst 18 · Critical Capital 3 · Steel For Fuel 8

  1. Steel For Fuel · 21 Sep 2026 For AI, energy is nothing, and energy is everything (reprise) by Andy Lubershane, Partner and Head of Research, Energy Impact Partners How much could an AI data center pay for electricity before the computing stops being worth doing?
  2. Catalyst · 3 Sep 2026 What comes after the data center backlash? with Brian Janous, co-founder and chief commercial officer, Cloverleaf Infrastructure Is the backlash against data centers something developers can actually fix, and if it is not, where does the buildout go instead?
  3. Critical Capital · 18 Aug 2026 AI is disrupting services. What does that mean for power demand? with Jake Saper, partner, Emergence Capital; enterprise software and AI infrastructure investor who began his career developing solar in India. The transcript renders his surname as both “Saper” and “Saber”; this note uses the dominant rendering. If AI starts to change how services such as law, accounting and insurance are actually delivered, what does that imply for how much compute, and how much power, the economy will end up needing?
  4. Catalyst · 6 Aug 2026 The electrician shortage slowing the AI boom with Raymond Hawkins, chief customer officer, Compass Datacenters How much is the shortage of electricians actually holding back data center construction, and what can anyone do about it?
  5. Steel For Fuel · 5 Aug 2026 On data centers and energy “affordability” by Andy Lubershane, Partner and Head of Research, Energy Impact Partners Are data centers the reason American electricity bills are going up, and is the country facing an affordability crisis?
  6. Catalyst · 9 Jul 2026 Inside the AI power wars with Jeremie Eliahou Ontiveros, leads infrastructure and power research, SemiAnalysis Everyone can describe how the AI companies differ on chips. How do they differ on power?
  7. Steel For Fuel · 6 Jul 2026 Data centers need the grid — so we have to make the grid better by Andy Lubershane, Partner and Head of Research, Energy Impact Partners If the grid cannot connect AI data centers fast enough, is building a power plant next to each one the way out?
  8. Critical Capital · 23 Jun 2026 The trillion-dollar collision between AI and the grid with Varun Sivaram, founder and CEO, Emerald AI; previously an executive at two clean energy developers and a senior advisor at the US State Department Can the grid absorb the data centers that AI wants to build, and if the answer is no, what is the cheapest thing to change?
  9. Steel For Fuel · 18 Jun 2026 The next industrial revolution and the “art of life itself” by Andy Lubershane, Partner and Head of Research, Energy Impact Partners If AI is the next industrial revolution, will it be one that makes existing work more efficient, or one that gives us capabilities we did not have?
  10. Catalyst · 28 May 2026 Building inference data centers on the high seas with Garth Sheldon-Coulson, co-founder and CEO, Panthalassa Could you power AI compute by putting the generator and the servers together on untethered steel structures bobbing in the deep ocean, and would it pencil?
  11. Critical Capital · 28 Apr 2026 Atoms, bits, and the trillion-dollar manufacturing race with Aidan Madigan-Curtis, partner, Eclipse Ventures; previously at Apple, then five or six years at Samsara Who wins when the digital economy runs into the physical one, and what would the United States have to do to compete again at making, powering and moving things?
  12. Catalyst · 23 Apr 2026 Live from Transition-AI 2026: Inside Google’s massive AI capex with Amin Vahdat, chief technologist for AI infrastructure, Google As AI shifts from training models to serving them, does that change the size, the reliability and the power profile of what Google builds with the $175 to $185 billion of capital expenditure it has planned for this year?
  13. Catalyst · 9 Apr 2026 The rise of flexible data centers with Varun Sivaram, CEO, Emerald AI Almost everyone now agrees data centers should be flexible grid assets. What is actually stopping it from happening at scale?
  14. Catalyst · 12 Mar 2026 AI scaling pathways: On grid, on edge, off grid, off planet with Jake Elder, senior vice president of research and innovation, Energy Impact Partners If demand for compute keeps climbing, which ways of siting and powering data centers can actually scale: large grid-connected sites, small sites at the edge, off-grid sites, or data centers in orbit?
  15. Catalyst · 18 Dec 2025 Will inference move to the edge? with Dr. Ben Lee, professor of electrical engineering and computer science, University of Pennsylvania; visiting researcher, Google How much AI inference will move out of giant centralized data centers into smaller local ones or onto personal devices, and what would that do to the energy picture?
  16. Catalyst · 4 Dec 2025 Who benefits from the AI power bottleneck? with Shanu Mathew, senior vice president and portfolio manager-analyst for US sustainable equity, Lazard Everyone in the AI and energy business says power is the bottleneck. Whose interests does that story serve, and how should you discount what you hear?
  17. Catalyst · 6 Nov 2025 Inside a $300 million bet on AI for physical R&D with Ekin Dogus Cubuk, co-founder, Periodic Labs; formerly Google DeepMind A year ago Cubuk was sober about using AI for materials discovery. What changed enough in twelve months to justify leaving Google DeepMind and raising $300 million to do it?
  18. Catalyst · 23 Oct 2025 Five big questions about the future of energy with Andy Lubershane, head of research and partner, Energy Impact Partners Five separate questions about where the AI-driven electricity boom goes next: when could today’s shortage turn into a glut, who wins and loses from it, and what should we expect from nuclear, enhanced geothermal and grid-enhancing technologies?
  19. Catalyst · 28 Aug 2025 The mechanics of data center flexibility with Varun Sivaram, founder and CEO, Emerald AI What does it actually take to make an AI data center flexible, and how much flexibility is really there?
  20. Steel For Fuel · 16 Jun 2025 Why does nobody know how much energy AI will consume? by Andy Lubershane, Partner and Head of Research, Energy Impact Partners Why do the analysts watching data centers disagree so widely about how much electricity those data centers will consume, even a few years out?
  21. Catalyst · 12 Jun 2025 The state of play of data center development with Chris Sharp, chief technology officer, Digital Realty Is the AI buildout spreading data centers out to wherever cheap power happens to be, or concentrating them further into the regions that are already full?
  22. Catalyst · 6 Mar 2025 A skeptic’s take on AI electricity load growth with Jonathan Koomey, founder and president, Koomey Analytics How much electricity will AI data centers actually use, and how much weight should anyone put on the forecasts saying the answer is enormous?
  23. Catalyst · 6 Feb 2025 Catalyst: 2025 Climate Tech Trends with Nat Bullard, co-founder, Halcyon; former chief content officer, BloombergNEF Entering 2025, what do the underlying data actually show about where energy and climate stand, and does a sudden jump in AI model efficiency break the load-growth story?
  24. Catalyst · 5 Dec 2024 Explaining the ‘Watt-Bit Spread’ with Brian Janous, co-founder and chief commercial officer, Cloverleaf Infrastructure; previously vice president of energy at Microsoft Why can data centers not get power, when the companies that want it are willing to pay almost anything for it?
  25. Steel For Fuel · 20 Nov 2024 For AI, energy is nothing, and energy is everything by Andy Lubershane, Partner and Head of Research, Energy Impact Partners If energy is only a small part of what computing costs, why is the AI industry suddenly worried about power, and what should the power sector do with that?
  26. Catalyst · 22 Oct 2024 Inside Crusoe’s energy-first approach to data centers with Chase Lochmiller, co-founder and CEO, Crusoe If you site computing around energy instead of siting energy around computing, what changes?
  27. Steel For Fuel · 26 Jul 2024 AI versus computing efficiency by Andy Lubershane, Partner and Head of Research, Energy Impact Partners Is computing hardware getting efficient fast enough to offset how fast AI is adding electricity demand?
  28. Catalyst · 13 Jun 2024 Under the hood of data center power demand with Brian Janous, co-founder, Cloverleaf Infrastructure; former vice president of energy, Microsoft Granting that AI will pull a large amount of new load onto the grid, how do we actually serve it, and what does serving it do to the grid and to the tech companies’ climate commitments?
  29. Steel For Fuel · 6 Jul 2023 Some “complete bullshit” about AI and energy by Andy Lubershane, Partner and Head of Research, Energy Impact Partners Would pursuing human-level AI break the energy system, and does the energy industry actually need models that big?