Motley Fool Money – "AI’s Power Problem" (October 27, 2025)

Host: Tim Byers
Guests: Nick Seiple, Seth Jason

Episode Overview

This episode dives deep into a less-discussed but critical facet of the ongoing artificial intelligence (AI) revolution: the immense power demands generated by AI, cloud computing, crypto, and broader electrification trends. Analysts Nick Seiple and Seth Jason join host Tim Byers to explore whether the energy grid can realistically keep up, how utility and tech giants are adapting, and how nuclear power—especially small modular reactors (SMRs)—might fit into the future energy equation for supporting AI.

Key Discussion Points & Insights

The Exponential Rise in Power Demand from AI

• Surging Needs: Current estimates say the U.S. will need to add about 80 gigawatts of new capacity annually to keep up with AI and related technologies (00:45).

• Historical Contrast: Nick notes that U.S. electricity demand was nearly flat (+0.1%) from 2005–2020, but that's changing fast due to AI and data center proliferation (01:38).

  "We’ve gone from a regime where we’ve kind of been treading water... to now we have new sources of load coming on from AI that we’re scrambling to meet."
  — Nick Seiple [02:32]

• Data Centers’ Footprint: Data centers were <2% of U.S. electricity demand in 2020; by decade’s end, they're projected to be 9–12% (01:55).

The Reality of Meeting Demand

• Gap Between Promises and Reality: Seth challenges the hype and urgency communicated by AI companies, remarking that some of it is "PR doom"—both to sell the importance of their products and to position themselves as essential (03:29).

  "Every week these AI CEOs are playing the congratulations, you said the biggest number game."
  — Seth Jason [03:32]

• Unknowns Remain: True demand is uncertain—efficiency gains and future software/hardware improvements could dampen the demand curves, but so far both power needs and AI deployments are in a race (05:30).

  "You don’t know when they're going to flatten into S curves, but...where they cross over, one of them is like electrical demand from AI the other one is, is power generation that can serve it. We don't actually know that."
  — Seth Jason [05:05]

• Infrastructure Bottlenecks: Gas turbines (for quick baseload) have 3–4 year backlogs; grid hookups for data centers can take five years (06:47, 10:36).

Nuclear Power & Small Modular Reactors (SMRs)

• SMRs Defined: Smaller, potentially mass-producible reactors (300 MW or less) that could allow for faster, cheaper nuclear power plant construction compared to traditional gigawatt-scale reactors (07:03).

  "The hope is with these small modular reactors, you’ll be able to do that [bring down costs]...if you look in other markets like China, they’ve been able to build nuclear reactors back to back to back using the same workforce."
  — Nick Seiple [08:13]

• Behind-the-Meter Solutions: Co-locating reactors with data centers to avoid grid competition is gaining momentum (09:30).

  "You can fit these on existing sites near the data center and...get around competing with the retail electricity customer for this electricity."
  — Nick Seiple [09:58]

• Case-by-case Complexity: Nuclear buildout or reactivation (like Brookfield’s and Microsoft’s projects) is highly site-specific, rife with regulatory, logistical, and financial hurdles (12:36).

  "It's not nearly as simple as they all make it sound. And Nick has said a bunch of times...most of these are going nowhere. A few of the better ones may get somewhere."
  — Seth Jason [13:34]

• Reactivating Old Plants: Companies are scrambling to reactivate dormant or abandoned plants—reflecting just how acute the power crunch has become (13:56).

  "It’s really all hands on deck trying to meet the power demand coming down the line."
  — Nick Seiple [15:28]

Nuclear Investment Fakers vs. Breakers

Game Segment: "Faker or Breaker" (16:45)

• Oklo (OKLO):
  • Seth: Faker – "In any of this nuclear space, I want a company that knows what they’re doing, has been around the block."
• NuScale Power (SMR):
  • Nick: Faker – Cites cost overruns, lack of real progress, despite "first mover" advantage. "I don't think NuScale is significantly closer to deploying a reactor today than it was five years ago." [18:17]
• Nano Nuclear Energy (NNE):
  • Seth: Faker – "Just about every company you looked at that had the word nano in it was something you needed to scream and run away from." [18:45]

Promising Players (20:00)

• Most credible U.S./Canadian SMR efforts:

  • GE-Hitachi BWRX-300 (Darlington, Canada) with BWX Technologies as a major supplier
  • Military projects (Project Janus): Antares Nuclear, General Atomics, Kairos Power (w/Google), Oklo, Radiant Industries, Westinghouse, X Energy

  "If you look to what the military is doing, it can give you a sign of some of these areas where we might see construction earlier than... SPACs that have come public..."
  — Nick Seiple [20:45]

Sizing Up the Timeline

• Short-term: Expect delays and hype outpacing reality—most public “nuclear startups” are years away, if ever, from deployment (19:14).
• Long-term (by 2030s): Real acceleration in nuclear and next-gen power buildout becomes necessary as AI demand outstrips all conventional tricks (15:33).

Notable Quotes & Moments

• "AI CEOs are playing the congratulations, you said the biggest number game."
  — Seth Jason [03:32]
• "It’s really all hands on deck trying to meet the power demand coming down the line."
  — Nick Seiple [15:28]
• "I love that everybody's so excited...but I'll believe it when I see it."
  — Seth Jason on Oklo [16:49]
• "Nano...was something you needed to scream and run away from."
  — Seth Jason on Nano Nuclear Energy [18:45]

Key Timestamps

| Timestamp | Topic | |-----------|-------| | 00:45 | AI and projected power generation needs | | 01:38 | Historical context: flat demand, now inflection | | 03:29 | Seth on AI company hype and reality | | 05:30 | Uncertainty of actual future demand and infrastructure constraints | | 07:03 | SMRs explained: what they are and why they're different | | 09:30 | The behind-the-meter nuclear/data center trend | | 10:36 | Grid connection delays as a major obstacle | | 12:36 | Site-specific nuclear plant reactivation issues | | 13:56 | Industry-wide “all hands on deck” for new and old power sources | | 16:45 | "Faker or Breaker" nuclear company segment | | 20:00 | The most credible nuclear projects and military connection |

Tone & Language

The episode is both sober and skeptical, rooted in technical realities and investing experience, but also wry and playful—especially in the “Faker or Breaker” segment and with pointed jabs at overhyped companies.

Takeaways for Investors

• Cautious Optimism: The huge energy requirements for AI are real, and energy infrastructure companies (especially credible ones in nuclear) may see significant tailwinds.
• Don’t Buy Just On Hype: Many nuclear "startups" are unproven and likely years away from revenue or deployment.
• Watch the Leaders: Keep eyes on GE-Hitachi, BWX Technologies, and companies working closely with governments/military.
• Timeline is Key: This will be a long slog, not an overnight boom—invest accordingly.