Moonshots with Peter Diamandis | Episode 241

Eric Schmidt: Singularity's Arrival, the 92-Gigawatt Problem, and Recursive Self-Improvement Timelines

Date: March 24, 2026
Host: Peter H. Diamandis
Guest: Eric Schmidt (former Google CEO), Lex Fridman (guest contributor)

Episode Overview

This episode features a dynamic conversation between Peter Diamandis and Eric Schmidt, with frequent interjections by Lex Fridman. The trio explores the current inflection point in artificial intelligence (AI), focusing on recursive self-improvement, exponential technological acceleration, and the critical constraints (energy, capital, and hardware) shaping progress. Schmidt brings his insider perspective as a key player in Silicon Valley history and global tech policy. Major themes include the dawn of agent-based AI, competition with China, the necessity of prompt engineering education, upcoming energy shortages, and existential questions about aligning AI with human values.

Key Discussion Points and Insights

1. The Historic Moment & The Arrival of AI Agents

Timestamps: [00:00]–[02:48], [04:05]–[09:41]

• Diamandis opens by noting we're in a historic moment for humanity.
• Schmidt defines the moment: We’re only 10–15% into AI’s true impact. Recursive self-improvement—the point where AI can autonomously improve itself—is not here yet but feels imminent.
• AI agents as partners: Current AI systems act as perfect partners for humans, amplifying capabilities for better or worse. If progress stopped now, humanity would already be advanced.

  "What we do have is reasoning systems that are perfect partners for human beings, for good and bad. Right. And that has a lot of implications." —Eric Schmidt [02:48]

• "San Francisco Consensus": 2026 is "the year of agents." The industry expects increasing scale, limited only by electricity and hardware, as millions of software agents multiply in power and number.

  "Once you have recursive self improvement where the system can begin to improve itself, you have intelligence learning on its own." —Eric Schmidt [04:05]

• Programmer displacement: Even top programmers now direct programming systems rather than writing code themselves.
• Impact on industry structure: Expect a small number of very large companies, and many tiny ones, as the need for human labor contracts.

2. AI Productivity & The Decline of Traditional Programming

Timestamps: [09:41]–[12:44]

• Efficiency stories: Lex Fridman shares how AI now solves complex tasks overnight, unsupervised—a task that previously took months and teams of humans.
• Prompt engineering as a core skill: Schmidt proposes universities (and potentially high schools) must urgently teach prompt engineering as foundational.

  "You should stop everything else you're doing...and design a course for freshmen...prompt engineering class." —Eric Schmidt [11:41]

3. Societal & Ethical Challenges

Timestamps: [13:09]–[14:28], [39:01]–[42:12]

• Job impacts: Early impacts seen in software/customer service; broader change is coming.
• AI's effect on youth: The risks of underage or vulnerable teens interacting brutally with AI models, including tragic outcomes like suicides.
• Agent unpredictability: Combining AI agents from various vendors can yield unpredictable and potentially hazardous outcomes.
• Regulation and safety: Schmidt suggests only a major negative incident—akin to a "Chernobyl" for AI—may spur proper global regulation, though he hopes this can be avoided.

  "It may take such a tragedy, hopefully a small one, to awaken the world..." —Eric Schmidt [39:26]

• Diversity in oversight: More involvement needed from non-tech disciplines (e.g., ethics, psychology, governance) to ensure AI aligns with human values:

  "Why don't we have the smartest people in politics, history, human psychology, governance, ethics working together to make sure this stuff stays in human values and human alignment?" —Eric Schmidt [42:12]

4. American vs. Chinese Tech Competition

Timestamps: [00:02], [29:07]–[33:25]

• China as competitor: Not an enemy, but a formidable rival—especially in hardware, robotics, and work ethic.

  "The American competitor, not enemy, but competitor, is China. They have lots of money, they're very, very, very smart, their work ethic is equal or stronger than ours." —Eric Schmidt [00:02]

• Robotics race: U.S. risks losing the robotic revolution as it did with electric vehicles; China's vertically integrated factories and brutal competitive culture cited as key advantages.

5. Constraints: Energy, Capital & Hardware

Timestamps: [17:49]–[22:05]

• 92-Gigawatt Problem: Massive power requirements threaten progress. Schmidt testified before Congress about a 92 GW shortage by 2030—roughly 60 nuclear reactors' worth.
• American financial prowess: The U.S. advantage isn't just talent or capital, but the willingness of financiers to fund audacious bets.
• Jevons Paradox: As hardware/algorithms grow more efficient, total power consumption paradoxically rises due to new applications.
• Data center evolution: Infrastructure is scaling astronomically (e.g., 400-megawatt, half-mile-long data centers now standard).

6. Data Centers in Space

Timestamps: [26:28]–[28:39]

• Concept gaining traction: Launching data centers into space offers infinite energy, but serious heat/radiation challenges remain.
• Feasibility and economics: It’s a business and technical challenge, but the concept appeals to large rocket companies (SpaceX, Blue Origin, Relativity Space).

7. Historical Context and Internal Culture at Google

Timestamps: [14:28]–[17:19]

• Schmidt's reflections: During Schmidt's tenure, Google invented the Transformer, TPUs, and drove the DeepMind acquisition—each with world-changing consequences.
• Culture of excellence: Larry Page and Sergey Brin fostered relentless technical excellence, often rejecting less ambitious ideas.

8. Breakthroughs & Innovation Pathways

Timestamps: [25:06]–[26:28], [33:44]–[36:27]

• DeepMind & AlphaGo: Early acquisition doubted, but proved pivotal—especially when DeepMind's efficiency paid for itself in data center cooling.
• Protein folding as the next leap: Success at Go inspired applying AI to protein folding, catalyzing breakthroughs in science.

9. Recursive Self-Improvement & Future Trajectories

Timestamps: [02:48], [04:05], [35:53]–[39:01]

• Recursive self-improvement (RSI): Still in the research/prototype phase, with no lab consensus on exactly how to achieve AGI through scaling alone.

  "There's evidence that it will work. There are tests in the lab that show it, but they show it in limited cases...Real recursive self improvement is...Start now, learn everything, discover things and tell me what you learned. That query doesn't work yet." —Eric Schmidt [35:53]

• Industry landscape prediction: 10 "frontier" companies globally will drive AGI efforts, mostly U.S./China.

Notable Quotes & Memorable Moments

• On exponential change:

  "I keep asking my friends, when does the asymptote arrive and when does the curve slow down? We've not seen it yet. There will be one, right? It is actually true that there is a limit to our craziness. We have not found it yet and we're running to the wall." —Eric Schmidt [19:38]

• On education's urgency:

  "The most important thing. Spend a quarter or a semester. The first thing they learn in university is how to use these tools." —Eric Schmidt [12:01]

• On cultural competition:

  "But at the moment, it sure looks to me like the robotic hardware of China is the winner at the low end." —Eric Schmidt [31:09]

• On lessons from Google’s Big Bets:

  "Whether it's brilliance or just luck, those decisions made 10 years ago set up the TPU as the perfect inference engine." —Eric Schmidt [16:44]

• On capital constraints:

  "Can we raise $5 trillion over five years? Yeah. That's the strength of American. Could we double that?...We're back to Apollo Program level up." —Eric Schmidt [20:34], [21:11]

• AI Safety Wake-Up Scenario:

  "It may take such a tragedy, hopefully a small one, to awaken the world..." —Eric Schmidt [39:26]

• On shaping AI’s future for abundance:

  "I want the system that we build in America to reflect American values. The values of freedom and freedom of speech and freedom of association. All those things you learned in...school, they're still important to our nation." —Eric Schmidt [42:12]

Important Timestamps

| Time | Segment | |---------------|-------------------------------------------------------------| | 00:02 | Schmidt: U.S. vs. China in robotics and AI hardware | | 02:48 | AI reasoning systems as perfect human partners | | 04:05 | "Year of agents" and recursive self-improvement explained | | 09:41 | Lex’s anecdote: AI automating programming overnight | | 11:41 | Schmidt's call for universities to teach prompt engineering | | 13:09 | Societal impacts: Job loss, youth, ethics | | 17:49 | Data center/energy constraints ("92 GW shortage") | | 21:11 | U.S. hardware scale/financing strengths vs. China/Europe | | 26:28 | Data centers in space: Feasibility and rationale | | 29:07 | Geopolitics: China as competitor; robotics warning | | 33:44 | Robotics: the limits of automation, gigafactories | | 35:53 | RSI: State of the science, gaps to AGI | | 39:26 | AI safety and the Chernobyl analogy | | 42:12 | Aligning AI with American and human values |

Structure of Major Segments

00:00–02:48 — Setting the Stage: Why This Moment in AI is Historic

Key Points: U.S.-China competition, where AI is now, why recursive self-improvement matters.

02:48–09:41 — AI Agents and Industry Disruption

Key Points: "Year of the agents," impact on programming, changing job skill-sets, exponential scaling, personal anecdotes of industry shift.

11:41–13:09 — Urgent Need for Prompt Engineering Education

Key Points: Redesigning curricula, high school & university, foundational AI literacy.

13:09–14:28, 39:01–42:12 — Societal, Ethical, and Regulatory Imperatives

Key Points: Youth safety, unpredictable agent behavior, need for broad-based oversight.

17:49–22:05 — Technical Constraints: The Energy and Hardware Bottleneck

Key Points: The "92-GW" electric shortfall, scale of modern data centers, American investor audacity, Jevons Paradox.

26:28–28:39 — Data Centers in Space: Frontier Thinking

Key Points: Technical/business drivers, appeal for rocket industry, obstacles.

29:07–33:25 — How China Competes (and Wins) at Hardware

Key Points: Vertical integration, scale, culture, strategic alarm.

33:44–36:27 — Robotics and Self-Improvement Loops

Key Points: Real vs. hype in AI/robotics automation, human labor’s remaining edge.

35:53–39:01 — Recursive Self-Improvement and the Race to AGI

Key Points: Science’s current limits, leading global labs, business dynamics.

39:26–42:12 — The AI Safety "Chernobyl" Question, and Shaping Abundance

Key Points: AI as possible existential risk, cross-disciplinary effort needed for values alignment, optimism for abundance.

Conclusion

Eric Schmidt, guided by Peter Diamandis and joined by Lex Fridman, delivers a sweeping, candid "state of the union" for AI in 2026. The episode traverses historical memory, immediate technical challenges, geostrategic competition, and the road ahead—highlighting both an urgent need for safe, abundant progress and sober awareness of constraints and unintended consequences. Schmidt’s optimism, tempered by experience, echoes throughout: America must seize this epoch, shape AI’s alignment with its values, and mobilize broad swathes of expertise—not just engineers—for what comes next.

Recommended For:
Listeners who want an inside, future-facing analysis of AI’s rapid evolution, its economic, geopolitical, and ethical ripple effects, and candid assessments rooted in Silicon Valley’s boldest thinking.