Prof G Markets — "How The AI Economy Could Collapse"

Date: September 29, 2025
Hosts: Scott Galloway (Prof G), Ed Elson
Podcast Network: Vox Media Podcast Network

Episode Overview

This episode examines the ongoing explosion of investment in the AI sector—and the potential risk that, underneath the hype and capital flow, the economics are being propped up by circular financing deals that could lead to a collapse reminiscent of past tech and infrastructure bubbles. Scott Galloway and Ed Elson draw parallels to the dot-com bubble, the railroad and electric grid booms, and raise antitrust and macroeconomic concerns around the AI investment mania. The show also touches on the broader state of the U.S. economy, highlighting growing inequality masked by record-breaking indices.

Key Discussion Points & Insights

1. The AI Spending Frenzy and Circular Deal Theory

[05:14–07:36]

• Recent weeks have seen massive AI-related spending news:

  • Nvidia committing $100 billion investment in OpenAI for data centers.
  • Oracle issuing $18 billion in bonds to finance AI spending.
  • OpenAI's Project Stargate launching, with plans for up to $1 trillion in infrastructure spending.

• Circular Deal Theory Explained:

  • Major AI players are investing in each other and then using those investments to buy each other's products and services.
  • Examples: Nvidia invests in OpenAI, OpenAI then buys Nvidia chips; Microsoft invests in OpenAI, which commits to buy compute from Microsoft; similar deals among Oracle, Amazon, Anthropic, etc.
  • This creates an appearance of immense demand and revenue, while the money simply cycles among industry giants.

  “The money is all moving in a circle. And so this got us to be a little bit concerned. It feels a little bit like financial engineering.” — Ed [06:54]

2. Historical Parallels: Echoes of the Dot-Com Bubble

[07:36–12:50]

• Scott likens current AI financial engineering to late-stage 1990s dot-com tactics:

  • AOL would invest in e-commerce companies, which in turn spent the funds on AOL services.
  • These "related party transactions" inflated the illusion of real demand and sustainable growth.
  • The illusion unraveled quickly once real demand failed to materialize.

  “This is kind of what you see in what I call late stage, trying to prop up shit… It's usually indicative of kind of late stage bubble if you will.” — Scott [10:55]

• Scott draws on direct experience:

  • His own ventures saw valuations sink when accounting for related-party revenue.
  • Dot-com hardware vendors (e.g. Cisco, Lucent) propped up telcos with loans; telcos would spend that back on equipment, creating artificial demand that collapsed disastrously when the market turned.

• Ed provides data on the fallout:

  • Over $2 trillion in market cap disappeared post-2000; the Dow Jones Communication Technology Index fell 86%; half a million jobs lost.
  • Many crashed companies (WorldCom, Global Crossing) never recovered, though the underlying technologies eventually did become essential.

  “Everyone's right. Of course, AI is going to happen. But what they're getting wrong is the timing on all of this. They are overspending now, but as soon as the music stops, you're going to see something so similar to what happened in 2000.” — Ed [17:42]

3. Are AI Valuations Getting Ahead of Reality?

[18:32–22:20]

• Scott highlights that AI may become transformational—just as the Internet was—but warns that valuations are far ahead of actual revenue and user adoption.

  • Citing MIT research: large corporations aren't seeing the expected returns from AI investments.
  • Many employees simply aren't using the AI tools, despite expensive enterprise deals.
  • Predicts OpenAI is likely prepping for a public offering to raise real capital, as private funds balk at sky-high valuations.

  “We have these valuations that are unsustainable based on current dollar volume. But we know this is the future. We know it's going to be huge. So let's do some kind of some shell games.” — Scott [20:44]

4. Big Tech’s Antitrust and Monopoly Concerns

[11:38–12:50]

• With Nvidia (dominant in AI chips) and OpenAI (dominant in LLMs) coordinating so closely, Scott flags antitrust red flags:
  • Custom chip design and close supplier ties could lock competitors out.
  • "If the premier chip maker is optimizing and giving better chips and... for OpenAI's business... They create a moat that no one else can compete with... very obvious antitrust violations." — Scott [11:50]

5. The Recurring Pattern of Overinvestment in Tech Revolutions

[25:46–29:28]

• Ed expands the analogy, citing over-investment cycles in:

  • Railroads (1800s)
  • Electrical grid (1920s)
  • Telco fiber in the dot-com era

• In each case, infrastructure investment overshot real, present demand, leading to bankruptcies and crashes, even as the long-term infrastructure became valuable.

• Financial mismanagement, not the fundamental technology, was the culprit. Ed singles out OpenAI as most at risk of being overextended.

  “If we are to see a crash in some form... what is the company that I think is going to get caught with their pants down? ... I think you gotta go with OpenAI.” — Ed [28:43]

6. Who Gets Burned, and What Happens Next?

[29:28–36:30]

• Winners and Losers:

  • It's late-coming shareholders who get burned if (when) the music stops.
  • Scott references the "Wintel" (Windows-Intel) duopoly as a historical template; predicts a similar centralization in AI.
  • OpenAI and similar companies may rely on building insurmountable infrastructure, locking up compute and power.
  • IPOs and debt raises are likely next, repeating Amazon's playbook—some giants will survive, many will not.

• Prediction from Scott:

  • The next phase is big tech using inflated stock as acquisition currency, buying up older, cash-flow-stable firms as insurance.

  “You’re going to just see extraordinary transactions here, not only of the new stuff, but also the next stage... is when these companies use their cheap capital to go buy more traditional technology and data firms that have cash flow and are more enduring.” — Scott [33:08]

7. Broader Economic Picture: Illusion of Prosperity, Reality of Inequality

[39:23–46:21]

• Headline data (record indices, GDP, earnings growth) is being driven by the top 10%, masking how the bulk of Americans are struggling.

  • Sales of cheap food are rising; searches for cheap eats are up.
  • Housing and medical insecurity are acute for low- and middle-income households.
  • Animal shelters are full as more pets are surrendered—a subtle recession indicator.
  • The "S&P 10" phenomenon: just 10 mega-stocks are responsible for more than half of recent index gains.

  “We’re becoming like a Venezuela or a third-world country where basically the top 10% have all the money... It gets really ugly down the food chain.” — Scott [41:51]

8. Where is This Headed? Scenarios for the Future

[46:21–54:37]

• Potential Outcomes:

  • The obvious scenario: current tech giants can’t sustain valuations, leading to a massive correction—either through genuine productivity gains or, more likely, an abrupt crash.
  • Darker scenarios: financial collapse and constitutional crisis fueled by inequality, polarization, and civil unrest.
  • Regional breakup of the U.S. into economic/cultural blocs, à la EU.
  • More likely: a populist political backlash leading to an AOC/Bernie Sanders-type figure dominating, not a centrist.
  • Both tech bubble and consumer data “smell of recession.”

  “America’s kind of down with [inequality] because our superpower is that we believe we’re going to be someday in that top 10, maybe that top 1%... Everybody in America knows the lottery is stupid, but, baby, my ticket’s a winner.” — Scott [43:38] “Both of these stories... they smell of recession.” — Ed [50:10]

9. Recession: Inevitable or Imminent?

[51:52–54:37]

• Scott cites Jamie Dimon: “A recession is something that happens every seven years.”

• The U.S. has gone 17 years without a significant recession—due for a contraction.

• Major concern is not just recession, but stagflation, disaffected youth, and possible unrest, given staggering generational inequality.

• Real systemic change could depend on whether dissatisfaction translates to votes or something more disruptive.

  “The bottom line is young men are more prone to scale the walls of the capitol... There’s no indices for self-harm, deaths of despair... One economic shock and... nine in ten young people don’t think the system is working.” — Scott [53:19]

10. Week Ahead & Big Predictions

[54:37–56:46]

• Economic data on the horizon: consumer confidence, unemployment, Nike earnings, expiring EV tax credit.

• Scott’s Prediction:

  • Expect some of the biggest M&A and investment deals in history within the next six months, as major tech companies spend their currently cheap/inflated currency.
  • Disney could be a takeover target, given its vulnerability and the "Megalodon" scale of potential tech acquirers.

  “You’re going to see some of the biggest, strangest transactions—some of the biggest M&A deals in history...” — Scott [55:10]

Notable Quotes & Memorable Moments

• On the Circular Logic of AI Valuations:

  “It feels eerily reminiscent of '99... It’s a head fake. It’s not really getting the $100 billion in the full-body contact violence.” — Scott [08:02]

• On Tech Bubble Parallels:

  “The more and more the AI story unfolds... this really is looking almost identical to what happened with the dot-com implosion.” — Ed [12:51]

• On Underlying Demand:

  “You don’t need to worry that AI isn’t going to have a big impact. It is... The question is, do the valuations reflect the actual dollar demand?” — Scott [19:36]

• On Economic Reality vs. Market Data:

  “One of the most damaging metrics... is the NASDAQ and the Dow Jones because it creates this illusion of prosperity.” — Scott [41:26]

• On the Cycle of Recession:

  “Jamie Dimon... said something that happens every seven years... We have been partying all night... the mother of all hangovers is coming.” — Scott [51:52]

Timestamps for Key Segments

• 05:14 – 07:36: Introduction of AI investment “circular deal theory”
• 07:36 – 12:50: Dot-com bubble parallels, financial engineering, and bubble dynamics
• 18:32 – 22:20: AI realities vs. hype, anticipated IPOs/public markets
• 25:46 – 29:28: Historical tech bubbles: railroads, electric grid, telco fiber
• 29:28 – 36:30: Winners/losers, debt risk, “winner takes all” AI predictions
• 39:23 – 46:21: Economic inequality, illusions in official economic data
• 46:21 – 54:37: Potential future scenarios—corrections, political fragmentation, and backlash
• 54:37 – 56:46: The week ahead; predictions on M&A and shifts in corporate strategy

Episode Tone and Style

• The hosts combine sharp financial analysis with irreverent, sometimes darkly comedic banter.
• Scott’s anecdotes (dot-com era, dinners with CEOs) anchor historical analogies.
• Ed brings data and historical context, echoing Scott’s skepticism and concern.
• Both are candid in their warnings and critical of financial media/market hype, maintaining their trademark “no mercy, no malice” style.

This episode is essential listening for anyone looking to understand the risks lurking beneath the AI gold rush—and for those seeking a sober (and at times entertaining) assessment of how market psychology, financial engineering, and inequality could shape the economy’s path in the months ahead.