Better Offline – "The Enshittifinancial Crisis: Part Three"

Host: Ed Zitron (Cool Zone Media and iHeartPodcasts)
Date: January 22, 2026
Length: ~40 minutes

Episode Overview

This third installment of the "Enshittifinancial Crisis" series dives deep into the financialization and looming instability in the current tech economy—especially focusing on the AI startup bubble, the unsustainable data center boom, and the dangerously centralized and debt-fueled nature of tech infrastructure. Host Ed Zitron weaves an engaging, heavily opinionated analysis, critically examining how venture capitalists, institutional investors, and major banks have become complicit in fueling a potentially catastrophic financial setup centered around artificial intelligence startups and the massive, debt-driven expansion of data center infrastructure.

Key Discussion Points & Insights

1. The AI Startup Valuation Bubble

[02:19 – 08:20]

• AI company valuations are at historic highs, but these numbers are often "on paper" with little substance behind them.
• Venture capitalists (VCs) are investing in AI startups—like OpenAI, Anthropic, Perplexity, Cognition, Cursor—at valuations ranging from $3B up to $29B, even as these companies hemorrhage cash and lack a clear path to profitability.
• Startups’ value is driven by VCs' ability to mark up their assets under management (AUM), which lets them charge higher fees to their limited partners.
• Zitron highlights how some VCs are investing with "little to no information" about companies' finances, referencing sources who claim investors past OpenAI's $100B valuation round have no "information rights" and thus receive only the numbers companies want to publicize.
  • Quote [07:30]: "They only know what new ridiculous stat Sarah Friar, CFO of OpenAI, has farted out..." – Ed Zitron.

2. The Delusion of Perpetual Growth

[08:20 – 10:30]

• VCs cling to the narrative that pouring money into growth, regardless of profitability, will yield returns—a belief increasingly detached from business fundamentals.
• Ed doubts OpenAI's revenue claims ("$20 billion annualized"), suspecting major exaggeration to prop up the myth of explosive AI growth.
  • Quote [07:58]: "They want you to think OpenAI made $20 billion in 2025—fuck that, I don't even think they made 10." – Ed Zitron.
• Predicts a wave of revaluations for AI startups: their next funding rounds (or exits) will slash valuations by at least half, possibly collapsing confidence in VC as an asset class.

3. Data Center Mania and the CoreWeave Case Study

[10:30 – 19:00]

• Focus shifts to the parallel, debt-fueled explosion in data center construction, primarily to support AI compute.
• Cites Bloomberg stats: 2025 saw $178.5B in US data center credit deals, more than double 2024.
• CoreWeave—an "AI Neo-Cloud" company—is presented as the poster child for risky, leveraged infrastructure:

  • Its business model: pre-sign massive contracts (sometimes before the infrastructure exists), use these as collateral to raise billions in debt, install Nvidia GPUs, then (eventually) start fulfilling contracts and getting paid.

  • CoreWeave's Debt Situation:

    • As of Q3 2025, $25B in debt on $5.35B revenue; deeply unprofitable.
    • Sits on $1.1B in deferred revenue (deposits for services not yet rendered).
    • Major investors/partners include Nvidia, which also serves as a backstop for certain debts.

  • Construction of these data centers routinely faces major delays. Zitron questions the official excuses (blaming weather, etc.), digging up evidence suggesting underlying financial issues.

  • Quote [14:09]: "This is a very important deal to know for literally any data center development you've ever heard of or seen." – Ed Zitron (on CoreWeave's unusual contract structures).

4. Debt Structures and Financial Meltdown Risk

[19:00 – 24:40]

• Outlines the web of financial institutions underwriting these data center loans (Blue Owl, JP Morgan, Goldman Sachs, MUFG, BNP Paribas, Morgan Stanley, Deutche Bank, etc.).
• This centralization is a major risk: if one institution falters or pulls out, the entire stack of deals could be destabilized.
  • Quote [22:27]: "It’s worrying how centralized this is. And the largest deals... All four of those companies have at some point funded CoreWeave. In fact, everybody appears to have funded CoreWeave at some point." – Ed Zitron

5. The Absurd Economics of Modern Data Centers

[28:59 – 34:00]

• Reveals the core flaws in the data center "gold rush":
  • Construction is slow, expensive, and prone to delay.
  • All the new AI-specific facilities will fill up with GPUs that become obsolete before they're even operational, thanks to Nvidia’s rapid upgrade cycles.
    • Example: as new facilities finish installing Blackwell chips, Nvidia will already be shipping the Vera Rubin generation, with incompatible rack formats once Kyber launches in 2027—instantly devaluing "new" assets.
  • Most demand is artificially concentrated in a handful of hyperscale deals with OpenAI and similar giants, who themselves are on shaky financial footing, often negotiating year-long or longer payment terms.
    • Quote [29:30]: "Every single major bank and financial institution has piled hundreds of millions, if not billions of dollars into building data centers that… only seem to lose money. Didn't think—no one, no one fucking think to check that one." – Ed Zitron.

6. Cascading Risks, Shaky Tenants, and Pullouts

[34:00 – 39:00]

• Many of the largest data center deals hinge on tenants (like OpenAI, Microsoft, Oracle) that may struggle to pay—or may be incentivized to walk away from contracts if service is delayed.

• Highlights damning examples, such as:

  • Microsoft's deal with Nebius—which may collapse if Nebius can’t deliver (with suspiciously unrealistic construction timelines).

  • Blue Owl Capital, a key financier, abruptly pulling out of a $10B Oracle data center project in Michigan, citing mounting debt risk and AI spending. Despite industry rumors, Zitron is skeptical that a new institutional player will simply "fill the gap" this time.

  • Quote [37:42]: "This is not a consumer mortgage... Stepping in would require billions of dollars in legal logistics and likely Blackstone talking to the very same banks who already said no. Why are they not doing this? And indeed, why are things looking shaky?" – Ed Zitron

• Warnings about the fragile financial state of supposedly bulletproof firms, e.g., Oracle has negative cash flow, low margins on GPUs, and $248B in data center leases pending—almost all to serve one major (questionable) customer: OpenAI.

Notable Quotes & Memorable Moments

• On AI Startup Bubble:
  • “A big beautiful number, the biggest you’ve ever seen, and your limited partners need to pay you a fee just to manage it.” [05:11]
• On Information Rights:
  • “Nobody who invested past OpenAI’s $100 billion valuation have any information rights…they literally do not know anything about the company.” [06:49]
• On Deferred Revenue:
  • “Coreweave’s sitting on $1.1B in deferred revenue. That’s income for services not yet rendered.” [18:26]
• On Data Center Construction:
  • “Data center developers are raising money months up to a year before they ever expect to make a penny of revenue, not profit.” [13:16]
• On Obsolescence:
  • “All of those data centers being built right now are being filled with Blackwell chips and by the time they turn on Nvidia will be selling its next generation Vera Rubin chips, making them obsolete.” [30:45]
• On Centralized Financial Risk:
  • “It’s worrying how centralized this is…literally every part of this piece is being funded by like 8 to 9 banks.” [22:15]
• On Blue Owl Pullout:
  • “Blue Owl pulling out is not the same as a regular deal…If it’s $10 billion, Blue Owl probably would put in 2 billion themselves. There’d be, like, maybe $8 billion of credit…” [37:58]

Key Timestamps

• 02:19 – Ed Zitron opens main episode content; lays out the AI/VC valuation bubble.
• 06:49 – Information rights (or lack thereof) for AI company investors.
• 07:30 – Critique of OpenAI’s public revenue numbers.
• 10:30 – Shift to data center bubble, introduction to CoreWeave.
• 13:16 – Explanation of how debt is used to finance data centers.
• 18:26 – Ed dives into CoreWeave’s deferred revenue and loan documents.
• 22:00 – 24:00 – Rundown of major banks funding these deals and the associated risks of concentration.
• 28:59 – Economic flaws in data center “gold rush”; hardware obsolescence.
• 32:40 – Questioning future rental demand and payment flows.
• 34:00 – Real-world hazards: Microsoft, Nebius, contractual collapse risk.
• 37:42 – Blue Owl pulls out; implications for financial confidence.
• 40:00 – Episode closes, tee-up for next week: the “data center apocalypse.”

Tone and Style

Ed Zitron infuses the episode with energetic skepticism, biting sarcasm, and frequent profanity—critically dismantling industry hype and calling out systemic financial recklessness. His commentary is direct, conversational, richly informed, and at times bleakly humorous.

Conclusion

Part Three of "The Enshittifinancial Crisis" lays bare how the modern tech and AI economy is underpinned by risky debt, opaque valuations, and speculative infrastructure investment. With major institutions shoulder to shoulder in a house of cards and core tenants (like OpenAI, Microsoft, Oracle) as much at risk as the companies serving them, Zitron warns of a looming collapse when reality inevitably catches up. Part Four, previewed at the end of the episode, promises to explore how this domino effect might actually unfold.