Oracle’s credit rating drops to one level above junk | Ed Zitron

The video (hosted by Isaac, featuring Ed Zitron) argues that the “AI bubble” is already showing financial and credit stress—and that the fallout will be harsher for companies and investors that bought deepest into the hype.

1) AI data-center monetization is starting to look like an exit strategy

Ed Zitron reacts to a Wall Street Journal report that data center owners are seeking to sell majority stakes in AI-related operating companies worth tens of billions. His view:

• Investors are trying to “cash out” while mania still exists.
• The timing conflicts with the claim that the AI buildout is early-stage and will only accelerate—so selling now suggests fear or uncertainty about future demand/capacity.
• He expects more attempts to sell or restructure projects before the bubble fully breaks, especially because building data centers takes years and is expensive to complete properly.
• He dismisses the optimistic narrative that these sales are merely “handing keys to bigger pockets” to finish infrastructure; he frames it as real financial de-risking.

2) Private equity is portrayed as a major risk multiplier

The discussion centers on who might buy these stakes—particularly private equity (PE) and related capital.

• Zitron claims PE is “in a losing streak” since 2021 due to overinvestment in software and a need to deploy capital.
• He argues PE tends to “rush” and uses debt, which is especially dangerous for data centers because they’re like building a small town—you can’t cut corners and still make them function profitably.
• If PE makes the bet and demand/profits don’t materialize, he expects losses—and possibly “continuation funds” or selling among PE firms to pass risk along.

3) Oracle’s credit downgrade is presented as a warning sign tied to OpenAI

The core news item: Oracle’s credit rating is downgraded to about one level above “junk.” Zitron interprets it as reflecting Oracle’s dependence on OpenAI.

Key points he makes:

• He says S&P Global cited OpenAI specifically as the driver: Oracle is building about 7.1 GW of AI data-center capacity for OpenAI under very large contract terms (he estimates hundreds of billions total).
• He argues the business logic is fragile because:
  • OpenAI would need to pay for the capacity, but OpenAI likely cannot afford it at the implied scale.
  • Oracle has warned in its own reporting that it may not be paid by OpenAI; if Oracle can’t get paid, it says repayment is unlikely from anyone else without discounting.
• He claims Oracle’s core non-AI business has plateaued/declined for a decade, while AI compute is compressing margins (he says gross margin dropped significantly recently).
• Therefore, the downgrade is framed less as “Oracle is temporarily troubled” and more as “Oracle cannot sustain the financial structure if OpenAI demand/payment falters.”

4) Oracle’s financial distress could spill over into corporate/founder personal finances (Larry Ellison)

Zitron extends the Oracle story to Larry Ellison’s exposure:

• He says Ellison uses margin loans backed by large Oracle stock holdings.
• If Oracle’s stock falls (e.g., markets fear OpenAI won’t pay), margin calls could force Ellison to pledge more shares or sell shares—creating downward pressure on Oracle stock.
• He links this to timing-sensitive corporate plans involving Ellison’s deal-related cash/financing needs (Paramount/Warner Bros. Discovery), suggesting the deal might be threatened if stock/cash positions worsen.
• He ultimately argues Oracle may be “bailed out” institutionally (not allowed to simply collapse due to systemic/public-sector importance), but that the AI data-center side could be restructured or sold off—while Oracle’s credit stress still worsens rapidly if downgraded further (“fallen angel” risk).

5) Broader theme: executives are making sweeping claims while the market ignores near-term reality

The video broadens to the behavior of major AI backers:

• SoftBank CEO Masayoshi Son dismisses bubble fears while claiming AI spending could reach $5T annually by 2040 and suggesting fusion is needed. Zitron’s reaction: such statements ignore present constraints and are treated as “mantra” by media instead of being challenged.
• Zitron also portrays Masayoshi Son as “gambling,” implying SoftBank’s leverage and holdings are exposed to the same AI-linked fortunes as Oracle/OpenAI.

6) Meta is expected to pressure prices—threatening already-loss-making rivals

Another highlighted risk: Meta may undercut OpenAI/Anthropic pricing substantially (reportedly via using ad revenue to subsidize AI offerings).

• Zitron argues this won’t immediately “kill” OpenAI/Anthropic overnight, but it will dilute demand and growth rates.
• Since Zitron claims both OpenAI and Anthropic are already running massive losses and have large compute commitments, he argues they can’t afford slower growth or margin compression.

7) Overall conclusion: the bubble ends with the question “was all this money worth it?”

Across Oracle, data-center exits, PE risk, Meta pricing pressure, and AI-spending claims, Zitron’s thesis is that:

• The AI economy is heavily dependent on a few counterparties (OpenAI/OpenAI-like demand).
• Capital intensity and commitment schedules make the downside hard to absorb.
• Media and markets may have been overly credulous about CEO-level optimism.
• When the correction comes, the central verdict will likely be whether massive spending produced sustainable profits—or just huge losses.

Presenters / contributors

• Ed Zitron (guest; writer; host of Better Offline podcast)
• Isaac (interviewer/host; “on the tech report with me is…”)