Podcast Summary: The Joe Rogan Experience Fan – OpenAI Running Losses Due to Microsoft Cloud Overdependence

Date: November 18, 2025
Podcast: The Joe Rogan Experience Fan
Host: The Joe Rogan Experience of AI

Episode Overview

This episode takes a critical look at newly leaked financial documents regarding the relationship between OpenAI and Microsoft. The host breaks down OpenAI’s skyrocketing compute costs, the complex revenue-sharing deals with Microsoft, and the broader implications for the rapidly expanding AI sector. Using insights from journalists like Ed Zitron and sourced leaks, the episode dissects OpenAI’s spending patterns, revenue growth, and what these numbers mean for the future of artificial intelligence investment.

Key Discussion Points & Insights

1. Leaked Documents Reveal OpenAI’s Payments to Microsoft

• [00:00–02:20]
  • Host discusses leaked internal documents (via Ed Zitron) exposing just how much OpenAI pays Microsoft for compute.
  • In 2024, OpenAI paid Microsoft $493.8 million as part of their revenue share.
  • By the first three quarters of 2025, this payment climbed to $865.8 million, with expectations to surpass a billion dollars by year’s end.
  • Memorable Quote:

    “That’s just in the first three quarters of this year. So we have a whole, you know, another third quarter or fourth quarter that hasn't been reported in those numbers yet.” (A, 01:24)

2. Understanding the 20% Revenue Share Deal

• [02:20–04:30]
  • OpenAI agreed to pay Microsoft 20% of its revenue, rooted in Microsoft’s initial $10B investment (totaling $13B over multiple injections).
  • Neither company formally acknowledged the percentage publicly, although it’s widely reported.
  • Microsoft, in turn, shares approximately 20% of Bing and Azure OpenAI-related revenue with OpenAI.
  • The complex partnership is further complicated by product integration: Bing is powered by ChatGPT, and Bing is embedded inside ChatGPT.
  • Memorable Quote:

    “There's kind of like a lot of... collaboration between the two companies.” (A, 03:25)

3. Microsoft’s “Net” vs. “Gross” Revenue Share

• [04:30–05:20]
  • Leaked payments refer to Microsoft’s net revenue share, not gross, meaning profits are calculated after deductions for payments already made to OpenAI. This complicates how much each side is truly earning.

4. Estimating OpenAI’s Revenue via Leaked Data

• [05:20–07:10]
  • By back-calculating based on the 20% split, last year’s OpenAI revenue was at least $2.5B; first three quarters of 2025 brought in about $4.33B.
  • Multiple reporting sources (e.g., The Information) offer similar but not identical numbers.
  • Sam Altman’s public statements suggest even higher future earning potentials (claims of “well more” than $13B projected for the annualized run rate).
  • Memorable Quote:

    “When Sam Altman was asked about their revenue, he said that is quote, well, more when he was referring to reports of $13 billion a year.” (A, 06:45)

  • Altman’s ambitious prediction: hitting a $20B annualized run rate by end of 2025 and possibly $100B by 2026.

5. Skyrocketing Compute (Inference) Costs

• [07:10–09:10]
  • Zitron’s sources estimate OpenAI spent $3.8B on inference in 2024, ballooning to $8.65B in the first three quarters of 2025.
  • Inference costs refer to the live compute needed to keep AI models responsive for users, and most of this was paid to Microsoft Azure.
  • Recent deals with other providers (AWS, Google Cloud, Oracle, etc.) hint at OpenAI’s attempts to diversify beyond Microsoft due to scale and costs.

6. Cash vs. Credit: The Real Cost Structure

• [09:10–10:30]
  • Training spends are largely made using cloud credits provided by Microsoft as part of investment deals (“tokens or credits”), reducing immediate cash drain during training.
  • However, inference costs (serving user queries, daily operations) demand large cash payments, driving losses.

7. Implications for the AI Sector: The “AI Bubble”?

• [10:30–11:30]
  • Despite massive revenue, OpenAI operates at a loss because of overwhelming inference costs.
  • If the market leader isn’t profitable, it raises questions about sustainability across the industry and fuels talk of a potential “AI bubble.”
  • Memorable Quote:

    “If OpenAI is really still running their models at a loss… that's like a massive deal for the investment world who is spending tons of money on these AI models. The valuations are absolutely insane.” (A, 10:50)

Notable Quotes (with Timestamps & Attribution)

• On cost escalation:

  “You can assume it’s going to be over a billion dollars that they're going to be paying Microsoft for this year alone.” – A (01:28)

• On Microsoft/OpenAI collaboration:

  “There's kind of like a lot of… collaboration between the two companies.” – A (03:25)

• On Sam Altman’s projections:

  “He said that by the end of this year, they’re going to end it above $20 billion in annualized revenue run rate.” – A (06:55)

• On the AI sector’s profitability:

  “If OpenAI is not running at a profit, that means a lot of other people are not as well.” – A (11:03)

Important Segment Timestamps

• OpenAI’s Payments to Microsoft – 00:00–02:20
• 20% Revenue Share Deal Structure – 02:20–04:30
• Net vs. Gross Revenue Share – 04:30–05:20
• Calculating OpenAI’s Revenue – 05:20–07:10
• Compute Cost Explosion – 07:10–09:10
• Cash vs. Cloud Credits – 09:10–10:30
• AI Bubble Concerns – 10:30–11:30

Tone & Style Notes

The host delivers facts with a colloquial, enthusiastic tone reminiscent of classic Joe Rogan discourse: a mix of technical specificity, skepticism about hype, and open-ended wonder about future implications. The episode is research-driven but accessible, blending leaked stats with plain-language explanations.

Conclusion

This episode delivers a critical, detailed look at the economics of generative AI. The mounting compute costs, aggressive revenue projections, symbiotic yet complicated relations with Microsoft, and the persistent gap between revenue and profit paint a picture of an industry on an edge—poised for either massive growth or a dramatic correction. The host urges listeners to watch this space, as these developments may determine not just the future of OpenAI, but the entire AI investment world.