🔴 OpenAI: What If the Money Runs Out?

Is the AI bubble about to burst?

In a widely discussed analysis published on April 14, 2025, Ed Zitron, founder of Where’s Your Ed At and a sharp critic of generative AI, draws an explicit parallel between OpenAI and Lehman Brothers before the 2008 crisis.

"OpenAI is a load-bearing company for the tech industry. Its collapse would trigger a contagion, just like Lehman Brothers did in the financial world."

In his view, OpenAI's survival determines not only the financial health of its partners (Microsoft, Nvidia, SoftBank, Oracle) but also the very narrative of generative AI as the new engine of economic growth.

"OpenAI isn’t just the leader of the generative AI market: it is the generative AI market."

I checked and followed closely what he was saying (his podcast is remarkably well sourced) and called my friends in the United States and in France to find out what is really going on.

🔍 My analysis

Despite these warning signs, I can see that OpenAI's o3 model has never performed better. Its abilities in reasoning, code generation, natural interaction and consistency in everyday use are superior to anything the company has produced so far. The user experience is stable, direct and remarkably effective.

To break into a disruptive market, between one-eighth and one-tenth of Big Tech's market value has been poured into generative AI (in cash, in models, in startups and in data centers). It is a colossal bet.

But it remains, to some extent, a calculated bet, unless world geopolitics and the systemic stakes we are discussing here get the upper hand.

Now it remains to be seen how OpenAI will navigate this year's legal and financial complexities.

One thing is certain: everyone needs it alive.

But while the general public is discovering o3, the competition is getting ready: DeepSeek R2, the next model from the Chinese company DeepSeek, is expected in early May 2025. Trained on an optimized stack, the model promises significant gains in performance and inference cost. It is already seen as a serious threat in technical communities, especially for enterprise and open-source use. So much so that Trump is trying everything to ban it.

1. A structurally loss-making model

OpenAI projects revenue of $12.7 billion for 2025. But its spending would reach $28 billion, including:

• $13 billion for compute on Azure (Microsoft),
• $2.4 billion for CoreWeave,
• $1 billion or more for Stargate,
• $3 billion to train new models,
• $2.5 billion in overhead (salaries, data, marketing, and so on).

"OpenAI spends $2.25 for every dollar it earns."
Ed Zitron

The rest of this analysis is reserved for our subscribers.