A new OpenAI Microsoft payment leak is shedding rare light on one of the most important — and most mysterious — financial relationships in the AI industry. After months of speculation about OpenAI’s runaway revenue and compute spending, leaked internal documents obtained by tech blogger Ed Zitron now offer the clearest picture yet of how the money actually flows between OpenAI and Microsoft.

According to the leaked figures, Microsoft received $493.8 million from OpenAI in revenue-share payments in 2024. Fast-forward to the first three quarters of 2025, and that number jumped to a staggering $865.8 million. The documents indicate that OpenAI is sending roughly 20% of its revenue to Microsoft — a share tied to Microsoft’s more than $13 billion investment in the AI company.

But here’s where the OpenAI Microsoft payment leak gets even more complicated.

A source familiar with the agreement told TechCrunch that the leaked numbers represent net revenue share, not the gross figure. That’s because Microsoft also shares revenue with OpenAI. Microsoft pays OpenAI for powering Bing with OpenAI models and for the Azure OpenAI Service, which thousands of developers rely on. Those payments reduce the total Microsoft reports internally.

And since Microsoft doesn’t break out this exact revenue in its public filings, the true scale of the financial exchange isn’t fully visible — even with the leak.

What the OpenAI Microsoft Payment Leak Tells Us About Revenue

Using the widely cited 20% split, analysts estimate that OpenAI generated at least:

• $2.5 billion+ in revenue in 2024
• $4.3 billion+ in the first nine months of 2025

But the real numbers are almost certainly higher. The Information previously reported that OpenAI made around $4B in 2024, while Sam Altman most recently said the company is generating “well more” than $13 billion a year and will end 2025 with a $20B+ run rate — with ambitions to hit $100B by 2027.

Where Is All the Revenue Going? Compute Costs.

The OpenAI Microsoft payment leak also highlights a concerning trend: OpenAI may be spending even more on inference than it earns.

Inference — the compute needed for ChatGPT to answer your prompts — is extremely expensive, and most of that spend is cash, not credit. According to Zitron’s analysis:

• $3.8B spent on inference in 2024
• $8.65B spent on inference in the first 9 months of 2025

Historically, Microsoft covered most of OpenAI’s training compute through credits, but inference is a cash burn, and it’s accelerating.

OpenAI has expanded beyond Azure and now also uses CoreWeave, Oracle, AWS, and Google Cloud. But Azure remains the core compute backbone — and the biggest cost center.

If the leak is accurate, OpenAI is potentially operating at a loss despite explosive revenue growth.

The Bigger Question: What Does This Mean for the AI Boom?

The OpenAI Microsoft payment leak fuels growing concerns about whether today’s AI economics are sustainable.

If the most valuable private company in AI — backed by billions from Microsoft — is still deeply in the red due to compute costs, what does that mean for every smaller AI startup? For investors betting on trillion-dollar AI markets? For the so-called “AI bubble” chatter echoing from Silicon Valley to Wall Street?

OpenAI insists it has a path to profitability. But the numbers in this leak suggest a much more delicate financial balancing act behind the scenes.

One thing is clear: the OpenAI–Microsoft relationship is not just strategic — it’s existential.