by Mighva Verma - 3 days ago - 3 min read
Big investments, high-stakes alliances, and a relentless race to build the world’s most powerful AI models at the core of this drama is the blockbuster partnership between OpenAI and Microsoft. Freshly leaked financial documents now reveal the eye-popping scale of the dollars moving between these two giants, and the findings highlight both the rewards and the wild risks in the red-hot artificial intelligence sector.
According to documents recently leaked to tech blogger Ed Zitron, Microsoft received $493.8 million in revenue-share payments from OpenAI in 2024. If that number makes you pause, get this: in just the first nine months of 2025, those payments exploded to a staggering $865.8 million. The reason? OpenAI is contractually obligated to share about 20% of its revenue with Microsoft, a direct result of Microsoft's eye-watering $13 billion investment in the startup, cementing their powerful alliance. While neither side has officially confirmed these percentages, the numbers provide the clearest glimpse yet into their financial relationship.
What makes this partnership even more intriguing is the mutual exchange. Microsoft, which uses OpenAI’s models to power Bing and its Azure OpenAI Service, actually pays OpenAI a similar revenue share, believed to be around 20%, from those business lines. However, the figures cited in the leaked documents are Microsoft’s net receipts, meaning deductions are already factored in for whatever Microsoft pays OpenAI in return. The tech giant keeps the specifics about Bing and Azure OpenAI Service revenue under wraps, making the total kickback figures even trickier to pin down.
Back-of-the-envelope calculations, working off that 20% figure, suggest OpenAI brought in at least $2.5 billion in revenue in 2024 and $4.33 billion in the first three quarters of 2025, though prior estimates put actual numbers even higher. OpenAI CEO Sam Altman has claimed the company’s revenue is well beyond $13 billion annually, with the potential to exceed a $20 billion run rate by year’s end and possibly a jaw-dropping $100 billion by 2027.
The story isn’t just about money earned, it’s also about money burned. Inference the cloud compute needed to generate AI responses is an outrageously expensive undertaking. Documents point to OpenAI spending about $3.8 billion just on inference in 2024, with spending rocketing to $8.65 billion in the first nine months of 2025. Most of that cash flows to Microsoft Azure, the backbone of OpenAI’s cloud muscle, with only a handful of secondary partners like CoreWeave, Oracle, AWS, and Google Cloud in the mix.
Some costs, like training new AI models, are buffered by Microsoft’s cloud credits (essentially non-cash), but nearly all inference costs are paid in hard cash. Disturbingly, these numbers imply OpenAI may be spending more on running its AI than it actually brings in.
The implications are enormous. If OpenAI, one of the world’s most celebrated AI unicorns, is still deep in the red just to keep its models running, the doubt and debate return: Is the AI gold rush as lucrative as it seems, or just a very expensive bet on the future? As investment dollars chasing AI ventures soar, the industry waits to see if the financial math finally adds up or if another tech bubble is in the works.
Neither OpenAI nor Microsoft has commented on the leaks. But one thing is certain: the business of building the world’s smartest machines is as much about relentless ambition as it is about relentless spending.