Executive Summary
Oracle's cloud infrastructure missed growth targets by 300 basis points, but the real story lies buried in a $300 billion backlog that may be mathematically impossible to fulfill. Bloomberg Intelligence analyst Anna Raghurana revealed that OpenAI—responsible for over 60% of Oracle's AI backlog—generates just $20 billion in annual revenue, yet has committed to spending $300 billion exclusively with Oracle. This creates a structural mismatch that no amount of optimistic modeling can resolve. The market is pricing Oracle as if this backlog converts linearly, but the physics of data center construction, power constraints, and OpenAI's competitive position suggest otherwise. Oracle's response—raising CapEx from $35B to $50B while burning $13B in free cash flow—signals desperation rather than confidence. The company needs private equity partnerships to finance this buildout, effectively admitting their balance sheet cannot support their own growth story. Meanwhile, Google's Gemini has closed the performance gap with GPT, undermining OpenAI's pricing power and ability to honor these commitments. This sets up a classic infrastructure trap: Oracle is building capacity for demand that may evaporate before the first server comes online.
Key Insights
what Matthew Bloxam, Anna Raghurana, Sam Fizzelli said“open AI currently or the order book is from open AI. By the end of this year, open AI will have a revenue run rate of about 20 billion. So everybody is saying that, okay, well, tell me if you have revenue of 20 billion, how are you going to spend 300 billion just with Oracle?”
This is a preview. Log in to see the full analysis including investment opportunities, risks, catalysts, and detailed insights.