Oracle's share price has been on a dramatic rollercoaster, driven by a single, once-obscure accounting metric now at the heart of the artificial intelligence investment frenzy. The metric, known as Remaining Performance Obligations (RPOs), represents future sales under contract but not yet recognized as revenue. For Oracle, this number has exploded to a staggering $523 billion as of November 30, creating both euphoria and deep skepticism on Wall Street.
The Meteoric Rise and Sharp Fall of Oracle's Stock
In early September, Oracle's stock experienced a seismic surge, skyrocketing 36% in a single day following an earnings report that revealed its RPOs had more than tripled to $455 billion. This peak, reached on September 10, marked an all-time high for the company. However, the optimism was short-lived. Since that pinnacle, the stock has plunged dramatically, losing 43% of its value. The catalyst for this reversal is growing investor unease over the quality and collectibility of those massive future sales promises, particularly a single, colossal commitment.
The $300 Billion OpenAI Question Mark
Central to the controversy is a mammoth, five-year contract with AI research leader OpenAI. While Oracle has not officially named the client, it is widely understood that approximately $300 billion of its RPO growth is tied to a deal to supply computing capacity to OpenAI. This commitment is now under intense scrutiny.
Analysts point to a critical mismatch: OpenAI's ambitious future commitments versus its current financial reality. A December 5 report by Moody's indicated OpenAI has about $1.4 trillion in commitments over eight years, yet it remains in a fundraising stage with comparatively modest revenue. The company has stated it is on pace for over $20 billion in annualized revenue by end-2025, a figure dwarfed by its obligations.
This has led to open skepticism from Wall Street. J.P. Morgan analyst Erica Spear titled a recent note "If You Build It, Will They Pay?" Similarly, D.A. Davidson's Gil Luria suggested Oracle should proactively restructure the OpenAI contract, arguing it is unlikely to be fulfilled and capital could be deployed more responsibly.
The "Squishy" Accounting Behind "Probable" Sales
The core of the debate hinges on an accounting judgment call: Is the collection of these contracted amounts "probable"? RPOs are not recorded on the main financial statements but disclosed in footnotes, and their recognition depends on management's assessment of this probability.
This is a subjective standard, higher than "more likely than not" and often interpreted by accountants as a 70% or greater likelihood. The rules do not provide a precise numerical threshold, leaving room for interpretation. As facts change—such as the status of a key investor's commitment—management's view of what is "probable" can also shift, making RPOs a dynamic and potentially volatile metric.
The situation is further complicated by the interconnected, or "circular," nature of deals in the AI sector. Oracle is a major customer of Nvidia for its chips. Nvidia, in turn, had announced a non-binding letter of intent on September 22 to invest up to $100 billion in OpenAI over many years, with OpenAI buying millions of Nvidia chips. However, this deal has not been finalized nearly three months later, with Nvidia cautioning in its latest report that "there is no assurance" the investment will be completed.
If Nvidia's investment falters, it raises serious questions about OpenAI's ability to fund its $300 billion commitment to Oracle, potentially triggering a reassessment of what future sales are truly "probable." The artificial intelligence boom is built on mind-boggling numbers, but their solidity ultimately rests on a series of complex, interdependent judgments.
