Revenue in the August 2025 quarter was $14.9B, less than 1% below consensus. Earnings per share were $1.47, missing by a penny, and operating margin was 41.8%, modestly ahead of forecasts. Guidance for November 2025 called for 12 to 14% revenue growth in constant currency and cloud revenue growth above 30%. On their own, these results would not justify a 5% move in shares, not 36%.

The true catalyst was the unprecedented disclosure of $455B in remaining performance obligations for AI workloads (cloud), up 359% from the prior year and more than double analyst expectations. Notably, according to reporting from The Wall Street Journal the day after the earnings release, the majority of these new bookings stem from a single, transformational contract: OpenAI has agreed to purchase $300B in cloud computing power from Oracle over five years, beginning in 2027.

The addition of this one deal helped drive over $300B in new bookings in a single quarter and has fundamentally reframed the Street’s Oracle investment thesis. Investors now see a multiyear growth profile tied to AI workloads anchored by OpenAI’s massive, long-term commitment, rather than incremental cloud gains.

While this backlog highlights Oracle’s increasing relevance in AI infrastructure, it also introduces new risks, given the significant concentration in one customer and the back-ended timing of revenue recognition. Nonetheless, this watershed agreement represents the largest cloud contract in Oracle’s history and shifts the narrative toward sustained, AI-driven growth.

Overflow no longer tells the full story. Breaking from its previous role as a supplementary cloud provider, Oracle has secured a landmark $300B cloud computing deal with OpenAI, propelling the company to center stage in the race for AI workloads. This is not about picking up slack from hyperscalers. It’s a bold, high-stakes play to become core infrastructure for the most prominent generative AI company.

The Wall Street Journal reports that the majority of Oracle’s new $317B in contract bookings this quarter stem from the OpenAI deal, which runs for five years beginning in 2027. For context, Oracle’s current total cloud run rate is just ~$25B versus AWS’s $120B+. The OpenAI contract compels Oracle to rapidly scale capacity, finance unprecedented infrastructure buildout (requiring substantial debt), and depend heavily on a single customer’s success.

Gone is the idea that Oracle’s growth will slow as “overflow” normalizes. Instead, the company’s multiyear growth now hinges on OpenAI’s adoption curve, competitive advantage, and financial health. While this could set Oracle on a path of record revenue and relevance if OpenAI delivers, it’s an all-in move and one that exposes Oracle to concentration risk, execution risk, and delayed revenue realization (since the majority of new revenue isn’t recognized until 2027 and beyond).

Oracle has shifted the narrative from “scraps from Big Tech’s table” to major AI infrastructure partner. Investors are betting that this will be durable and transformative, but the sustainability of this model over the long run is far from certain.

The forward guidance is ambitious: $18B in Oracle Cloud Infrastructure revenue for fiscal 2026 (77% growth), then surging to $32B in 2027, $73B in 2028, $114B in 2029, and $144B in 2030. This is a long-term bet on OpenAI, and by proxy, on the continued explosive expansion of AI infrastructure demand at a scale never seen before.

Three months ago I suggested we were in the third inning of the AI buildout. After June earnings I revised my view that we were in the second inning. I’m now debating if we’re in the first inning.

The size of Oracle’s backlog highlights that AI demand is stretching beyond available supply and mind blowing levels. This supports the view that the AI infrastructure cycle remains in an early phase.

And the view that we’re still early goes beyond the AI infrastructure buildout that should have elevated growth rates for the next few years.

It speaks to how disruptive AI will be. The level of infrastructure investment is rocketing past high expectations. That means the hardware brain of AI is going to be bigger than we thought six months ago. A bigger AI brain gives AI native companies the horsepower to build experiences that will dramatically change our lives. How much change? The just like Oracle’s guidance, the sky is the limit.

The overflow dynamic also validates the hyperscalers’ case. If their customers are resorting to Oracle, it implies that their own pipelines are even larger. In time, they will expand to accommodate more of this demand. Until then, the industry will experience a period of sustained investment and constrained supply. That environment supports pricing power and visibility across the ecosystem.

Disclaimer