Oracle spent most of the last decade as the database company that missed the cloud. Then, over the past two years, it turned into one of the most talked-about names in AI infrastructure, and the stock re-rated to match. The question is whether the move is justified, or whether a legacy software company got swept into a theme. The honest answer is that the backlog is real and enormous, and the stock has already priced in the assumption that Oracle converts it cleanly. Those are two different bets, and only one of them is close to a sure thing.
What Actually Changed
Oracle’s cloud infrastructure business, OCI, was a distant fourth behind Amazon, Microsoft, and Google. What it had going for it was a reputation for high performance networking and lower pricing, which turned out to matter a lot for training large AI models. Oracle leaned into that, signed capacity deals with AI labs and other customers, and began building data centers at a pace that would have been unthinkable for this company a few years ago.
The result shows up in one number: remaining performance obligations, which is the dollar value of contracted revenue that Oracle has signed but not yet delivered. That figure went from tens of billions to well over one hundred billion, and then, after a set of very large AI capacity agreements, the company disclosed a backlog several times larger still. Management has used that backlog to guide toward cloud revenue growth accelerating sharply over the next few fiscal years, with total revenue targets that would roughly double the company by the end of the decade. Whether that scale of spending is sound or a warning sign is the question I worked through on whether the AI build-out rhymes with 1999.
The Backlog Is the Bull Case, and Also the Risk
A signed contract is not delivered revenue. To turn the backlog into cash, Oracle has to build the data centers, secure the power, install the chips, and operate all of it at a margin that justifies the capital. That is a very large construction and operations program running on a compressed timeline, and Oracle is doing it while its free cash flow has turned negative because the spending comes before the revenue. The company is funding that gap with debt, and its debt load was already meaningful before this build-out started. The Oracle financials page is the one to watch here, because the story lives in the revenue ramp, the capex line, and the interest bill.
There is also customer concentration. A large share of the headline backlog is tied to a small number of AI customers, and at least one of them is itself not yet profitable and is dependent on continued fundraising. That dependence runs through the whole build-out, which I mapped in a piece on who actually profits from the AI infrastructure spending wave. If any of those customers renegotiates, delays, or cannot pay on the original schedule, the backlog number that the stock is built on becomes less solid than it looks. This is not a reason to assume the worst. It is a reason to treat the backlog as a range of outcomes rather than a fixed figure.
Two Ways the Next Three Years Go
If Oracle executes, the next three years look like the fastest growth this company has posted in its modern history, cloud revenue compounding at rates normally reserved for much smaller companies, and the stock grows into a multiple that currently looks stretched. If the build-out runs behind schedule, or a key customer stumbles, or the margins on all this rented compute come in thinner than hoped, then the market is holding a legacy software business plus a capital-heavy, lower-margin infrastructure arm, at a price that assumed the good version. The spread between those two outcomes is unusually wide for a company this size.
| Question | Bull answer | What to watch |
|---|---|---|
| Does the backlog convert on time? | Yes, cloud revenue accelerates through FY27 and FY28 | Quarterly OCI revenue vs the guided ramp |
| Can Oracle fund the build? | Debt is manageable against the contracted revenue | Free cash flow, interest coverage, debt issuance |
| Are the customers solid? | Diversifying beyond the first big AI deals | Customer concentration disclosure |
| What is the margin on rented compute? | Improves with scale and utilization | Cloud gross margin trend |
What Would Change My Mind
I do not own Oracle, and I am watching rather than buying. The transformation is real, and I respect a company this old rebuilding its growth profile the way it has. But the stock trades at a multiple that needs the backlog to convert close to plan. The Oracle quote page carries the live valuation, and the quant rating tool shows the Valuation factor flashing the same caution this article does. Two or three quarters of the guided cloud acceleration actually printing would move me from watching to buying. A soft quarter that knocks the stock down the way infrastructure names get knocked down would do it faster. For now the price asks me to trust the forecast, and I would rather see it first.
Gavin Thorne writes on equity strategy and company-level research. This article reflects his personal research process and is intended for informational purposes only. It does not constitute investment advice.