Oracle Corporation ORCL

Because its cloud backlog exploded. In September 2025 Oracle disclosed that remaining performance obligations had jumped to $455 billion and projected $144 billion of fiscal-2030 cloud revenue; the stock rose 36% in one session, its best day since 1992. By the end of fiscal 2026 that backlog reached $638 billion, up 363% year over year, and OCI revenue grew 93% to $5.8 billion in the May quarter. The market stopped pricing Oracle as a slow database vendor and started pricing it as the fourth hyperscaler — even after the shares gave back nearly half their peak to about $184.

RPO — remaining performance obligations — is contracted cloud work Oracle has booked but not yet delivered, and at $638 billion it is close to nine times the company's $67.4 billion of annual revenue. It matters because the stock now trades on it: management expects to recognize roughly 12% of the backlog as revenue over the next twelve months, so it is a visible base for the fiscal-2027 guide of about $90 billion. The caution is that a backlog is a forecast with a contract attached, not cash — and Bank of America estimates more than half of it traces to a single customer, OpenAI.

Yes — free cash flow was about negative $23.7 billion in fiscal 2026. Oracle generated a record $32 billion of operating cash flow but spent $55.7 billion on capital expenditures, up 162%, to build the AI cloud. Management guides to roughly $70 billion of net capex in fiscal 2027 and says it expects to raise about $40 billion more in debt and equity, including a $20 billion share program, on top of $43 billion of debt raised last year. Unlike the established hyperscalers, Oracle funds the build-out largely with external capital rather than its own cash flow.

By Bank of America's estimate, more than half of Oracle's $638 billion backlog traces to a single customer, OpenAI. That concentration runs largely through the roughly $300 billion Stargate capacity commitment, and it cuts both ways: it is the engine behind the RPO surge but also leaves Oracle unusually exposed to one counterparty. A report that OpenAI had missed internal targets sent the shares lower, and DA Davidson cut its price target to $200 in late 2025 citing the single-customer driver of the backlog — though it raised the target to $225 after the June 2026 results.

They span an unusually wide range around a price near $184. The consensus twelve-month target sits around $253 across roughly 43 analysts, but the panel runs from a low near $155 to a high of $400 at Guggenheim. After the June 2026 results, targets moved both ways — Bernstein raised to $325 and Barclays to $250, while Wedbush cut to $240 and Scotiabank to $241 on the capital-raise and cash-flow concerns. That roughly two-and-a-half-fold spread captures the core disagreement about how to value a record backlog.

Because a one-time gain flattered the earnings underneath it. The trailing GAAP multiple near 32 times rests on per-share earnings that include a $2.7 billion pre-tax gain from selling Oracle's Ampere stake, booked in the quarter ended November 2025, which lifted that period's earnings to $2.10 a share. Strip out the gain and trailing profit — and the multiple — look richer, not cheaper. The forward multiple, around 23 times the fiscal-2027 non-GAAP guide of about $8.05, is the cleaner read, but it still capitalizes a backlog that has not yet reached the income statement.