Oracle · one lens
Hamilton Helmer on Oracle
Oracle has signed large AI capacity deals, but its rising build costs and latest financing talks leave the barrier protecting those deals unproven.
Oracle’s latest financing talks sharpen the question in Helmer’s 7 Powers: does Oracle earn a significant cost or price benefit that a rival cannot copy? The October 8 report says Oracle is talking with Apollo and Goldman Sachs about financing a large chip purchase. The Wall Street Journal describes Broadcom’s financing effort for OpenAI separately and gives no borrowing amount for Oracle. Oracle has not closed a deal in that report.
Oracle’s fiscal 2026 figures in the company dossier put capital expenditure at $55.7 billion, versus $21.2 billion in fiscal 2025, and free cash flow at negative $23.7 billion, versus negative $394 million the year before. Oracle needs those purchases to yield cash flow, not just capacity. Helmer’s Scale Economies test asks whether Oracle’s volume lowers unit costs enough to let it price where a smaller rival cannot profit. The record provides neither Oracle’s cloud unit costs nor a rival’s comparable costs.
Oracle has evidence of demand beyond OpenAI. The Tencent lease report describes a roughly $7 billion agreement over five years for chip access through Southeast Asian data centers. The Wall Street Journal’s OpenAI contract report describes roughly $300 billion over about five years. Oracle cannot establish a Scale Economies barrier from either contract’s size. Oracle also cannot establish a Cornered Resource from Tencent’s access to chips unavailable in China: the record does not show that rivals cannot obtain comparable supply.
Oracle’s long database history makes Switching Costs worth testing, but OpenAI’s reported rental commitment does not show what a customer would spend to leave after its contract ends. Oracle faces a separate delivery test in New Mexico. Bloomberg’s Project Jupiter report says Oracle sent Blue Owl a notice seeking to delay payments if the project misses its planned 2028 launch. Oracle has not demonstrated hard-to-copy Process Power through that notice.
The TEXXR coverage record counts 27 Oracle articles in the completed third calendar quarter of 2026, versus an average of 20 articles per quarter across the preceding eight quarters. The record measures writing about Oracle, not construction progress. Oracle has sizable contracts; the record has yet to show Helmer’s Sustainable barrier behind them.
The seven Powers, one by one
| Power | Status | Why |
|---|---|---|
| Scale Economies | contested | large capacity commitments appear in the record, but comparable Oracle and rival unit costs do not |
| Network Economies | absent | not argued in the record |
| Counter Positioning | absent | not argued in the record |
| Switching Costs | contested | OpenAI’s reported multiyear commitment shows a contract, not a demonstrated cost of leaving Oracle |
| Branding | absent | not argued in the record |
| Cornered Resource | contested | Tencent’s reported chip access does not establish supply that rivals cannot obtain |
| Process Power | absent | not argued in the record |
An editorial application of Hamilton Helmer's published framework — not their actual view, affiliation, or advice.
Where this lens comes from
Hamilton Helmer's framework is set out at the lens page, drawn from 7 Powers. This page applies it to Oracle and nothing else — the company's full record is in the dossier.