Oracle's massive capital spending increase—jumping to $70 billion（約11兆円） for fiscal 2027 from $50 billion（約8兆円）—has spooked investors and exposed a critical tension in the AI build-out: enormous upfront costs requiring companies like OpenAI and Anthropic to go public while capital demands remain steep.

What happened: Oracle reported strong quarterly earnings with 20% revenue and earnings growth and 47% cloud revenue growth, but the stock dropped about 11% in premarket trading because capex spending jumped to $56 billion（約9兆円） in fiscal 2026 (up from $21 billion（約3.4兆円） a year ago) and is projected to reach $70 billion（約11兆円） in 2027. The company's backlog of remaining performance obligations stands at $638 billion（約100兆円）, though over half of that revenue is expected more than three years out.

Why it matters: Oracle is saddling itself with nearly $130 billion（約21兆円） in debt (growing fast) and negative free cash flow for the first time in a long time, betting that OpenAI, Anthropic, and other AI companies will be able to pay for their committed spending. With over 50% of that $638 billion（約100兆円） backlog tied to OpenAI alone (down from virtually all of it previously), the company faces concentration risk—though the backlog has grown by $85 billion（約14兆円） sequentially, showing demand is broadening beyond one customer.

What to watch: Both OpenAI and Anthropic are preparing to go public—Anthropic recently raised $65 billion（約10兆円） and confidentially filed its IPO paperwork, while OpenAI has also filed confidentially. Participants in the podcast debate whether this rush reflects genuine capital desperation or an opportunistic move to raise money while market valuations and IPO sentiment remain favorable; the timing clusters three of the largest IPOs on record (Anthropic, OpenAI, and SpaceX) closely together.