Strange Fall of Oracle | How a $300 Billion AI Deal Changed Everything?

Summary

The video explains Oracle’s strategic shift from a high-margin database software company to a capital-intensive AI infrastructure provider—and why that transition has made investors more concerned about debt, cash flow, and execution.

Business strategy and evolution

• Oracle’s original value proposition: In the late 1970s, banks and large organizations struggled with fragmented data stored across separate systems. Oracle’s commercial SQL database helped them query and manage information in one place, improving decisions such as lending and credit risk.
• Asset-light software model: For decades, customers bought and operated their own servers while Oracle sold the software. The video describes this as a “golden rule”: let customers own the hardware.
• Cloud disruption: Amazon Web Services popularized renting computing power by usage rather than buying servers upfront. This gave startups access to scalable infrastructure without large initial investments. Oracle was slow to embrace cloud computing; the video says it did not seriously invest until 2016. By then, new software-license sales had fallen 15% in a year, and Oracle’s revenue had declined from $38.2 billion to $37 billion.
• AI infrastructure pivot: Oracle’s reported deal to provide computing capacity to OpenAI made the company look like an AI infrastructure leader. But unlike its former software model, this strategy requires substantial spending on land, data centers, power, and Nvidia chips before the revenue arrives.

Frameworks and business playbooks

• “Build first, sell later” J-curve: Infrastructure projects incur large upfront costs, followed by a period of losses or low returns before the investment pays back. The video’s example spends $100 and earns $30 per year, reaching profitability only in year four.
• Infrastructure stress test: The narrator tests Oracle’s exposure through three downside scenarios:
  1. Additional borrowing increases annual interest costs.
  2. Refinancing older, cheaper debt at higher rates raises interest expense.
  3. Delayed customer payments, incomplete power connections, or price competition reduce operating profit while costs continue.
• Capacity and timing risk: Building facilities is not enough; power connections and customer readiness must align with construction and chip deployment.
• Talent-hiring promotion: A sponsor segment advertises Series A Hiring’s process: define a founder’s needs by phone, have the firm screen candidates, and deliver five shortlisted candidates within five days.

Key metrics cited

The figures below are presented as stated in the subtitles; some revenue figures appear to use different periods or definitions.

• Oracle reportedly generated $17 billion in profit and held $31 billion in cash.
• Its stock was said to have risen 87% in 2025, reached a market value near $922 billion, and later fallen about 56% from its peak. The opening also describes a 50% decline over the prior year.
• Oracle reported $455 billion in remaining performance obligations (RPO), compared with Wall Street expectations of roughly $180 billion.
• The reported OpenAI agreement was worth about $300 billion over five years, with payments starting in 2027—about $60 billion annually once underway.
• At the time of the deal, OpenAI’s annual revenue was described as $12 billion, while Oracle’s operating cash flow was $21.5 billion. The video later says OpenAI raised $122 billion at an $852 billion valuation, and its annual recurring revenue had approached $70 billion.
• Oracle’s backlog later reportedly reached $664 billion, including an additional $209 billion after the OpenAI deal.
• Oracle was said to have spent $56 billion on construction and equipment in 2026, followed by another $28 billion in the next three months—$84 billion over 15 months—while receiving $55 billion in cash.
• The video contrasts software gross margins of nearly 70% with estimated margins of 30–40% for the infrastructure model.
• In the stress test, Oracle’s interest costs rise from $5.7 billion to $8.5 billion after a hypothetical $40 billion loan at 7%, then to about $9.5 billion after refinancing. If profit falls from $27 billion to $20 billion, interest would consume nearly half of profit.
• Oracle’s refinancing costs are described as rising from about 1.65% in 2021 to around 6.5% in 2026, with roughly $17 billion of older debt due for repayment over the next two years.
• The credit-default-swap example says insurance on $1 million of Oracle debt rose from about $5,000 to $20,000, with the indicator reaching 200 basis points.

Risks and takeaways

• Backlog is not cash: RPO represents contracted future business, not revenue already collected. The OpenAI payments were not expected to begin until 2027.
• Customer concentration and affordability: The scale of the OpenAI commitment is large relative to the customer’s revenue as described at the time of signing, leaving investors focused on whether OpenAI can meet the payments.
• Debt magnifies downside: Interest payments remain due even if a data center is delayed, underused, or earns less than expected.
• Technology can outpace construction: If newer chips make facilities or equipment less competitive before they pay back, Oracle may need further investment or could fail to earn an adequate return.
• Execution matters as much as demand: Oracle’s opportunity depends on delivering infrastructure, securing power, keeping it technologically current, and converting its backlog into cash and profits.

The video’s conclusion is not that Oracle’s business is necessarily failing, but that its AI growth strategy has changed its financial risk profile. The outcome depends on whether demand, customer payments, and infrastructure returns arrive quickly enough to justify the borrowing and upfront spending.

Presenters and sources mentioned

Think School narrator; Oracle’s quarterly results and RPO figures; OpenAI deal and funding details as described in the video; AWS and Oracle historical business developments; market and credit-default-swap data cited in the narration; and the Series A Hiring sponsor segment.