甲骨文披露高达2,480亿美元的租赁付款承诺，几乎全部与AI资料中心与云端容量相关，将在现在至2028财年之间开始生效，且多数尚未入表。这个数字比9月财报附注所
公司预期本财年至5月的资本支出将翻倍以上至500亿美元，约等于其同期预测营收的75%。分析师估计未来五年累计资本支出将超过3,000亿美元，自由现金流可能连续四年为负；在BBB评级与约1,050亿美元净负债之下，新租约长达15至19年，按中位数推算年租金约146亿美元，约为过去12个月223亿美元营运现金流的三分之二。
管理层主打规模化成长：云端基础设施部门到2030年年营收目标1,660亿美元（超过去年的16倍），租赁AI算力的毛利率估计30%至40%。但其5,230亿美元已签约收入积压中约75%将在60个月内入帐，而据报约3,000亿美元的OpenAI合约分布约五年，凸显「租期更长、客期更短」与提前解约费风险；再加上设备六年折旧、可能中途升级、对未来算力供给过剩的疑虑、股价自9月高点下跌44%，以及融资成本上升，2,480亿美元的租赁曝险成为信心压力测试。
Oracle disclosed $248 billion of lease-payment commitments, largely for AI data centers and cloud capacity, starting before fiscal 2028 and mostly off balance sheet. The figure is almost $150 billion higher than September’s footnote disclosure, a scale that exceeds peers. Investors worry the long property leases outlast much shorter customer contracts, including big AI clients such as OpenAI.
Oracle expects capital spending to more than double to $50 billion in its fiscal year ending in May, about 75% of forecast revenue. Analysts see cumulative capex above $300 billion over five years, with free cash flow likely negative for four. With a BBB rating and roughly $105 billion net debt, new leases running 15–19 years imply about $14.6 billion annual rent—around two-thirds of $22.3 billion operating cash flow.
Management pitches scale: cloud-infrastructure sales of $166 billion by 2030 (over 16× last year) and 30–40% gross margins on rented AI capacity. Yet 75% of its $523 billion contracted backlog is due within 60 months, while a reported $300 billion OpenAI deal spans about five years, leaving lease-duration risk and potential termination fees. Add six-year equipment depreciation, upgrade risk, fears of an AI capacity glut, a 44% share-price drop since September’s peak, and rising borrowing costs, and the $248 billion rent exposure looks like a major confidence test.
