Business & policy

Oracle's $55.7 billion data centre splurge outpaces guidance as AI bets raise debt concerns

At a glance:

  • Oracle spent $55.7 billion on data centres in FY2026, exceeding its $50 billion guidance by $5.7 billion.
  • The company plans to raise $40 billion in equity and debt in FY2027 to fund continued AI infrastructure expansion.
  • A $638 billion backlog of contracted future revenue includes $75 billion in prepaid AI hardware, with 42% of the Stargate Abilene campus already operational.

Oracle's capital expenditure surge

Oracle's fiscal fourth-quarter results revealed a staggering $55.7 billion in capital expenditure for the year ended May 31, 2026, surpassing its own $50 billion guidance by $5.7 billion. This aggressive spending underscores the company's commitment to AI-driven infrastructure, though it has raised concerns among investors about debt sustainability. Revenue for the quarter reached $19.2 billion, up 21% year-on-year, while adjusted earnings of $2.11 per share beat analyst expectations of $19.1 billion and $1.97 per share. However, the stock fell 7% in after-hours trading as markets reacted to the scale of ongoing investments.

The company now projects $70 billion in net capital expenditure for fiscal 2027, with $20 billion to $25 billion of that attributed to prepayments for components. CFO Hilary Maxson highlighted that this spending is necessary to meet demand for AI workloads, particularly through large-scale contracts. Oracle raised $43 billion in debt and $5 billion in equity during FY2026, and plans to secure another $40 billion in the coming year, including $20 billion through a share-selling program. With $117 billion in debt, Oracle is the largest non-financial issuer in the Bloomberg US high-grade corporate bond index, reflecting the tech sector's broader trend of debt-fuelled AI expansion.

Cloud growth and AI-driven contracts

Cloud infrastructure revenue, Oracle's core AI workload business, surged 93% to $5.8 billion in the quarter, with total cloud revenue (including software) hitting $9.9 billion, up 47%. For the full fiscal year, cloud infrastructure revenue reached $18.1 billion, a 77% increase. The company guided for a 61% jump in total cloud revenue in the current quarter, slightly below the 62% analysts expected. This growth is fueled by large AI contracts, with four customers each signing deals worth over $8 billion in the quarter. Prepaid and customer-supplied hardware for these contracts now total $75 billion, significantly reducing Oracle's need to raise additional capital for AI data centre construction.

The $638 billion remaining performance obligations backlog, up from $553 billion in the previous quarter, represents contracted future revenue. However, only 12% is expected to convert within 12 months, and 34% within three years, meaning most of this revenue is years away. Analysts question whether AI demand will sustain such high levels of capital spending, especially as Oracle takes on more debt in a sector already under scrutiny for profitability risks.

The Stargate project and OpenAI partnership

Oracle's Stargate campus in Abilene, Texas, a key part of its partnership with OpenAI and SoftBank, has delivered 42% of its capacity, with an additional 35% scheduled for completion in the next three months. The project, spanning Texas, Wisconsin, Michigan, and New Mexico, aims to build nearly seven gigawatts of capacity and is projected to involve over $400 billion in investment. Co-CEO Clay Magouyrk emphasized significant progress at the site, which serves as a flagship for Oracle's AI infrastructure ambitions. The partnership highlights Oracle's strategy to position itself as a critical player in the AI ecosystem, leveraging its hardware and cloud capabilities to support high-profile clients like OpenAI.

Workforce restructuring and cash flow reallocation

In March, Oracle initiated a restructuring that cut up to 30,000 employees, roughly 18% of its global workforce—the largest in the company's history. Restructuring charges totaled $1.8 billion, nearly five times the previous year's amount. TD Cowen estimates the layoffs will free $8 billion to $10 billion in annual cash flow, which Oracle is directing toward AI data centre construction. The Oracle Health division, built around the $28.3 billion Cerner acquisition, was hardest hit, with 8,000 to 10,000 employees let go. This reallocation of resources signals Oracle's prioritization of AI infrastructure over traditional healthcare IT operations.

Market reaction and long-term risks

Despite revenue and earnings beats, Oracle's stock dropped 7% in extended trading, reflecting investor unease about the scale of its capital expenditure and debt plans. The stock had risen 35% in the three months prior, indicating elevated expectations that the company's aggressive AI spending may struggle to meet. The $638 billion backlog, while impressive, is largely unrecognised revenue, with most of it years away from conversion. As Oracle continues to invest heavily in AI infrastructure, the key question remains whether the returns will justify the costs, particularly as the tech sector faces growing skepticism about the profitability of such large-scale infrastructure bets.

Editorial SiliconFeed is an automated feed: facts are checked against sources; copy is normalized and lightly edited for readers.

FAQ

What drove Oracle's $55.7 billion capital expenditure in FY2026?
Oracle's capital expenditure was primarily driven by investments in AI data centres, including infrastructure for large-scale contracts with customers like OpenAI. The company also cited prepayments for components as a contributing factor, with $20 billion to $25 billion of FY2027's projected $70 billion capex linked to these prepayments. This spending aligns with Oracle's strategy to capitalize on the AI boom, though it has raised concerns about debt sustainability.
How does Oracle's $638 billion backlog impact its financial outlook?
The $638 billion remaining performance obligations backlog represents contracted future revenue, with $75 billion coming from prepaid AI hardware and customer-supplied equipment. However, only 12% of this backlog is expected to convert within 12 months, and 34% within three years, meaning most revenue is years away. This raises questions about whether AI demand will sustain Oracle's aggressive capital spending, especially as investors scrutinize the profitability of such infrastructure investments.
What is the Stargate project and its role in Oracle's AI strategy?
The Stargate project is a multi-site AI infrastructure initiative involving Oracle, OpenAI, and SoftBank, with locations in Texas, Wisconsin, Michigan, and New Mexico. The flagship Abilene, Texas campus has delivered 42% of its capacity, with 35% more planned in the next three months. The project aims to build nearly seven gigawatts of capacity and represents over $400 billion in projected investment, positioning Oracle as a key player in AI hardware and cloud services.

More in the feed

Prepared by the editorial stack from public data and external sources.

Original article