Oracle's $100B AI data center buildout draws scrutiny on debt

Oracle's debt-financed push to build AI data centers worldwide has loaded billions in liabilities onto its balance sheet, a New York Times investigation found.

Oracle's debt-financed global AI data center buildout has piled billions in liabilities onto its balance sheet, a New York Times investigation published July 31 found, raising questions about the sustainability of the industry's capital-spending boom. The company, led by billionaire chairman Larry Ellison, has borrowed heavily to construct facilities around the world, betting that surging demand for artificial-intelligence computing will outpace the cost of financing.

"Oracle's expansion is a wager that AI demand will keep growing faster than the interest bill on the debt it took on to build," said Rachel Kim, an analyst covering AI infrastructure at Edgen. "If that demand stalls, the leverage becomes a problem not just for Oracle but for every hyperscaler chasing the same contracts."

The scale of the bet is visible in Oracle's market moves. The company added roughly $100 billion in market value on a major announcement earlier this year, and it reshuffled its executive ranks, naming Giancarlo Magouyrk and Douglas Sicilia co-CEOs while Safra Catz moved to executive vice chair. The exact size of Oracle's debt load tied to the data center program was not yet disclosed in the investigation, but the report described the liabilities as "ominous" relative to the company's cash generation.

The Cash-Strapped AI Buildout

Oracle is not alone in stretching its balance sheet. The broader AI infrastructure buildout is getting cash-strapped, with hyperscalers and cloud providers committing tens of billions of dollars to data centers, graphics processing units and power capacity before revenue from those assets materializes. Oracle's rivals — Microsoft, Amazon and Alphabet's Google — have each pledged record capital expenditures, but Oracle's smaller revenue base makes its debt burden proportionally heavier.

The competitive stakes are direct. Oracle has positioned its cloud as a home for AI workloads, competing for enterprise customers against Amazon Web Services and Microsoft Azure. Every dollar Oracle spends on data center capacity is a dollar it cannot return to shareholders or use to service existing obligations, and the company's ability to win AI contracts now determines whether that spending pays off.

What the Debt Means for Investors

For investors, the investigation sharpens a question hanging over the entire AI trade: how much of the sector's capital spending is justified by real demand, and how much is a self-reinforcing bubble? Oracle's stock, which has rallied on AI optimism, now carries the added risk of a credit downgrade if its leverage keeps climbing. The company's data center capacity, measured in megawatts, has expanded rapidly, but the revenue per megawatt remains unproven at scale.

The implications extend beyond Oracle's own shares. If the AI capex cycle falters, the fallout would hit chipmakers like Nvidia, whose data center revenue depends on hyperscaler spending, as well as power utilities and construction firms tied to the buildout. Oracle's debt load is a leading indicator for the sector: if the company must slow its expansion to service its liabilities, it would signal that the industry's spending pace is not sustainable.

Oracle has not yet disclosed a timeline for when its data center investments will generate returns sufficient to cover its financing costs. The company's next earnings report will be closely watched for signs that its AI revenue growth is keeping pace with its debt service.

This article is for informational purposes only and does not constitute investment advice.