Oracle raises $40B for AI infrastructure as investors grow more selective

Oracle is launching a major new fundraising effort — including a $20 billion share offering — to expand its AI data center capacity, a move that analysts say is accelerating a more discerning approach among investors toward AI infrastructure-related bonds.

Oracle plans to raise a total of $40 billion (about 60.8 trillion won) through bond and share issuances, including the previously announced $20 billion share offering, CNBC reported Wednesday.

The company had already raised $43 billion in debt and $5 billion in equity during fiscal year 2026, a pace that has stoked investor concerns about whether AI demand can justify such massive capital outlays.

Despite the heavy financing burden, Oracle posted adjusted earnings per share of $2.11 and sales of $19.18 billion (about 29.16 trillion won) for the fourth quarter of fiscal year 2026 (March through May), topping analyst expectations of $1.96 in EPS and $19.1 billion in sales.

Sales rose 21 percent from a year earlier. Oracle maintained its fiscal year 2027 sales outlook at $90 billion (about 136.8 trillion won) while raising its adjusted EPS forecast to $8.05.

Still, free cash flow for the full fiscal year came in at negative $23.7 billion (about 36.02 trillion won), and the share price tumbled in after-hours trading after the additional fundraising plan became public.

Amid all this, Bloomberg reported that Wall Street investors are moving away from blanket buying of bonds issued to fund AI data center construction by major tech companies, instead becoming more selective about which deals they back.

A recent Citigroup analysis found that bonds tied to Oracle, Meta, Google and Nvidia have largely tracked their parent companies' share price performance and attracted strong demand. By contrast, some Microsoft-related bonds issued with a non-callable, bullet-maturity structure saw weaker buying interest — a sign of investment differentiation driven by liquidity risk. The trend points to a broader shift in which investors are expected to scrutinize not just a tenant's creditworthiness but also the complexity of individual data center projects and their debt repayment structures.