Oracle needs $45-50 billion this year to build cloud infrastructure for customers including OpenAI, Meta, NVIDIA, and xAI. The database giant announced its 2026 funding plan on February 1, revealing a 50-50 split between bond issuances and equity programs.
The scale reflects Oracle's position: $455 billion in booked cloud services it hasn't delivered yet, and Q3 commitments of $7 billion pushing projected infrastructure revenue growth to 70% this fiscal year. The company is collaborating with OpenAI on the Stargate project - a $450 billion initiative targeting over 8 gigawatts of data center capacity.
The financing structure matters here. Oracle is issuing senior unsecured bonds (one-time, early 2026), mandatory convertible preferred securities ("modest portion"), and up to $20 billion through a new at-the-market equity program. Big Red emphasized twice that this maintains "investment-grade" status - notable because investors sued in February over claims Oracle concealed its debt needs.
The market is skeptical. Oracle's stock dropped from $328 in September 2025 to around $164 by late January 2026. Credit default swap costs hit five-year highs in December. The concern: Oracle's fortunes are increasingly tied to OpenAI, which is unprofitable and hasn't detailed its own infrastructure financing.
Three things to watch:
- Whether Oracle can monetize that $455 billion backlog fast enough to justify both debt burden and equity dilution
- Stargate project execution - 1 gigawatt in Michigan alone is substantial infrastructure risk
- Customer concentration - if OpenAI's plans change, Oracle's expanded capacity becomes a very expensive bet
The company is betting it can compete with AWS, Google, and Microsoft (each generating $250+ billion annually vs. Oracle's $57 billion) through AI-focused infrastructure. History suggests Oracle is better at database margins than cloud infrastructure ones. We'll see if this time is different.