Oracle's $165B AI Megacampus Bet Hits Cost Overruns — BBB Rating, $55.7B FY26 Capex, New Debt/Equity Raise Looms

Multiple outlets reported that Oracle is dealing with multi-billion-dollar cost overruns across its $165B AI data-center buildout, including the Wisconsin and New Mexico/“Project Jupiter” sites. The key hard confirmation we can anchor in filings: Oracle capex jumped to $55.7B in fiscal 2026 (ended May 31, 2026) — from $21.2B in fiscal 2025 — which is the financial signature of “overrun-like” execution risk even before we debate the exact “$165B campus” project-accounting.

Beyond headline capex, the most credit-relevant details are the long-dated commitments and capital-markets instruments that sit behind future build and power/network capacity. In Oracle’s FY2026 10-K, the company discloses additional lease commitments and unconditional purchase obligations tied largely to data center arrangements — plus explicit equity and notes activity in early 2026.

Cost overruns don’t only inflate capex totals — they also alter timing. In hyperscaler-like builds, timing is the difference between “spend now, monetize soon” and “spend now, monetize after power capacity is ready.” That timing shift increases leverage and reduces near-term free cash flow, which is exactly what rating models punish.

Reports tied New Mexico/Project Jupiter cost overruns to local energy and infrastructure concerns. That fits a broader data-center supply chain reality: GPUs are “component constrained,” but data centers are “system constrained” by interconnection, substations, and utility upgrade lead times.

Oracle’s downstream is cloud demand (enterprise + developers) consuming OCI capacity. The risk isn’t only that Oracle spends too much; it’s that capacity monetization per dollar of capex can fall if usable availability ramps slower than expected or at higher cost.

Even without naming every vendor in the filings, the commitments Oracle discloses effectively identify the supply-chain nodes the buildout depends on: data-center power (utility upgrades, grid interconnect equipment) and long-lived cloud infrastructure purchases. Downstream is the cloud consumption layer that turns deployed capacity into recurring revenue.

Reports say Oracle’s credit rating slide to BBB coincided with a Wisconsin Public Service Commission dispute involving financial guarantees that could cost more than $100M annually. We can’t fully validate the exact “BBB downgrade + $100M” chain from Oracle’s SEC filing extract alone in this session, but we can validate Oracle’s capex intensity and its credit-sensitive borrowing structures — the exact ingredients that make regulators and utilities tighten collateral rules.

Oracle didn’t wait for the FY2026 cash squeeze to appear in hindsight. It established an ATM equity program up to $20B and issued large notes plus mandatory convertible preferred depositary shares in early 2026, which is consistent with an approach of keeping funding optionality open during a long infrastructure ramp.

The evidence we can hard-anchor from filings supports one thing strongly: Oracle has entered a heavy infrastructure build phase (FY2026 capex $55.7B vs $21.2B), backed by long-dated lease/power commitments and early-2026 capital-market financing. The critical question is whether OCI revenue ramp and utilization keep pace with infrastructure readiness — especially where the power grid is the bottleneck.