Oracle Flags Data-Center Risks to Its AI Infrastructure

Oracle's fiscal year 2026 annual report, filed with the SEC in late June, discloses that capital expenditures rose to $55.7 billion for the year ended May 2026, up from $21.2 billion in fiscal 2025, and that the company is guiding to $90 billion to $95 billion in capex for fiscal 2027 (Oracle 10-K; Gizmodo). The filing states, "To grow our OCI business, which requires increased computing capacity, we must incur significant capital and operating expenditures." Gizmodo, which first reported on the filing, catalogued dozens of specific risks Oracle listed, including overbuilding, customer defaults, stranded capacity, power and GPU shortages, permitting and construction delays, tariff and export-control exposure, and cybersecurity and reputational risk.

Oracle has committed to large-scale AI infrastructure deals, including capacity commitments tied to OpenAI and Meta, and is a partner in the Stargate data-center project alongside OpenAI and SoftBank. Gizmodo reports Oracle explicitly flagged counterparty risk, noting that some of its customers "may be highly leveraged" and that Oracle could face "risks of non-payment and non-performance" - relevant given that AI labs such as OpenAI and Anthropic continue to spend well beyond their revenue. Oracle shares have fallen sharply over the past month amid broader investor caution about AI-infrastructure spending, per Gizmodo's reporting.

Teams evaluating cloud AI capacity commitments should treat Oracle's disclosure as a template for the risk categories that matter in practice: power availability and grid strain, GPU supply timing, fixed-price power contracts versus volatile energy costs, and permitting or zoning delays that can push commissioning dates. Procurement and finance teams negotiating long-term capacity or reserved-instance commitments should factor in the possibility of stranded capacity or renegotiated terms if demand growth slows relative to the scale of committed spending.

Watch whether other hyperscalers (Microsoft, Google, Amazon, Meta) disclose similarly detailed AI-infrastructure risk language in upcoming filings, how Oracle's fiscal 2027 capex guidance of $90-95 billion tracks against actual spending, and whether GPU-hour pricing or training-queue times shift as a leading indicator of capacity strain. Oracle's own credit exposure to AI-lab customers is also worth tracking as those companies report their financials.