Oracle Stock Plummeted by 25% in the First Half of 2026, and This Dire AI Warning Might Be Why

Oracle’s AI data-center demand looks massive but quality is questioned: remaining performance obligations (RPO) surged 363% YoY to $638B, yet Oracle warned it expects to convert only ~12% of RPO into revenue over the next 12 months. The article flags counterparty concentration (about $300B of RPO tied to OpenAI) given OpenAI’s ~$25B annualized revenue and heavy reliance on external funding (recent $122B capital raise), creating non-payment/non-performance risk. With Oracle carrying $122B in long-term debt and planning an additional ~$40B via debt/equity to fund more capacity—and shares down ~25% in 1H 2026 and >50% from last year’s high—the piece frames a high-risk demand/demand-release scenario for AI infrastructure economics.

Analysis

The market is still valuing ORCL’s AI buildout as if backlog were cash, but the real variable is funding quality. When customer concentration is paired with levered data-center capex, equity behaves like a spread trade on the customers’ access to capital and Oracle’s ability to refinance at acceptable terms; that is a much weaker setup than a normal software backlog. Near term, the first damage usually shows up in multiples and credit spreads, not in GAAP revenue, so the stock can keep derating even before any contract default becomes visible.

Relative winners are the balance-sheet fortress names that can self-fund AI capex and monetize it across broader platforms, especially MSFT and META. NVDA and AMD are less clear-cut: they benefit if capacity demand migrates to healthier buyers, but Oracle slowdowns could also create a temporary air pocket in GPU orders and delay revenue recognition. Over the next 1-3 months, the key catalyst is whether Oracle must raise capital on punitive terms; over 6-18 months, the risk is stranded AI infrastructure if utilization normalizes below current contract assumptions.

Consensus may be missing that the problem is not just demand durability but liquidity optionality. If Oracle can prove faster cash conversion, lower leverage, and stable customer funding, the thesis weakens materially; absent that, the equity remains vulnerable to a financing-driven de-rating even if AI demand stays strong. The move is not obviously overdone until the market sees evidence that RPO quality is financeable without recurring balance-sheet stress.

Key Decisions for Investors

• Short ORCL on rallies or buy 3-6 month put spreads; thesis is leverage + customer-funding risk compressing the multiple before revenue disappointments show up. Falsify if Oracle secures funding at tight spreads and cash collections accelerate.
• Pair trade: long MSFT / short ORCL for the next 1-3 months. MSFT has the stronger balance sheet and broader monetization, while ORCL is more exposed to funding and customer-concentration risk.
• Treat NVDA and AMD as conditional beneficiaries rather than immediate longs: wait for evidence that Oracle’s capex slowdown is temporary before adding. If ORCL’s financing path deteriorates further, the read-through is weaker GPU demand timing, not necessarily lower long-run AI demand.
• Watch ORCL credit spreads and any announced debt/equity financing as the leading indicator. A widening in bonds/CDS would be a stronger sell signal than the equity move alone.