Oracle shares tumble on earnings. But there's a silver lining for our AI chip and power stocks

Oracle beat May-quarter expectations on both revenue and earnings, but the stock is pressured by plans to raise about $40 billion in debt and equity in fiscal 2027, including $20 billion of already announced at-the-market issuance. Remaining performance obligation jumped 363% year over year to $638 billion, with management citing surging AI demand and nearly 1 GW of FY27 Q1 delivery. The article is constructive on AI infrastructure demand, but mixed for Oracle shares because capital intensity and future dilution/depreciation remain major concerns.

Analysis

The key takeaway is not Oracle’s own equity story; it is that the AI supply chain is still constrained by demand, not by capital. When contracted backlog is expanding faster than delivery capacity, the bottleneck shifts downstream to GPUs, networking, power, and data-center electrical gear, which should keep pricing power elevated longer than consensus expects. That supports the higher-quality picks-and-shovels names over the platform layer, where capex intensity and depreciation are becoming the main source of earnings risk.

The second-order winner is the power stack. If delivery is approaching near-term gigawatt scale, incremental load is large enough to tighten lead times for turbines, switchgear, transformers, and high-density interconnect components, which is constructive for GEV, ETN, and GLW even if the market initially focuses only on semis. A less obvious implication is that older-generation accelerators remain monetizable, which extends useful life curves and reduces near-term replacement urgency, helping utilization-heavy suppliers like NVDA and ARM while capping the fear of rapid obsolescence.

The vulnerable names are the hyperscalers and Oracle itself because the market is moving from a scarcity narrative to a unit-economics narrative. If customer economics weaken or AI software pricing compresses, the very backlog that looks bullish today can become a solvency question in 2-4 quarters, especially for smaller customers funding usage with venture capital or loss-making model businesses. That makes the current pullback in ORCL more of a financing-discount issue than a demand issue, but it also means the broader AI trade can derate quickly if one large customer disappoints or restructures commitments.

Key Decisions for Investors

• Add to NVDA on weakness over the next 1-3 sessions; use the Oracle selloff as a sentiment dislocation entry. Risk/reward favors a 3-6 month hold as backlog validation should keep utilization tight and support multiple resilience.
• Initiate a pair trade: long GEV / short ORCL for 1-2 quarters. Thesis: power and equipment suppliers monetize AI capex immediately, while ORCL carries financing and depreciation overhang.
• Add ETN and GLW on any 2-5% pullback. These names are leveraged to the physical buildout with less direct exposure to AI pricing wars; downside should be more limited than cloud/platform peers.
• Reduce exposure or hedge MSFT, AMZN, GOOGL, and META via short-dated puts into the next earnings cycle. The risk is not demand collapse, but multiple compression if capex stays elevated without clear revenue inflection.
• Speculative long ARM versus INTC over the next 3-6 months. ARM benefits from broadening compute demand across generations, while INTC remains a higher-risk beneficiary with slower AI monetization and execution uncertainty.