AVGO, ORCL and MU: comparing three AI infrastructure stocks after a sector selloff

Broadcom, Oracle and Micron fell roughly 4.5%-4.9% in a coordinated technology-sector selloff, with the stocks trading 26.1%-56.4% below their 52-week highs. The article frames Broadcom as the strongest sentiment-reversion candidate on $26.9B of free cash flow and 77.3% FY2025 gross margin, Micron as the lowest-valuation AI-memory play at 13.1x forward P/E, and Oracle as the highest-risk option due to 273.6% debt/equity and -$23.7B levered FCF. Despite the pullback, cited analyst target upside ranges from 47.8% for Broadcom to 59.7% for Oracle, though Oracle's leverage and cash-burn risks remain material.

Analysis

The key signal is that AI infrastructure equities are selling despite a potentially favorable disinflation impulse from lower energy prices. That points to factor-level de-risking—duration, crowded AI ownership, and capex-payback anxiety—rather than an oil-driven earnings shock. In the next several sessions, high-beta semis should remain vulnerable to systematic selling; a lower real-rate backdrop becomes supportive only if Treasury yields actually decline rather than credit spreads widen.

AVGO is relatively insulated from commodity-cycle risk but is most exposed to a reassessment of hyperscaler capital-intensity and custom-silicon concentration. Its valuation can hold only if the next results demonstrate that accelerator and networking demand is broadening across customers; a flat AI revenue outlook would likely compress the premium multiple further. NVDA is the cleanest read-through: any evidence that hyperscalers are shifting incremental workloads toward internal ASICs is a relative positive for AVGO but not necessarily an absolute positive if total AI budgets decelerate.

MU has the strongest balance-sheet optionality, but memory pricing—not HBM certification—will determine near-term earnings revisions. The market is likely underpricing the possibility that conventional DRAM/NAND supply discipline weakens as peers chase AI-adjacent demand; HBM margins cannot fully offset a broad memory-price reversal. Conversely, sustained DRAM contract-price increases through the next quarterly negotiation cycle would make MU's current multiple look less cyclical and support upside over 3-6 months.

ORCL is the weakest risk-adjusted expression of AI spending because its equity outcome depends on external financing conditions as much as backlog conversion. Negative free cash flow is tolerable during buildout only while debt markets remain open and cloud capacity produces accelerating utilization; any upward revision to capex, delayed capacity delivery, or wider credit spreads could force a multiple reset well before revenue recognition catches up. The contrarian view is that lower energy may ease inflation and financing pressure, but that benefit requires lower long-end yields, not merely lower crude.

Key Decisions for Investors

• Do not buy the broad AI dip immediately; wait 3-5 trading days for AVGO/MU relative performance versus NVDA and the SOXX ETF to stabilize. A continued underperformance of SOXX versus the S&P 500 despite falling Treasury yields would invalidate a simple rate-relief long thesis.
• Initiate a 1-3 month pair: long AVGO / short ORCL in equal dollar exposure after a sector bounce. AVGO has higher-quality cash conversion, while ORCL carries materially greater capex, refinancing, and execution sensitivity; exit if ORCL reports clear free-cash-flow improvement or if AVGO's next AI revenue outlook fails to reaccelerate.
• Accumulate MU only in tranches, with a 3-6 month horizon, contingent on independently tracked DRAM contract prices remaining positive into the next pricing round. Use a 7-10% position-risk stop or protective puts because a reversal in DRAM pricing can drive earnings estimates down faster than the headline valuation implies.
• Use ORCL as a watch rather than a standalone long until management provides evidence of cloud-capacity utilization and a credible path to positive levered FCF. A widening of its credit spread, incremental debt issuance, or capex guidance increase would be a short/avoid trigger; falling long-term yields and improving FCF would be the required catalyst to revisit.