‘A giant going all-in on AI infrastructure’: BofA reinstates Oracle at Buy

BofA reinstated Buy on Oracle with a $200 price target (~30% upside) citing accelerating AI infrastructure and cloud demand and a $553 billion RPO backlog. Analysts forecast revenue growth of 17%, 33% and 46% for FY26–FY28 and expect ~1,200bps of gross margin compression, while flagging >57% of RPO scheduled beyond three years, customer concentration (e.g., OpenAI), a capex ramp to ~ $50B in FY26 and negative free cash flow through FY29 requiring external financing.

Analysis

Oracle’s pivot to hyperscale AI infra is a dual-edged sword for the ecosystem: it creates a durable, sticky demand stream for high-end GPUs, power/cooling systems, and specialized networking, which should continue to benefit Nvidia, TSMC, and tier-1 data-center OEMs for multiple years. But large forward procurements from a single incumbent customer suppress spot availability and can spike input costs for smaller cloud players and enterprises that can’t commit long-duration contracts; expect pricing dislocations in the procurement market for 6–24 months as capacity catches up.

The clearest operational risk is cadence: converting long-duration contractual backlog into recognized revenue requires synchronized deliveries of GPUs, racks, and interconnects, plus partner software integration — any slippage creates cascading margin pressure and cash burn. Near-term catalysts to watch are tranche-level delivery notices, quarterly backlog-to-revenue conversion rates, and any financing events; a missed GPU delivery window or a downgraded capex plan would likely compress equity materially within days-to-weeks, while successful scale should play out over 18–36 months.

Consensus underappreciates the refinancing and working-capital mechanics here: front-loaded capex means Oracle must bridge several years of negative FCF if conversions lag, which increases dilution/default tail-risk relative to a pure software comparible. That makes asymmetric trades attractive — buying convex upside to successful execution while limiting exposure to execution/financing failure is the prudent route for the fund given the opaque customer concentration and multi-year buildout timeline.

Key Decisions for Investors

• Long ORCL equity + protective put: Buy ORCL stock size A (mnemonic exposure consistent with portfolio sizing) and purchase a 12–18 month protective put (strike ~10–15% out-of-the-money) to cap downside. R/R: asymmetric — ~30–40% upside if execution accelerates with downside limited to put premium over 12–18 months.
• Relative-value pair: Long ORCL / Short AMZN (equal notional cloud exposure) for 12–24 months to express conviction that committed AI infrastructure spend will reallocate incremental enterprise demand toward OCI. Target 30% relative outperformance; risk is a broad AI demand surge that lifts both legs — set a 20% stop-loss on the short leg.
• Gamma play on GPU squeeze: Buy NVDA 6–12 month calls (moderate size) to capture continued GPU tightness/price power as hyperscale builds ramp. Hedge 20–30% of notional by buying ORCL short-dated puts (6 months) to protect against a scenario where Oracle’s heavy pre-orders normalize GPU pricing and compress vendor equity multiples.