TD Cowen reiterates Oracle stock Buy rating on cloud momentum By Investing.com

Oracle beat on quarterly revenue and EPS, with total revenue up 21% year-over-year and EPS at $2.11 versus $1.96 expected. Oracle Cloud Infrastructure revenue rose 92% in constant currency, remaining performance obligations jumped 363%, and TD Cowen reiterated Buy with a $300 target while lifting estimates on AI-driven growth. Offsetting the strength, SaaS growth of 9% came in below estimate and gross margin pressure kept management from raising guidance.

Analysis

The market is still pricing Oracle like a software rerating story when the real earnings engine is a capacity-constrained infrastructure buildout. That matters because infrastructure providers with visible backlog and pre-funded demand usually de-rate on gross-margin fear before they re-rate on operating leverage; the key inflection is when capex intensity peaks and revenue converts faster than depreciation. If management proves it can keep incremental supply online without a second margin reset, the next leg is not a modest multiple expansion but a regime change in long-duration EPS estimates.

The second-order winner is the AI supply chain: hyperscaler alternatives, networking, power, land, and grid-adjacent beneficiaries should all see follow-through as customers diversify away from single-vendor concentration risk. The likely loser is pure-play enterprise software peers that cannot show a similar backlog-to-revenue conversion rate, because investors will increasingly compare every AI narrative against hard contracted demand rather than pilot announcements. That raises the bar for names selling “AI-enabled” features without true infrastructure economics.

Near term, the main risk is that the stock has already moved to a valuation where any hint of delivery slippage, delayed capacity, or margin disappointment can trigger a sharp drawdown. Over the next 1-3 quarters, the setup is highly event-driven: results that confirm accelerating remaining obligations and stable build economics should squeeze shorts, while any commentary implying supply bottlenecks or capital needs could compress the multiple quickly. The consensus may be underestimating how much of the current rerating depends on execution credibility rather than just AI demand.

The contrarian take is that the best risk/reward is not chasing the common-stock breakout after a big run, but owning the volatility around the next few print dates. If backlog converts cleanly, upside from here is still meaningful because earnings estimates are likely too low; if it does not, the stock can mean-revert hard as a duration trade. In other words, this is a fundamentals-confirmation trade, not a momentum trade.

Key Decisions for Investors

• Stay long ORCL on pullbacks, but only with a 3-6 month horizon; upside case is estimate revisions driving another 15-25% re-rate if capacity delivery remains on schedule, while downside is a fast 10-15% gap down on any margin/capex miss.
• Buy ORCL call spreads into the next earnings cycle rather than outright shares; this captures upside from a confirmation print while limiting drawdown if the market is already front-running the backlog story.
• Pair trade: long ORCL / short a basket of large-cap enterprise software names with weaker AI monetization evidence over the next 1-2 quarters, as capital should continue rotating toward businesses with contracted infrastructure demand.
• For a cleaner AI infrastructure expression, add exposure to power/grid beneficiaries and networking suppliers on any ORCL weakness; the second-order spend is likely to broaden beyond compute as capacity ramps.
• Tighten risk if ORCL loses post-earnings support or management signals a delay in new capacity; that would be the first cue to reduce because the stock is now more sensitive to execution than narrative.