2 Growth Stocks With Strong Earnings Prospects to Buy in October

NVIDIA reported fiscal Q2 revenue of $96.2 billion, up 106% year over year, and guided fiscal Q3 revenue to about $108 billion as AI-infrastructure demand remains strong; consensus forecasts FY2027 revenue growth of 88.0% and EPS growth of 93.9%. Quanta Services posted Q2 revenue growth of 41.1% to $9.56 billion and a 71% increase in adjusted EPS to $4.24, with backlog rising roughly 49% to $53.4 billion. The article favors both stocks despite a higher-for-longer rate backdrop, with 2026 PCE inflation projected at 3.7% and the federal-funds rate at 4.1% at year-end.

Analysis

The actionable distinction is duration of cash flows. NVDA remains a high-beta expression of hyperscaler capex, where even a modest deceleration in cloud-provider spending or a gross-margin reset can overwhelm another beat given elevated expectations. PWR monetizes the same AI buildout indirectly: incremental data-center load requires transmission, substations and interconnection work, making its demand less dependent on GPU replacement cycles and potentially more resilient if AI hardware spend rotates from training capacity toward power delivery.

PWR's backlog should not be valued as fully contracted earnings: utility project timing, permitting and labor availability can defer revenue conversion, while fixed-price execution creates margin risk if wages or equipment costs reaccelerate. Still, grid capex is supported by multi-year reliability mandates and electrification demand, providing a 6-18 month earnings-duration advantage over many AI-adjacent industrials. Beneficiaries beyond PWR include transmission-equipment suppliers ETN, HUBB and GEV; the more direct data-center power beneficiaries are VRT and CEG, though each carries greater valuation sensitivity to AI demand assumptions.

Near term, the October earnings setup favors companies that can raise forward estimates rather than merely meet consensus. The contrarian concern is that both names are already positioned as obvious growth winners: higher real yields can compress their multiples even with intact estimates. For NVDA, evidence of customer concentration, lower incremental gross margin, or a material gap between guidance and sell-side expectations would falsify a continuation trade; for PWR, backlog conversion, segment margin and 2027 utility award cadence matter more than headline revenue.

The preferred expression is quality-duration diversification rather than adding undifferentiated AI beta. A relative-value long PWR versus NVDA reduces exposure to a broad capex or rates shock while retaining exposure to the data-center power bottleneck; reverse only if hyperscaler capex guidance accelerates materially while utility award timing weakens.

Key Decisions for Investors

• Initiate a 3-6 month long PWR / short NVDA dollar-neutral pair on a 5-8% relative pullback in PWR; target 10-15% relative outperformance as grid-related awards and backlog conversion gain attention. Stop if PWR's next two quarters show material margin erosion or backlog growth stalls, or if NVDA's major customers collectively lift AI capex guidance.
• Maintain NVDA exposure only through a defined-risk post-earnings structure: buy 3-4 month call spreads financed in part by selling farther out-of-the-money calls, rather than chasing spot. Require upside guidance and stable gross-margin outlook; exit if forward revenue expectations fail to rise after results, as multiple compression becomes the dominant risk.
• Build a 6-18 month basket of PWR, ETN and HUBB on rate-driven weakness, with PWR as the project-execution component and ETN/HUBB as equipment-content exposure. Size below a pure AI allocation because transmission permitting and utility procurement can shift quarterly revenue timing.
• Monitor 10-year real yields and utility/interconnection award announcements over the next 1-3 months. A sustained real-yield rise without estimate revisions is a de-risking signal for both names; a drop in yields combined with raised earnings estimates is the cleaner catalyst to add.