Earnings Season Is Two-Thirds Over. Here's How It's Going and What It Means for the Market.

S&P 500 earnings season is running ahead of expectations, with 84% of reporting companies beating EPS estimates versus a 10-year average of 76%, and aggregate earnings growth tracking around 27% year over year. Alphabet, Amazon, Meta, and Nvidia were major contributors to the upside, helped by one-time gains at Alphabet and Amazon, a tax benefit at Meta, and Nvidia's $68.1B fiscal Q4 revenue with EPS of $1.62, about 6% above estimates. The piece reinforces that megacap tech and AI-linked names remain the main drivers of the market rally.

Analysis

The market is effectively being funded by a narrow set of balance sheets that can still convert AI capex into near-term earnings power, but the composition of the beat matters more than the beat itself. The current quarter is being flattered by one-off marks, tax items, and investment gains, which means the index-level EPS print is overstating the durability of operating leverage. That creates a subtle risk: passive inflows may keep rewarding the same leaders, while forward estimates for the rest of the index remain too sticky and vulnerable to revisions once the accounting noise fades.

Second-order beneficiaries are not just the obvious megacaps, but the compute, networking, power, and semiconductor supply chain tied to continued AI infrastructure spend. If these firms sustain capex, suppliers with exposure to racks, photonics, interconnects, memory, and power management should see a longer runway than the headline earnings data implies. The more important signal is that the largest tech platforms are still willing to spend aggressively despite already elevated margins, which keeps the AI arms race intact and raises the bar for smaller software and internet peers that lack distribution and balance-sheet flexibility.

The risk window is 1-2 months, not 1-2 days: as companies report the next quarter, the market will begin separating recurring EPS growth from transitory items. If AI monetization or ad/recommerce trends do not offset the normalization of these one-time boosts, the narrative can flip from “earnings breadth improving” to “earnings quality deteriorating.” That would likely hit high-multiple software first, then spill into the broader megacap basket if guidance disappoints.

Consensus is probably underpricing how concentrated the rally is becoming. The index can continue higher with only a handful of names working, but that usually makes the tape more fragile, not less, because breadth deteriorates while index-level momentum masks it. The better contrarian trade is to fade broad-market complacency while staying long the AI infrastructure winners that are actually monetizing the spending cycle.

Key Decisions for Investors

• Long NVDA on any post-earnings pullback over the next 2-6 weeks; treat it as a quality growth long with asymmetric upside if capex remains structurally elevated, but keep a tight stop if guidance implies a second-half digestion in orders.
• Pair trade: long AI infrastructure basket (NVDA/selected semicap equipment/networking names) vs short unprofitable software over the next 1-3 months; thesis is that spending is still real, but monetization is accruing to picks-and-shovels, not application-layer names with stretched multiples.
• Buy 1-3 month put spreads on an equal-weight S&P proxy or high-beta software ETF; this is a breadth hedge against index-level earnings concentration unwinding as one-time boosts roll off.
• Maintain a tactical long in META and AMZN only if ad/recommerce and cloud commentary confirm operating leverage beyond accounting items; otherwise trim into strength, as these names are the most exposed to sentiment reversal when growth normalizes.
• Avoid chasing GOOGL strength blindly; use call spreads instead of outright longs for the next earnings cycle, since the market may be overestimating how much of current EPS momentum is recurring versus mark-to-market noise.