Goldman expects U.S. equity market to continue making new highs By Investing.com

Goldman Sachs set a year-end S&P 500 target of 7,600, implying about 7% upside, citing 12% index gains since March 30, continued earnings growth, and AI investment spending. The bank expects 2026 and 2027 EPS growth of 12% and 10%, respectively, with AI contributing roughly 40% of S&P 500 EPS growth this year. It also sees beneficiaries of AI capex, especially power-infrastructure-linked names, as the clearest opportunity, while warning that market breadth remains unusually narrow.

Analysis

The market takeaway is not simply that AI spend is still alive; it is that capex is becoming a defensive moat for the largest platforms. That favors the handful of firms with balance-sheet capacity and in-house demand vectors, while pressuring smaller model vendors, enterprise software names without proprietary distribution, and any supplier that cannot prove it is directly embedded in hyperscaler roadmaps. The second-order effect is that the AI trade is narrowing from “software beta” into “utility-like infrastructure beta” where power, networking, and land/construction bottlenecks can re-rate faster than the application layer.

The bigger risk is that consensus is extrapolating a capex supercycle without distinguishing between spend and monetization. If cloud and model revenues do not visibly accelerate over the next 1-2 quarters, investors will start discounting efficiency rather than growth, which would compress the multiples of the highest-duration AI beneficiaries first. That matters because the market’s breadth is already fragile; a minor miss from one of the capex leaders could trigger a fast unwind in crowded passive and growth positioning over days, even if the long-term AI thesis remains intact.

For Amazon, the partnership deepens the strategic narrative around compute demand and bargaining power versus peers, but the cleaner expression is still via the ecosystem that enables the buildout: power, semis, and data-center supply chain names with backlog visibility. The contrarian read is that the best risk-adjusted upside may now sit outside the obvious mega-cap AI winners, because expectations there are elevated while the infrastructure beneficiaries still trade on underappreciated multi-year demand elasticity. If the cycle persists, the trade should extend to grid upgrades, thermal management, and electrical equipment before it meaningfully broadens into the rest of software.

Key Decisions for Investors

• Long AMZN vs short a basket of AI-adjacent software with weaker proprietary distribution for 3-6 months; express through a pair trade focused on relative multiple expansion versus AI monetization skepticism.
• Buy selective infrastructure beneficiaries tied to power and data centers for 6-12 months; risk/reward is favorable because backlog and pricing power should re-rate before earnings fully catch up.
• Use call spreads on AMZN into the next hyperscaler capex print; upside is tied to partner validation, while downside is cushioned if the market rotates away from broader AI enthusiasm.
• Fade overowned high-duration software names on any post-earnings strength; if capex remains high but revenue conversion lags, these names are vulnerable to a 10-15% multiple compression move.
• Keep a tactical hedge via short index breadth exposure over the next 1-2 months; narrow leadership makes the market vulnerable to a fast drawdown if one or two capex leaders disappoint.