Mag-7 earnings are the first real “broadening” stress test—and the cap-weight is already doing the heavy lifting

The YTD gap between an equal-weight S&P 500 exposure and the traditional cap-weighted exposure is a live “regime meter.” When the cap-weight lags, it usually means the market is willing to underwrite faster or higher-quality earnings from a broader set of companies. When the gap starts narrowing during major earnings windows, it typically means the market is re-accepting that the leaders’ cash-flow narrative dominates the index again—even if that leadership comes with capex intensity.

To anchor the event-based framing, we focus on the two first Mag-7 earnings reactions referenced in the brief: Meta and Microsoft. From the earnings financials we pulled for this session, Meta delivered Q2 revenue of $60.801B but net income of $15.848B, while also carrying very large operating expense and R&D relative to revenue. For Microsoft, the latest quarterly income statement in this session shows Q4 revenue of $90.007B and net income of $35.766B, with operating income of $40.603B—evidence of a “beat-capable” operating model despite AI build-out that still requires ongoing investment.

An equal-weight portfolio is structurally less sensitive to the index’s largest names. In practice, that means the equal-weight/cap-weight divergence can persist even when mega-caps are volatile—unless investors decide that mega-cap earnings quality is the real determinant of index-level forward guidance. The cap-weighted S&P 500 effectively buys “investor consensus on the leaders’ earnings durability.” When AI capex rises, the first-order effect is not always revenue; it’s operating leverage, depreciation/amortization expectations, and the market’s tolerance for longer payback periods.

Even though the index test is “broadening vs. concentration,” the underlying mechanism is capex timing. AI infrastructure spending is layered: customers (hyperscalers and platforms) pull forward compute demand; semiconductor and networking suppliers ship hardware; and the customer P&L absorbs the build-out via depreciation, R&D, and data-center operating costs. That’s why an earnings miss with resilient revenue is often interpreted as “capex is not yet converting into the earnings multiple you paid for.” Conversely, a beat can reset expectations that capex is translating into monetizable utilization.

In this session’s Meta dataset, Q2 operating income is $18.775B against Q2 revenue of $60.801B while R&D runs at $21.656B—an unusually large expense envelope relative to revenue. That is compatible with markets treating AI investment as front-loaded. For Microsoft, the latest-quarter operating income is $40.603B on $90.007B of revenue with far smaller R&D in dollars in this dataset ($9.997B), consistent with a higher capacity to absorb AI build-out without immediate operating leverage deterioration.

Over 1–3 years, the regime question becomes structural: can the leaders’ AI capex translate into durable operating leverage and cash conversion? That’s where the equal-weight trade either compounds (if monetization broadens beyond mega-caps) or collapses (if cap-weight leaders demonstrate sustained earnings durability and smaller stocks fade).

The quiet version of “broadening” is easy to believe when leaders are stable. The hard test is when leaders print under AI capex pressure. In this session’s verified earnings data, Meta shows a profile consistent with cost intensity and compressed free cash flow in the latest quarter, while Microsoft shows strong operating income and net income capacity. The investor implication is straightforward: the equal-weight/cap-weight divergence is most fragile when mega-cap earnings quality is perceived to deteriorate faster than revenue growth.