AI Valuation Gains Are Skewing S&P 500 Earnings: Microsoft, Amazon, Alphabet Analysis

The Accounting Mechanism Behind the Distortion

Microsoft, Amazon, and Alphabet all hold equity stakes in private AI companies—OpenAI, Anthropic, and in Alphabet’s case, SpaceX. As those private valuations climb, accounting rules require these companies to mark their holdings to market and record the gains on their income statements, typically under other income.

This is not revenue from selling software, cloud services, or AI products. It is an unrealized gain—a number that reflects what a private company is theoretically worth today, not cash that has changed hands.

When Anthropic and OpenAI are each valued at just under $1 trillion in private markets, even a modest percentage stake translates into a very large line item on a quarterly earnings report.

Why This Matters for the Broader Market

The Magnificent Seven names account for roughly 35% of S&P 500 second-quarter revenue, according to available data. When a handful of companies at that scale book large non-operating gains simultaneously, the effect on aggregate index metrics is disproportionate.

The beat rate tells a similar story. Companies reported earnings approximately 7% above analyst expectations this quarter, compared to a long-term average of around 4.4%. Part of that outperformance traces directly to these investment gains, which most analysts exclude from their non-GAAP forecasts.

In other words, the upside surprises were partly a function of accounting treatment, not operational acceleration.

The Volatility Risk Works Both Ways

Mark-to-market gains are not permanent. SpaceX, for example, has declined roughly 50% from its post-IPO high. That means Alphabet is likely to record a significant reversal on its SpaceX position when it reports the following quarter—unless an Anthropic IPO, which has been filed confidentially with the SEC, produces an offsetting gain.

This introduces a new kind of earnings volatility for mega-cap tech: one driven not by product cycles or customer demand, but by the fluctuating valuations of private companies in which they hold stakes.

What to Watch Going Forward

• IPO timing matters. If Anthropic or OpenAI lists publicly within the next year as expected, the mark-to-market mechanism changes. Gains and losses become more visible and more frequent.
• Non-GAAP remains the cleaner signal. Analysts are already excluding these items from forecasts. Readers tracking AI tool adoption or platform growth should weight operational metrics—cloud revenue, API usage, enterprise contract growth—more heavily than headline earnings.
• Concentration risk is compounding. Big Tech is simultaneously the infrastructure layer for AI, the primary investor in frontier AI labs, and a dominant share of the S&P 500. That means valuation swings in private AI companies now have a direct transmission mechanism into broad market earnings metrics.

The underlying business performance across Microsoft, Amazon, and Alphabet appears genuinely strong. But the reported numbers this quarter are not a clean read on that strength. Knowing the difference is the starting point for any serious analysis of where AI investment is actually going.