AI Paper Wealth Surges in US Earnings Season: OpenAI and Anthropic Equity Gains Inflate Profits, Wall Street Turns to Scrutinizing Cash Flow

The Q2 2026 US earnings season is wrapping up with jaw-dropping profit growth, but Wall Street's excitement is mixed with unprecedented caution. On the surface, S&P 500 earnings surged nearly 50% year over year, as tech giants delivered stellar results. Looking closer, however, Microsoft, Amazon, and Alphabet—Google's parent—have recorded massive paper gains on their equity stakes in OpenAI, Anthropic, and even SpaceX. This "paper wealth," which generates no real cash flow, is quietly reshaping how the market prices the AI trade: AI unicorns have yet to list on exchanges, yet their valuation gains have already "gone public" inside corporate income statements.

Earnings Surge Nearly 50%, but About a Fifth Is "Book Water"

According to the latest FactSet tally, S&P 500 earnings rose about 47.4% year over year in Q2 2026. By that number alone, this is one of the strongest earnings seasons in recent years. But strip out Alphabet and Amazon, and growth drops immediately to about 28.8%—much closer to Wall Street analysts' prior baseline expectation of roughly 24%.

Goldman Sachs, a Wall Street heavyweight, offers a more direct estimate: S&P 500 headline EPS rose about 45% year over year in Q2, with 19 percentage points coming from Alphabet and Amazon's unrealized equity investment gains. Excluding those, actual earnings growth is about 26%. LSEG data also confirms this—after removing investment gains from unlisted AI companies, earnings growth becomes notably milder, falling to about 29%.

In other words, a considerable portion of this season's "excess" earnings growth did not come from selling software, ads, or cloud services, but from mark-to-market revaluation of tech giants' stakes in AI startups. Because the magnitude of earnings beats has been inflated as a result (this quarter averaged 7% above expectations, versus a long-term average of 4.4%), market skepticism over the "quality" of this earnings season has grown accordingly.

What Lurks Inside the Big Three's Income Statements?

Amazon: $53.4 Billion in "Non-Operating" Gains

Amazon recorded roughly $53.4 billion in non-operating pre-tax gains this quarter, saying "a portion" indeed came from its investment in Anthropic. Thanks to that, earnings rose more than 240% year over year; but excluding investment gains, growth is closer to 17%. Meanwhile, Amazon's quarterly sales climbed about 20%, operating profit rose about 43%, and AWS-related revenue surged 37%—core business remains strong, just far less "explosive" than the headline numbers suggest.

Alphabet: $98 Billion in "Other Income"

Google parent Alphabet recorded about $98 billion in "other income" this quarter, driven mainly by unrealized appreciation on its equity securities, including Anthropic and SpaceX (a roughly 5% stake), which went public in June. Net profit growth consequently soared nearly 300% year over year; excluding the SpaceX and Anthropic gains, growth is closer to 23%. Even without revaluation gains, operating profit still rose 30%, operating margin climbed to 34%, and cloud computing continued to expand rapidly.

Microsoft: Relatively Moderate Contribution

Microsoft's boost from Anthropic equity gains has been relatively limited—earnings growth was pushed up about 10 percentage points. The company confirmed a $3.2 billion net gain from Anthropic and $480 million from OpenAI. Excluding this impact, net income and EPS still grew 22% and 23% respectively, while cloud revenue rose 27%.

It bears emphasizing that these gains are fully compliant with accounting rules, classified as unrealized fair value changes under "other income." But on the indirect method cash flow statement, such paper gains are typically deducted from net income—the company receives no cash from them, yet the income statement looks far richer.

Market Pricing Logic Shifts: From "Belief Phase" to "Validation Phase"

This trend matters because the seven big tech giants carry enormous weight in the S&P 500—according to LSEG, they account for about 35% of S&P 500 Q2 revenue and roughly a third of the large-cap index's weight over the past year. More importantly, earnings growth is highly concentrated: the top ten companies contributed nearly 80% of EPS growth. When just a few companies' "paper gains" can swing the entire index's earnings picture, the market's pricing logic is undergoing a profound shift.

The divergence in stock reactions this season is the best illustration. Apple and Meta both delivered revenue growth, yet their shares tumbled: Apple fell over 9% post-earnings, losing more than $350 billion in market value last week; Meta dropped over 9%, shedding about $85 billion. Microsoft and Amazon, by contrast, were rewarded—Microsoft added over $600 billion in market cap last week, and Amazon joined the $3 trillion club.

Jefferies experts note that the market is now reacting negatively to rising capital expenditures, a warning sign for companies pouring heavily into AI infrastructure. As the AI narrative enters late 2026, the "unconditional faith" phase is ending. Investors are beginning to scrutinize capital expenditure efficiency, margin resilience, and the certainty of AI monetization paths. Companies with sustained free cash flow and clear AI monetization are more likely to command a premium; those requiring continuous heavy investment with slower payoff—even with record revenue—are more likely to see their stocks fall after earnings.

"Paper Wealth" Cuts Both Ways: The SpaceX Lesson

What deserves even more caution is the two-way nature of these equity gains. SpaceX shares have fallen about 50% from their peak since its June IPO. D.A. Davidson managing director Gil Luria warns that at SpaceX's current trading price, Alphabet's mark-to-market gains are likely to reverse sharply when it reports September-quarter results; if Anthropic successfully goes public in September, it could partially offset that hit, but it is still too early to tell.

Both Anthropic and OpenAI have confidentially filed IPO paperwork with the SEC and are expected to hit public markets within the next year. Until then, the deeply intertwined loop of "capital investment—cloud orders—valuation rises—book gains" between tech giants and AI unicorns means that every swing in private valuations will be transmitted directly into giants' income statements, creating a procyclical amplification effect.

Real Growth Still Exists: AI Monetization Paths Are Clearing

Of course, this earnings season is not "false prosperity." Excluding one-time equity gains, cloud and AI-related revenue still grew strongly: Amazon's cloud business grew 37%, Microsoft's cloud grew 27%, Alphabet's operating margin rose to 34%, and Microsoft 365 Copilot paid users surpassed 30 million. These figures show that AI revenue generation is forming a real business loop through cloud computing, advertising efficiency gains, and enterprise software.

Goldman Sachs expects capital spending by major AI providers to rise from about $755 billion in 2026 to about $920 billion in 2027, and tech giants may also issue around $400 billion in investment-grade bonds in 2027. That means the cash burn continues, and the market's patience with AI return on investment is being repeatedly tested.

How Investors Should Respond: Cash Flow Is the Touchstone

For investors, this earnings season offers several lessons. First, do not judge solely by EPS—strip out equity revaluation gains when assessing earnings quality, and focus on operating profit, operating cash flow, and free cash flow. Second, watch the procyclical risk of "paper wealth": any pullback in private valuations will hit giants' income statements directly. Third, the earnings season isn't over—Nvidia is expected to report on August 26, and AMD and other chip giants have already reported. Capital expenditure and backlog data across the AI compute supply chain will be a key variable for the second half.

For investors in Southeast Asia, US equity allocation likewise requires "filtering the real from the fake": rather than chasing stories of simply beating EPS, focus on high-quality names with clear AI monetization and solid free cash flow. ETF tools tracking indices such as the Nasdaq 100 can also help diversify single-stock concentration risk. After all, when the tide goes out, what truly determines valuation is never book numbers, but the real business that customers are willing to pay for.