Big Tech just closed out its second-quarter 2026 earnings season, and the results looked, on paper, like a triumph. Microsoft, Amazon, and Alphabet all posted numbers that blew past Wall Street‘s estimates, and headlines celebrated growth rates that haven’t been seen in years.
Dig one layer deeper, and a strange pattern connects all three reports. A meaningful chunk of that growth has nothing to do with ads, cloud contracts, or software subscriptions. It comes from something these companies don’t even sell.
Paper gains from Anthropic and OpenAI are padding profits
S&P 500 earnings grew more than 48% year over year last quarter. Exclude the investment gains tied to private AI stakes, and that growth falls to roughly 29%, according to recent analysis, bringing underlying earnings growth much closer to expectations.
Gil Luria, managing director and head of technology research at D.A. Davidson, put it plainly. “The headline earnings numbers were very much inflated by equity gains in OpenAI, Anthropic and SpaceX,” he told CNBC, adding that these swings tend to even out over time, which is why most analysts strip them out of non-GAAP estimates.
Related: Jeff Bezos just named Amazon’s next big pillar
That stripping-out hasn’t stopped the near-term effect. Companies this quarter reported earnings 7% above expectations, well above the long-term average beat of 4.4%. A meaningful share of that upside surprise traces directly back to markups on venture stakes, rather than operating performance.
The effect matters more than a typical accounting footnote because of how much these companies already dominate the index. The so-called Magnificent 7 accounted for roughly 35% of S&P 500 second-quarter revenue, Armstrong Fleming & Moore noted, a share that has held near a third of the large-cap index for more than a year.
When a handful of stocks carry that much weight, their accounting choices move the whole market’s narrative.
Amazon, Alphabet, and Microsoft each got a boost
Amazon’s case is the starkest. The company booked a $53.4 billion pre-tax gain “primarily from” its stake in Anthropic, helping push reported net income to $62.6 billion, up more than 243% from a year earlier.
Without that gain, underlying operations-driven net income was closer to $20.8 billion, up roughly 14% year over year. Amazon has invested $13 billion in Anthropic with the potential for up to $20 billion more, and has also been building exposure to OpenAI through a separate cloud infrastructure arrangement.
Alphabet’s quarter ran through two names most investors weren’t fully tracking. Google owned about a 6% stake in SpaceX going into Q2. SpaceX went public in June at a $1.77 trillion valuation.
Alphabet also holds roughly 14% of Anthropic, with $13.3 billion already invested and commitments of up to $30 billion more. When Anthropic’s valuation tripled to $965 billion during the quarter, both positions moved sharply.
Together, they produced a $98 billion “other income” gain, boosting Alphabet’s net income to $112.1 billion, up 298% year over year. The actual advertising and cloud businesses grew 24%, according to Fortune.
Microsoft saw a smaller but still meaningful lift. A $3.2 billion gain, mostly from its Anthropic investment in fiscal Q4 earnings, added roughly $0.27 per share to earnings growth. The company separately accounts for the impacts of its OpenAI investment in its non-GAAP figures, underscoring how both AI relationships can affect its reported financial results.
This quarter’s results are not a new phenomenon. Goldman Sachs flagged a comparable dynamic back in the first quarter, when S&P 500 earnings growth neared 25% on paper and fell to roughly 17% after Amazon and Alphabet’s investment gains were excluded, as TheStreet reported.

Drago/Getty Images
The gains are real on paper, less certain in cash
These figures typically land in a catch-all line called “other income,” which makes them easy to overlook and inconsistent to compare across companies. That reporting gap is part of why the swings caught many analysts off guard this quarter.
There’s also a circularity worth watching. Alphabet’s $98 billion in other income came largely from gains on its equity investments in private AI companies. Anthropic, whose valuation tripled from $380 billion to $965 billion during Q2 after a $65 billion fundraising round, has committed to purchasing at least five gigawatts of computing capacity from Google Cloud.
Alphabet’s capital helps lift Anthropic’s valuation, while Anthropic’s spending helps lift Alphabet’s cloud numbers in return.
Reuters Breakingviews summed up the dynamic bluntly, noting that SpaceX and Anthropic together account for half of the roughly $280 billion rise in S&P 500 net earnings this quarter, even though both companies are still losing money.
At the same time, Big Tech’s massive infrastructure spending is incinerating cash flow, making the logic behind celebrating the paper gains harder to defend.
The reversal risk is also real. Alphabet cannot currently sell a single share of its $94 billion SpaceX stake, which is subject to short-term sale restrictions per its Q2 10-Q filing. A decline in SpaceX’s market price or Anthropic’s private valuation could reverse some of those earnings gains in a future quarter, according to Benzinga.
What this means for the numbers ahead
Both Anthropic and OpenAI have filed confidentially with the SEC and are moving toward potential public listings, although neither has committed to a firm debut date.
Anthropic is currently valued at $965 billion. OpenAI’s most recent private valuation was roughly $300 billion. Together with SpaceX’s $1.77 trillion public market cap, the current AI investment pipeline that has been flowing through Big Tech earnings represents trillions in paper wealth that will eventually need to be reconciled with actual revenue and cash flow.
Anthropic alone has grown its annualized revenue from roughly $9 billion to $47 billion by early 2026, a pace that makes its private valuation swings unusually large and significant for its corporate backers.
Investors reading next quarter’s headline numbers would do well to check the footnotes first, the ones most analysts already strip out before publishing their forecasts.
The underlying businesses are still growing. The question is whether the market keeps rewarding paper gains at the same pace once these labs actually start trading.