The U.S. stock market looks expensive; the S&P 500 currently has a trailing P/E ratio around 27. I recently discussed the idea that market participants are overly optimistic about future E. But you could argue that today’s high P is justified by the robust growth in actual E over the past year. Surely these official quarterly financials are rock-solid evidence of an AI-powered profits boom? Here’s BlackRock:[1]
Wow. It’s the first word that comes to mind to describe Q1 2026 U.S. company earnings. S&P 500 earnings growth is looking set to reach 28% year over year (yoy), more than double the consensus estimate of 12% at the start of the reporting season.
A 28% growth rate would indeed be impressive if it were real. But arguably, it’s not. Generally accepted accounting principles (GAAP) currently require that earnings include a mark-to-market adjustment for unrealized gains on equity. Rising equity prices in both public and private markets are now materially impacting the reported earnings of public companies. According to calculations by Wang (2026), the unrealized gains reported by just three companies (Alphabet, Amazon, and Nvidia) added $69B to the reported earnings of the S&P 500 in Q1 2026.[2] According to him, if you remove this $69B, the trailing annual growth of S&P 500 quarterly earnings is not 28% but instead is around 16%.
For example, Amazon owns shares in Anthropic. Anthropic (currently a private company) completed a funding round in February 2026 in which its valuation approximately doubled, increasing the value of Amazon’s stake. This event greatly impacted Amazon’s reported Q1 2026 net income of $30.3B, because Amazon had $16.8B of pre-tax non-operating gains from its stake in Anthropic.
And it’s not over yet. We can expect further blowout earnings for Amazon in the coming quarter, because Anthropic completed another funding round in May 2026, this time more than doubling in valuation. Alphabet may also receive a large earnings boost from its SpaceX stake, thanks to the SpaceX IPO of June 2026.
Earnings come in many flavors. First, there are GAAP earnings. Since the adoption of an accounting rule called ASU 2016-01, GAAP earnings for U.S. firms must include unrealized capital gains and losses from certain equity holdings. Second, there are “street earnings” created by analysts, which attempt to measure underlying earnings by removing non-recurring items. And third, there are sometimes earnings metrics created by the reporting companies themselves, such as the “Bitcoin per share” measure used by Strategy or the infamous “Community Adjusted EBITDA” devised by WeWork.
Street earnings became prevalent in the 1990s according to Bradshaw and Sloan (2002). The academic accounting literature has debated the differences between street and GAAP earnings. One view is that street earnings are systematically too optimistic (“earnings before bad stuff”). An alternative view is that street earnings are a better measure of long-term profits and that rational investors should just ignore GAAP earnings.
I’m not saying that ASU 2016-01 is obviously a bad idea. It was implemented following the GFC with the laudable goal of increasing transparency and preventing firms from obscuring the risk exposures arising from their holdings. It’s factually true that Amazon profited in Q1 2026 thanks to its ownership in Anthropic, and that information needs to be disclosed somewhere. In a world of rational and attentive investors, no one would be confused by the inclusion of unrealized gains in GAAP earnings, because these gains are fully disclosed in SEC filings.
The problem is that we live in a world where inattentive investors are likely to get confused. Warren Buffett said that the rule:[3]
will severely distort Berkshire’s net income figures and very often mislead commentators and investors …. That requirement will produce some truly wild and capricious swings in our GAAP bottom-line … media reports sometimes highlight figures that unnecessarily frighten or encourage many readers or viewers … I expect considerable confusion among shareholders for whom accounting is a foreign language.
Are investors actually confused by unrealized gains? Amornsiripanitch et al. (2025) argue that they are indeed confused:
When firms’ summary performance measure includes changes in unrealized gains and losses (UGL) from financial asset holdings, investor inattention causes the firms’ stock returns to overreact to changes in UGL.
And even if investors are not confused at the firm level, they might well be confused at the aggregate level. The bubble scenario is that this confusion impacts the aggregate stock market, with investors seeing rising E as a justification for rising P, which then mechanically feeds into rising E. Feedback loops, self-fulfilling prophecies, and “naturally occurring Ponzi schemes” have always been a feature of stock market bubbles.
This particular feedback loop was not a part of the bubbles of 1929 or 1999, because ASU 2016-01 had not then been adopted. However, it may have played a small role in the COVID bubble. For example, in 2021 Shopify had a stake in Affirm, while Amazon had a stake in Rivian. Both Affirm and Rivian had IPOs in 2021, resulting in increased valuations that goosed the reported earnings of Shopify and Amazon. When the market declined in 2022, the reported earnings of Shopify and Amazon were negatively impacted. He who lives by ASU 2016-01 dies by ASU 2016-01.
I previously discussed circular AI deals and said they’d be concerning if they resulted in “distorted information about fundamentals such as earnings” but that “I’m not seeing these features.” Okay, now I’m seeing these features: circular deals do have the potential to confuse investors about earnings. ASU 2016-01, created with the best of intentions after the collapse of the housing bubble, may have the unintended consequence of helping create a new bubble.
Endnotes
[1] “The equity outlook after more ‘magnificent’ earnings,” BlackRock, May 26, 2026.
[2] References to this and other companies should not be interpreted as recommendations to buy or sell specific securities. Acadian and/or the author of this post may hold positions in one or more securities associated with these companies.
[3] Berkshire Hathaway Shareholder Letter, February 2017.
References
Amornsiripanitch, Natee, Brandon Goldstein, Zeqiong Huang, David Kwon, and Jinjie Lin. "Net income aggregation, investor inattention, and portfolio holding decisions: Evidence from the insurance industry." The Review of Corporate Finance Studies (2025).
Bradshaw, Mark T., and Richard G. Sloan. "GAAP versus the street: An empirical assessment of two alternative definitions of earnings." Journal of Accounting Research 40, no. 1 (2002): 41-66.
Wang, Baolian. “The $69 Billion Mirage: How an Accounting Rule Inflated S&P 500's Q1 Earnings by 12%.” Substack, June 2, 2026.
