NextFin News - Alphabet reported the largest quarterly profit in its history in the second quarter of 2026, a $112.1 billion bottom line that grew 298% from a year earlier. Amazon's net income more than tripled to $62.6 billion. Microsoft's earnings swelled on stakes in the very AI startups those companies are racing to dominate. The common thread is not better ads, faster cloud growth, or cheaper chips. It is a roughly $160 billion mark-to-market boost from Big Tech's stakes in other AI companies — gains booked on paper, with no cash changing hands, and no operator actually managing the line item that drove the quarter.
The gains are legitimate under U.S. accounting rules, and they are real on the income statement. But they sit below operating profit, and they expose a circularity at the heart of the AI boom: the companies pushing the market higher are increasingly dependent on each other's valuations. When Alphabet, Amazon and Microsoft revalue their holdings in Anthropic, OpenAI and SpaceX, they are not cashing out a winner. They are marking up an asset whose value was set by the next round of funding — often funded by the same hyperscalers whose earnings the markup then flatters.
The Numbers Behind the Headline Profit Surge
Alphabet's second-quarter filing shows a net gain of $98.0 billion in "other income," which the company said was "primarily the result of net unrealized gains on our equity securities." That single line added $77.1 billion to net income after taxes and accounted for $6.26 of the company's $9.11 in earnings per share. Without it, Alphabet's profit growth collapses from 298% to roughly 23%.
The two biggest drivers are named in the filings. Alphabet's stake in SpaceX was carried at a fair value of $94.0 billion at June 30, 2026, an effective 4.9% ownership after SpaceX's June 12 initial public offering at a $1.77 trillion valuation. Its Anthropic holding rode a private-market revaluation from $350 billion to $965 billion in the same period. Amazon, for its part, recorded $53.4 billion of non-operating pre-tax other income "primarily from our investments in Anthropic." Microsoft booked a $3.2 billion gain from Anthropic and a $480 million gain from OpenAI.
Add them up — Alphabet's roughly $99 billion in equity-securities gains, Amazon's $53.4 billion and Microsoft's $3.7 billion — and the hyperscalers absorbed about $160 billion of paper profits in a single quarter. That is larger than the entire U.S. IPO market raised in 2025, when proceeds totaled roughly $45 billion. It is also roughly a third of what the four biggest hyperscalers plan to spend on AI infrastructure in 2026, a combined outlay that has been estimated at between $700 billion and $760 billion. In effect, the AI build-out is being partly financed by the appreciation of the assets bought with the cash it consumes.
The scale of the distortion is visible in the gap between the income statement and the balance sheet. Amazon's $53.4 billion pre-tax gain was nearly twice its $27.5 billion of operating income for the quarter. Alphabet's $98 billion gain was more than twice its $40.8 billion of operating income. Microsoft's $3.7 billion of disclosed investment gains sat alongside $40.6 billion of operating income. At all three companies, the investment line is no longer a rounding error in the footnotes. It is the largest single driver of the quarter.
The Accounting Mechanism: Why a Private Funding Round Becomes Your Profit
The mechanism is straightforward and often misunderstood. Under U.S. GAAP, equity stakes in private companies that do not have a readily determinable fair value are carried under the "measurement alternative." They are revalued only when an observable transaction occurs — a new funding round, a secondary sale, or an IPO. When Anthropic raised capital at a higher valuation, or when SpaceX listed, the holders of those stakes were required to recognize the change in value in net income for the quarter in which the observable event occurred.
That is why a funding round thousands of miles from Mountain View or Seattle flows straight into Alphabet's and Amazon's earnings per share. It is also why the gains are non-cash. On the cash-flow statement, the same revaluations are reversed out when reconciling net income to cash from operations. No dollar entered the treasury. No customer bought anything. A valuation multiple moved, and accounting rules translated that movement into profit.
This creates a distinctive distortion: headline net income and EPS growth can diverge sharply from operating performance. Amazon's reported earnings jumped more than 240% year over year; strip out the Anthropic gain and the growth rate is closer to 17%. Alphabet's near-triple-digit profit growth falls to about 23% on the same adjustment. Operating income — the line operators actually manage — told a different story: Amazon's operating profit rose 43% to $27.5 billion, and AWS revenue accelerated 37%, its fastest growth in 18 quarters.
The asymmetry cuts both ways, and that is the part investors have not fully priced. The same rule that booked $160 billion of gains this quarter will book losses when funding rounds price flat or down. SpaceX shares fell roughly 50% from their IPO high and about a third since the second quarter ended, setting up what analysts expect to be a mark-to-market reversal for Alphabet in the September quarter. A successful Anthropic IPO, expected around October, could offset it — but the offset would be another revaluation event, not a change in the underlying dynamic.
The Aggregate Distortion: How a Few Stakes Moved the Entire S&P 500
The distortion is large enough to move the aggregate earnings picture for the broad market. Tajinder Dhillon, head of earnings and equity research at LSEG, calculated that S&P 500 earnings growth in the quarter was about 48% year over year with the gains included. Excluding the private-company markups at just Amazon and Alphabet, growth drops to roughly 29% — much closer to the roughly 24% analysts had been forecasting.
In other words, close to 20 percentage points of the index's headline earnings growth came from a handful of revaluations at three companies. That is a concentration risk disguised as broad-based earnings strength. Investors buying the S&P 500 on 48% earnings growth are, in effect, buying exposure to Anthropic's and SpaceX's valuation multiples — multiples that can move as violently in one direction as the other.
The concentration is even tighter than the headline suggests. Three names — Anthropic, OpenAI and SpaceX — account for virtually all of the $160 billion. If Anthropic's next funding round prices below the $965 billion mark that drove this quarter's gains, Amazon's and Alphabet's "other income" lines turn negative in the same stroke. There is no diversification benefit inside the distortion, because the same funding cycle moves all three assets together.
The Circularity: What's Funding AI Is Increasingly More AI
The deeper issue is the loop the gains reveal. Alphabet and Amazon invest billions in Anthropic. Anthropic spends billions of the capital it raises on Google Cloud and AWS compute. That spending shows up as cloud revenue growth for the backers, helping justify the next funding round at a higher valuation. The higher valuation then flows back into the backers' income statements as profit.
"It's circular," said Matt King, founder of Satori Insights. "What's funding A.I. is now increasingly more A.I."
The loop is not a fraud — it is a feature of how capital-intensive frontier technology scales. But it means a portion of Big Tech's reported profitability is self-referential. The same dollars circulate through the ecosystem, booked as revenue on one balance sheet and as an investment gain on another. Robert Willens, a tax and accounting consultant, put the point more sharply in an interview: "It's interesting that they're able to control or influence the value of one of their own assets."
The self-referential quality is visible in the numbers. Alphabet has committed up to $40 billion to Anthropic, a mix of cash and long-term cloud infrastructure commitments. Amazon's original $8 billion Anthropic investment is now worth more than $70 billion on its books. Those stakes are not passive. The companies that hold them also sell the compute the portfolio companies consume, and they book the consumption as cloud revenue. The valuation of the asset and the revenue of the holder are tied to the same underlying spending wave. When that wave slows, both the investment line and the cloud line feel it — from opposite directions.
Cyclical or Structural: A Call on What Reverts
The central question for investors is whether this is a cyclical accounting wave that will revert, or a structural feature of the AI era that is here to stay. The answer is both, and separating them matters.
The cyclical leg is clear. Mark-to-market gains on private stakes are inherently mean-reverting because they are tied to funding cycles. When capital is cheap and AI enthusiasm is high, rounds price up and holders book gains. When the cycle turns, the same rule books losses. The evidence is already visible: SpaceX shares fell roughly 50% from their IPO high and about a third since the second quarter ended, setting up what analysts expect to be a mark-to-market reversal for Alphabet in the September quarter. A successful Anthropic IPO, expected around October, could offset it — but the offset would be another revaluation event, not a change in the underlying dynamic.
The structural leg is equally real. Big Tech is no longer just a buyer of AI innovation; it is a permanent holder of a large, illiquid venture portfolio embedded in its capital structure. Alphabet's $94 billion SpaceX position and its up-to-$40 billion commitment to Anthropic, Amazon's $8 billion investment now worth more than $70 billion, and Microsoft's stakes in OpenAI and Anthropic are not temporary trading positions. They are strategic assets that will produce volatile, valuation-driven earnings line items for as long as these companies remain privately held or newly public. That volatility is a permanent addition to Big Tech's earnings profile, not a one-time event.
The distinction matters because the cyclical leg cuts the headline growth rate back toward the mid-20s, while the structural leg means even the adjusted numbers will carry more volatility than the market has priced in. Investors treating 48% S&P 500 earnings growth as the new baseline are making a cyclical error. Investors assuming the adjusted 29% is a stable floor are underestimating the structural volatility now embedded in these balance sheets.
The Counter-Thesis: Strong Earnings With or Without the Marks
The strongest argument against the distortion narrative is that Big Tech's underlying earnings were strong even before the marks. Jeff Kilburg, founder and CEO of KKM Financial, called the investment gains "sprinkles" on top of an already strong earnings season and described the underlying profit jump as "jaw dropping" even excluding private assets. Amazon's operating income rose 43%, Google Cloud jumped 82% to $24.8 billion, Microsoft's Azure grew 43% and full-year Azure revenue surpassed $100 billion for the first time, and Alphabet's operating income climbed 30% to $40.8 billion. These are not numbers that need accounting assistance to look healthy.
The counter-thesis holds, with an important caveat. It is true that operating performance was robust, and that operating income is the cleaner boundary between business results and investment accounting. But the market does not trade on operating income alone. Headline EPS — the number most models, index funds and momentum strategies key off — was materially inflated. When the cyclical leg reverses, headline growth will fall faster than operating growth, and the market's reaction will be driven by the headline it actually priced. The "sprinkles" argument is correct about the operating business and wrong about the market's pricing anchor.
There is a second, subtler weakness in the counter-thesis. The investment gains and the operating growth are not independent. A meaningful share of the cloud acceleration that the counter-thesis points to as proof of organic strength is itself funded by the capital raised in the same funding rounds that generate the marks. Strip out the portfolio companies' compute spending and cloud growth is still strong, but not as strong as the headline suggests. The two stories — operating strength and investment markup — are not competing explanations. They are two accounting views of the same capital flow.
The falsifying signal is specific and observable: if Alphabet, Amazon and Microsoft each report flat or positive mark-to-market adjustments for two consecutive quarters while maintaining double-digit operating growth, the distortion thesis is wrong and the gains should be treated as a durable earnings component. If instead the September and December quarters show mark-to-market reversals that cut headline growth back toward the mid-20s while operating income holds, the thesis is confirmed.
What to Watch Next
Three signals will determine whether the $160 billion boost was a peak or a new normal.
Short term (the September and December quarters): Watch the "other income" line at all three companies. SpaceX's post-IPO decline points to a reversal for Alphabet; an Anthropic IPO, expected around October, could provide an offsetting mark. The net of those two events will set the tone for headline EPS. A second observable trigger: if Anthropic's IPO prices below $965 billion, the write-down risk extends beyond Alphabet and Amazon to every holder of AI venture stakes.
Medium term (operating cash flow): The gap between net income and operating cash flow will widen in quarters with large marks and narrow when they reverse. A sustained divergence — rising net income with flat operating cash flow — is the clearest warning that reported profitability is drifting from cash generation. Investors should track the reconciliation line that reverses unrealized gains; it is the most honest number in the filing.
Long term (structural volatility): As Anthropic, OpenAI and other private AI leaders list, the marks will crystallize into realized gains and the accounting noise will eventually subside. What remains is a permanent feature: Big Tech balance sheets carrying large, strategic, illiquid stakes whose valuations will swing with funding cycles. The IPO wave itself is a second-order risk. Goldman Sachs has estimated the 2026 AI-driven IPO pipeline — led by SpaceX, Anthropic and OpenAI — could generate roughly $160 billion in listings, quadrupling the prior year's market. That is a large absorption test for public markets, and the pricing of those debuts will set the next round of marks for the holders still waiting to list.
The base case is that headline earnings growth reverts toward the mid-20s once the marks normalize, while operating growth holds in the high teens to low 20s. The upside case is a successful AI IPO wave that keeps revaluations positive through 2027, extending the headline boost. The downside case is a funding-cycle pause that turns the marks negative and cuts headline growth well below operating growth for several quarters.
The $160 billion did not make Big Tech richer in cash. It made the market's favorite earnings metric richer in noise. The AI boom is funding itself, and until the loop closes with real revenue from real customers outside the circle, the cleanest read of Big Tech's earnings is the line the marks sit below — not the one they inflate.
Explore more exclusive insights at nextfin.ai.
