Damodaran argues AI capex is distorting reported earnings and cash flows.
Aswath Damodaran wrote on 8 October 2026 that the trillions of dollars going into AI capital expenditure (capex) are showing up across the financial statements of US companies. Aggregate net income at US publicly traded companies rose from $576 billion to $904 billion between the second quarter of 2025 and the second quarter of 2026, a growth rate of 57%, he writes, while the cash returned to shareholders grew far more slowly.
He set out the analysis in a post on his blog, Musings on Markets, titled "Stock Prices, Earnings and Cashflows: The AI Effect plays through!" Sharing it on X, he said he looks at how the spending has percolated through financial statements, "distorting book capital, earnings and cash flows." The figures below are his, for all US publicly traded companies.
**Who Spends and Who Books the Revenue**
Damodaran splits the market in two. The builders of the architecture are the hyperscalers, which he names as Meta, Alphabet, Amazon and Microsoft. Their spending, he writes, will cause lower earnings "at least until the capex starts paying off," as well as a hit to their free cash flows.
The same money becomes revenue for the suppliers. He lists chipmakers, with Nvidia and TSMC leading the way, network equipment makers, power plant builders and data center real estate developers. He notes gray zones: Amazon spends on AI capex but also benefits from its cloud business, and Nvidia, while selling the chips, is also investing directly or indirectly in data centers.
Capex in the second quarter of 2026 was up $133.4 billion from the second quarter of 2025, an increase of almost 36%, he writes. He says the top ten hyperscalers account for more than $2 trillion of AI capex, and that technology, communication services and consumer discretionary each grew capex by more than 50% in the quarter-to-quarter comparison.
*Business explainer: Capex and depreciation.* As we read it, Damodaran's point is about where AI spending lands in the accounts: as cost and lower free cash flow at the builders, and as revenue and earnings at the suppliers. The gap he draws attention to is between earnings, up 57% in the second quarter on his figures, and dividends and buybacks, up about 5% and about 9%.
**Bigger Balance Sheets**
Accounting puts capex on the balance sheet as assets and raises the book value of whatever funded it, Damodaran explains. Across all US stocks, book equity increased almost 13% between the second quarter of 2025 and the second quarter of 2026, and total debt is up almost 8%, he writes.
In dollar terms, he puts the increase at $1.8 trillion in book equity and $1.9 trillion in book debt over the same period. In technology, book equity rose almost 30% and total debt about 18.8%, by his figures.
**One Company's Capex Is Another's Earnings**
The spending lifts earnings for a simple reason, Damodaran argues: money spent on capex by a company building AI architecture "will become revenues (and earnings) for other companies that supply the building blocks for the architecture." He calls Nvidia the biggest beneficiary of the AI capex boom so far.
There should be higher amortization at the builders, he writes, but "the longer amortization schedules being used by many of them is reducing the current hit to earnings at these companies."
His figures: aggregate earnings rose from $511 billion to $692 billion in the first quarter of 2026, a growth rate of 35%, and from $576 billion to $904 billion in the second quarter, a growth rate of 57%. More than half of all companies reported declines in net income, he notes. Technology, financials and communication services saw the biggest increases, while health care, utilities and real estate lagged.
*Chart note: Changes as stated by Aswath Damodaran, for all US publicly traded companies. Earnings, capex and book values compare a 2026 quarter with the same quarter of 2025 (Q1 for one earnings bar, Q2 for the rest). Dividends and buybacks compare the last twelve months with calendar 2025, so the bars are not one series.*
**Cash Returned Has Not Kept Up**
Damodaran writes that the "higher earnings across firms will be partially or even fully offset by capital expenditures across firms," so free cash flows grow more slowly than earnings, and those free cash flows fund dividends and buybacks.
Dividends in the last twelve months are up about $43.3 billion, or about 5%, from calendar 2025, and stock buybacks are up about $106.7 billion, or about 9%, he writes. Those step-ups "clearly have not kept up with the earnings growth in 2026."
For much of the last two decades, he says, US companies returned 80% or more of their earnings to shareholders. In his words, "in the last twelve months leading into 2026, the companies in the S&P 500 returned 63% of their earnings to shareholders," which he calls "a low not seen since 2004."
**The Market Backdrop**
The ten-year Treasury rate rose from 4.75% to 5.29% during September 2026, Damodaran writes, a move that would put it in the top 10% of 770 monthly rate changes between 1962 and 2026. The market still added $2.5 trillion in market capitalization in September, and almost $1.5 trillion of that came from technology, he says.
His implied equity risk premium at the start of October 2026 was 3.70% over the ten-year Treasury rate, for an expected return on equities of 8.99%. He writes that the premium dropped below 4% for the first time this year.
**If AI Disappoints**
In his best case, AI investors earn returns on their capex above their cost of capital, lenders are made whole, and the companies emerge "more capital intensive than they used to be." In his worst case, accountants write off large portions of the AI capex, and companies that leaned on debt face defaults and distress.
He expects that "the market will lead in this process and accounting will follow." In his words, "waiting to act until accountants write off AI capex to sell your AI company implies that you waited too long."
He also discloses his own position. He holds five of the Magnificent Seven, all except Tesla and Nvidia, and says new money added to his portfolio in the last year or two has gone mostly into cash.
**The Structural Read**
On his numbers, aggregate net income rose 35% and 57% in the first two quarters of 2026, while dividends rose about 5% and buybacks about 9% over the last twelve months. The periods differ, so we read the comparison as directional rather than one-for-one.
His balance-sheet figures show how the spending was funded: book equity up $1.8 trillion and book debt up $1.9 trillion in a year, and in technology, total debt up about 18.8%.
His own caveat is that joint ventures and off-balance-sheet vehicles sit outside these numbers, and that private AI companies are not in his data at all. On his account, the aggregates are a floor on the effect, not the full picture.
*Aswath Damodaran, Musings on Markets (October 2026)*
"In effect, the only point on which you have a consensus on is that companies have collectively invested a huge amount in AI capex."
**Three Implications**
**AI builders:** Damodaran writes that the hyperscalers' capex will cause lower earnings until it starts paying off, and a hit to their free cash flows, which are calculated after capex.
**AI suppliers:** In his account, the same spending becomes revenue for chipmakers, network equipment makers, power plant builders and data center developers, with Nvidia the biggest beneficiary so far.
**Shareholders:** He writes that S&P 500 companies returned 63% of their earnings in the last twelve months, against 80% or more for much of the last two decades.
**The Business Engineer Lens**
This story maps onto the Business Engineer framework, "The AI Capex Map & The State of AI Hyperscalers."
The map puts it this way: "the capex race is no longer a corporate spending cycle. It is a sovereign-scale financing problem dressed in tech-company clothing — and the financial architecture is starting to invert."
As we read it, Damodaran's aggregates show that financing story from the accounting side: $1.9 trillion more book debt in a year, and cash returned to shareholders falling to 63% of earnings on his S&P 500 figure.
**What Is Not Established**
Damodaran flags two limits himself. His figures are "data from publicly trad
*[Editor's note: The source text ends mid-sentence here. The remainder of this section, and the bottom-line section, were not included in the material provided.]*