Why did the Nasdaq fall 2.6% on October 22, 2026, even as a majority of stocks on the New York Stock Exchange advanced? A sharp reassessment of the cost of artificial intelligence investments hit some of the market’s most influential companies. Meta shares fell 7%, semiconductor stocks tumbled, and the S&P 500 declined 1.21%. The Dow Jones Industrial Average, however, finished marginally higher. The contrast is a reminder that “the stock market” is not one uniform investment, especially when a relatively small group of technology companies carries so much weight in popular indexes.
The final US session figures were reported by Reuters on October 22 and Associated Press: the Nasdaq Composite closed at 26,008.86, down 694 points or 2.60%; the S&P 500 ended at 7,663.43, down 93.74 points or 1.21%; and the Dow rose 11.89 points, or 0.02%, to 52,867.98. These are October 22 closing figures, not live prices for October 23. The events that drove the decline were intertwined: Meta’s earnings, its extraordinary capital spending, wider concerns about AI valuations, and unease about inflation and interest rates.
What triggered the Nasdaq selloff on October 22?
The most visible pressure came from Meta Platforms, whose shares fell 7% after its latest earnings release. The company had reported rapid revenue growth and more advertising activity, but investors focused on the cash required to build AI infrastructure and on plans to create a new class of non-voting shares. Other major AI-related businesses faced selling pressure at the same time, pulling the Nasdaq down more sharply than the broader market.
According to Reuters, the Philadelphia Semiconductor Index fell 6.7%, its largest single-day percentage drop since June 23. Intel, AMD, Sandisk and Marvell each declined by more than 8%. These figures illustrate why the technology-heavy index suffered a particularly severe move. They do not mean that every semiconductor company suddenly experienced weaker demand or that a fundamental change in AI adoption was established in one session.
Market participants also considered rising oil prices linked to Middle East tensions and the possibility that persistent inflation might require a restrictive Federal Reserve policy. That is a second layer of pressure: high long-term borrowing costs can make far-off expected earnings less valuable in current terms. It is nevertheless impossible to assign a precise share of the day’s market move to one cause based only on commentary after prices have changed.
Why did Meta fall after reporting 26% revenue growth?
Meta’s official Q3 2026 financial results show that quarterly revenue rose 26% year over year to $64.60 billion. Daily active people across its family of apps averaged 3.64 billion in September, up 3%. Ad impressions grew 10%, while the average price per ad increased 14%. Those are not the numbers of a collapsing advertising business. They show that Meta’s core commercial engine was still growing strongly.
Operating profit, however, fell 37% to $12.99 billion. Importantly, the company recorded a $9.89 billion charge associated with a previously announced agreement with US state attorneys general concerning social media and teen mental health. Meta said that without that charge, operating income would have been approximately $22.88 billion. It would therefore be misleading to say that the entire drop in reported operating profit was caused by AI spending alone.
Net income rose to $9.86 billion from $2.71 billion, with the comparison affected by significant differences in tax charges between the two periods. These accounting effects are exactly why an investor should read the underlying release rather than deciding whether results were good or bad from a single earnings-per-share headline. The market’s larger concern was not merely the completed quarter but the scale of future investment required to deliver Meta’s promised AI capabilities.
How much is Meta actually spending on AI infrastructure?
Meta reported $34.91 billion of third-quarter capital expenditures when principal payments on finance leases are included. Its operating activities generated $32.79 billion in cash, while free cash flow was negative $2.12 billion under Meta’s published non-GAAP definition. One year earlier, quarterly free cash flow was positive $10.63 billion. This change attracted considerable attention because Meta has historically generated substantial cash from advertising while funding its own expansion.
Negative free cash flow is not equivalent to bankruptcy, nor does it mean the company ran out of cash. Meta reported $82.45 billion in cash, equivalents and marketable securities at the end of September. It means that, under its chosen measure, spending on property, equipment and finance leases exceeded the cash generated by operations during this quarter. If those investments lead to highly profitable services later, some of today’s spending may be justified. If returns disappoint, the same figures become evidence of costly overinvestment.
Management expects 2026 capital expenditures, including finance-lease principal, of $135 billion to $145 billion, narrowing an earlier forecast range of $130 billion to $145 billion. This is a substantial commitment. It reflects equipment, data centers and other infrastructure whose useful economic lives and future returns cannot be assessed from one earnings report. Meta argues that AI is already helping its core advertising business through recommendations and monetization, but the incremental value of specific long-term projects is difficult for outside investors to isolate.
What does Meta’s new Class C stock mean?
Meta also announced that its board approved the establishment of publicly listed, non-voting Class C common stock, expected to become effective in January 2027. The plan includes a one-time stock dividend of one Class C share for each existing Class A or B share, with details to be finalized. Meta expects the market price of its Class A stock to reflect an effect comparable to a two-for-one split once the distribution occurs.
A stock dividend of this kind is not a guaranteed increase in an investor’s economic wealth. Ownership is divided among additional shares, and market prices adjust. More consequential is the ability to issue non-voting equity for future capital raising, employee compensation or acquisitions without conferring the same voting rights. Meta says the structure is intended in part to help fund the scale of AI investment while preserving its founder-led governance model.
The concern for shareholders is not simply that a new ticker or share class might exist. Future issuance can dilute the economic interests of existing shareholders depending on its terms and use. The potential benefits depend on whether raising capital produces investments whose returns exceed their cost. Governance also matters: non-voting shares make it possible to finance expansion while concentrating voting influence among the holders of other share classes.
Why did the S&P 500 fall while the Dow stayed almost flat?
The explanation begins with index construction. The S&P 500 is weighted by companies’ market capitalization, so very large technology companies influence its return more than many smaller members. The Nasdaq Composite has substantial exposure to technology and growth businesses. The Dow is price-weighted and contains only 30 companies, which means its movements need not mirror those of either index.
On October 22, Reuters observed that declining stocks outnumbered advancers on the Nasdaq, while advancers slightly outnumbered decliners on the New York Stock Exchange. The AP market summary likewise noted that many US stocks outside technology rose. Procter & Gamble, Lockheed Martin, Blackstone and Spotify all gained after company-specific news, according to Reuters, demonstrating that the selloff was not evenly distributed throughout the economy.
For someone holding a broad-market ETF, this distinction is practical. The fund’s performance depends on which companies and sectors dominate its index, not on the percentage of stocks that happen to rise. A small number of very large stocks can pull a market-cap-weighted index down even when many constituents advance. Diversification spreads company-specific exposure, but concentration can still emerge at the index level as particular sectors become more valuable.
Does a 2.6% Nasdaq decline prove the AI bubble has burst?
No. The decline represents a sharp one-day repricing, not an experiment that settles the question of AI’s long-term economic value. AI might deliver substantial productivity improvements while some companies still spend too much or generate disappointing investment returns. A technology can succeed while investors in certain securities lose money, and a richly valued stock can be vulnerable to modest downward changes in expectations.
Reuters reported that the semiconductor index was more than 26% below its June peak after the session, yet still up approximately 51.5% for 2026 to date. Both parts of that observation matter. A major drawdown can coexist with strong year-to-date performance, particularly in a volatile sector. Interpreting such figures requires a defined starting date rather than whichever comparison supports a dramatic narrative.
It is also too early to conclude that spending on AI infrastructure is inherently wasteful. Capital expenditure can depress near-term free cash flow while producing productive assets for years. The challenge is distinguishing durable demand from the risk of overbuilding, pricing pressure and rapid changes in hardware. Investors cannot resolve that distinction by looking at a single quarter’s cash outflow or an individual day’s stock-market reaction.
What should investors watch after the selloff?
Useful questions begin with measurable trends. Will AI-related capital expenditures grow faster than the revenues and profit they enable? Can Meta continue improving its advertising business while financing enormous infrastructure commitments? Will large cloud providers disclose returns from their investments in enough detail for outsiders to evaluate them? How do chip suppliers’ orders and margins respond if big customers revise expansion plans?
The next set of earnings results across cloud computing, semiconductors and enterprise software will offer more evidence. Other variables matter too: bond yields, central-bank policy, geopolitical risks and the availability of financing. An investor who interprets every market movement as an AI story can overlook broader conditions that change the valuation of many different assets simultaneously.
There is a related lesson in InsightArea’s analysis of Tesla’s Q3 2026 results: the relationship between business performance, capital expenditure and share prices is more complicated than an earnings headline. Tesla’s cash-flow surprise was evaluated against expectations about its future projects, while Meta’s cash-flow deficit raised concerns about the sheer cost of its expansion. Neither result can be reduced to a simple rule that more spending is always good or always bad.
For long-term investors, the October 22 selloff is a reminder to understand what they actually own, including the hidden concentration of a technology-heavy ETF. It does not establish what the market will do next. The financial question is whether future cash flows justify current valuations under a realistic range of outcomes, and that judgment remains uncertain even when the underlying technology is genuinely transformative.
By Costin Liculescu, creator of InsightArea. This article explains financial news for educational purposes and is not personalized investment advice.
Sources: Meta Investor Relations: Q3 2026 earnings, published October 21; Reuters: US market close, October 22; Associated Press: October 22 index performance; Reuters global-markets report carried by Euronext. Price changes and index levels refer to the October 22 US closing session, not the October 23 trading day.
Comments are closed.