The September Effect is real for stock markets overall, but AI chip stocks have proven surprisingly resilient in recent years, suggesting that strong earnings and relentless demand for computing power can overcome seasonal selling pressure. Since 1928, the S&P 500 has lost an average of 1.1% during September and finished lower roughly 56% of the time, while the Nasdaq Composite has dropped 0.9% on average since its inception in 1971. Yet semiconductor leaders like Nvidia, Broadcom, Advanced Micro Devices (AMD), and Taiwan Semiconductor Manufacturing (TSMC) have shown mixed results when September arrives, with some years delivering gains despite the broader market headwinds.

What Exactly Is the September Effect?

Wall Street has transformed a recurring market pattern into financial folklore. The story goes like this: portfolio managers return from summer vacation, they rebalance their funds, and investors rotate out of growth stocks into defensive positions like consumer staples and utilities. This self-fulfilling prophecy has shown up consistently enough that traders treat September as a legitimate seasonal headwind rather than mere coincidence. The Dow Jones Industrial Average has dropped 1.1% on average during September since 1897 and finished the month positive only 42% of the time, reinforcing the pattern's staying power.

However, the pattern is not universal. Not every September delivers a bloodbath for stocks. The key insight is that September's weakness is real enough to respect but not necessarily powerful enough to dictate long-term investment decisions, especially when other forces like strong earnings and insatiable demand for AI infrastructure are at play.

How Have AI Chip Stocks Actually Performed During September?

The AI revolution has now experienced three September periods: 2023, 2024, and 2025. The results tell a nuanced story. In September 2023, both the VanEck Semiconductor ETF (SMH) and the iShares Semiconductor ETF (SOXX) dropped about 7%, which was ugly but still beat Nvidia's 10% decline and matched the performance of TSMC, AMD, and Marvell Technology. In September 2024, both ETFs finished the month flat, underperforming the positive performances of major AI chip stocks but insulating investors from losses seen in select laggards. Most remarkably, in 2025, the VanEck Semiconductor ETF gained 12% in September while the iShares Semiconductor ETF soared 11%, with the broader chip complex posting much higher gains.

These results suggest that strong earnings, falling interest rates, and insatiable demand for AI compute can overpower the historical September pattern. The lesson is straightforward: individual chip stocks can move much higher or much lower relative to a basket of stocks, but diversified chip-themed ETFs can provide exposure to the AI theme while reducing the whiplash from single-stock volatility.

Ways to Navigate Chip Stock Volatility Without Picking Individual Names

• Concentrated Chip Exposure: The VanEck Semiconductor ETF (SMH) holds roughly 26 chip names with Nvidia comprising approximately 23% of the fund, alongside major holdings in TSMC, Broadcom, Micron Technology, AMD, and ASML. This fund is up 51.4% year to date and 88.4% over the past year, but it carries significant volatility and is best suited for satellite positions rather than core holdings.
• Diversified Tech Exposure: The Invesco NASDAQ 100 ETF (QQQM) provides broader exposure to mega-cap technology companies including Apple, Microsoft, Nvidia, Alphabet, Amazon, Meta, Broadcom, and Tesla, plus roughly 90 other names spanning software, biotech, and consumer sectors. Returns are more moderate at 15.5% year to date and 24.72% over the past year, making it suitable as a core holding for investors seeking AI participation without single-stock risk.
• Global AI and Technology Exposure: The Global X Artificial Intelligence & Technology ETF (AIQ) tracks a broader index with roughly 89 positions and reaches beyond U.S. borders, holding Korean memory maker SK Hynix at 7.11%, Micron at 5.77%, AMD at 4.80%, and Samsung Electronics at 4.79%, plus international names like Alibaba and Tencent. Performance has been strong at 23.99% year to date and 40.01% over one year, though foreign holdings add currency and geopolitical risk.

For investors already in retirement, a common framework is capping thematic tech exposure at 10% to 15% of the portfolio combined, with QQQM doing most of the heavy lifting and SMH plus AIQ serving as smaller accelerators. This sizing approach treats volatile semiconductor positions like the satellites they are, protecting retirement plans from the damage that an ugly quarter in early retirement years can inflict.

Should September Volatility Change Your Investment Strategy?

The practical move is not to dump all chip stocks when September arrives. Instead, investors need to decide how much volatility they can actually tolerate. If you own individual stocks, sizing them so that a 10% drop does not force you to panic-sell is critical. AI chip stocks are growth names with frothy valuations and high liquidity, meaning they get sold first when the market gets uneasy. This is not the same thing as a fracture in the AI infrastructure thesis.

Data center build-outs are accelerating, hyperscalers are investing in custom application-specific integrated circuits (ASICs), and memory demand is not going away just because September arrived. The last two years proved that when fundamentals are strong, seasonal patterns can be overridden. Investors need to accept that September has never been a reliable signal that a correction or crash is on the way. At best, it is a seasonal phenomenon that sometimes shows up and drags the market down, but strong earnings and insatiable demand for AI compute can powerfully counteract that pressure.

The market's worst month has a history of being followed by better ones, including the traditional year-end rally. While the September Effect is real enough to respect, it should not be influential enough to dictate decisions that can affect your portfolio for the rest of the year. The key is matching your position sizing to your actual risk tolerance and time horizon, then letting the fundamentals of AI infrastructure demand do the heavy lifting.