History Says September Is Brutal for Stocks. Is This the Year AI Stocks Prove It Wrong?

Welcome to September -- bloody, red, loss-infused September. I'm being dramatic, but since 1928, the S&P 500 has averaged a 1.2% decline in the month, the only month on the calendar with a negative long-run average. Over nearly a century, the market has only gained ground in Septembers 44% of the time.

However, I think that with some help from artificial intelligence stocks, this September will be one of the less common ones that ends in the black. And the reason has to do with what already happened over the past eight months.

Image source: Getty Images.

The setup nobody expected in January

2026 has been the year of the great rotation. What started as modest reshuffling turned into one of the most significant capital moves in two decades, with large amounts of money leaving a handful of AI megacap stocks and spreading into energy, industrials, materials, staples, and small caps. By March, analysts were describing it as a shift from silicon to steel.

The numbers behind it are stark. The iShares Russell 1000 Value ETF (NYSEMKT: IWD) returned roughly 32% over the past year, compared with 14% for the iShares Russell 1000 Growth ETF (NYSEMKT: IWF). July delivered a violent rotation where the energy sector gained 12.13%, and the technology sector fell 7.96%. Energy and tech stood 25 percentage points apart at one point.

That matters because it means the AI trade entering September is not the crowded, everyone-owns-it position it was a year ago, or earlier this year. A lot of the weak money has already left.

Why the money will come back

Here is the part I find most compelling. The rotation happened while AI fundamentals were improving, not because they deteriorated.

Hyperscalers have repeatedly revised their capital expenditure estimates for 2026 upward, from roughly $600 billion in November 2025 to $620 billion in January to about $750 billion by mid-year. Four of the five hyperscalers raised their 2026 capex forecasts during a single earnings week, lifting their combined spending expectation from $670 billion to $725 billion in five days. Goldman Sachs now projects global AI investments will top $1 trillion this year.

Demand is not theoretical. Alphabet's Google Cloud grew by 63% year over year, Microsoft's AI run rate hit $37 billion, and Amazon Web Services posted its fastest growth in 15 quarters. Every hyperscaler describes its market as supply-constrained.

The September catalyst calendar

This is where timing works in favor of AI names. Broadcom reports its next set of results after the close of trading on Sept. 2, Oracle on Sept. 8, and Adobe on Sept. 10. Micron Technology will follow later in the month.

Broadcom carries the heaviest weight. Last quarter, its guidance freeze triggered a stock drop, so investors will be watching its AI chip revenue and fiscal 2027 AI guidance specifically. Oracle has form here, having risen 36% in a single September session in 2025 on the back of backlog disclosure.

A month full of concrete disclosures from tech companies with signed contracts is a different environment than a month where the market is driven by pure sentiment. When positioning is light and fundamentals are strong, earnings reports become a reentry trigger rather than a risk event.

What I am watching

The honest counterweight is monetization speed. AI services generate roughly $25 billion in direct revenue, compared with the infrastructure being built to deliver them. Capex as a share of sales will reach 86% for Oracle and 47% for Microsoft this year. That gap has to close eventually.

But that reckoning is a 2027 or 2028 problem. It is not a September 2026 problem. What this particular September has going for it is a set of conditions that rarely line up at once. The AI trade is no longer crowded, because eight months of rotation pushed a lot of money out of it and into energy, financials, and value names. The companies doing the spending keep raising their budgets rather than trimming them, and they are doing it in public, on earnings calls, with numbers attached. Demand still outstrips supply badly enough that every hyperscaler says so out loud. And this month's calendar is packed with companies that have to show their work.

Put simply, the sellers have already sold. What is left is a group of businesses with contracted revenue and improving fundamentals heading into a month where they get to prove it. History says September is often brutal, and history is usually worth listening to. This time, though, I think the setup will matter more than the season.

Micah Zimmerman has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Adobe, Alphabet, Amazon, Broadcom, Goldman Sachs Group, Micron Technology, and Oracle. The Motley Fool recommends the following options: long January 2028 $330 calls on Adobe and short January 2028 $340 calls on Adobe. The Motley Fool has a disclosure policy.