Authored by Lance Roberts via RealInvestmentAdvice.com

For the past few months, the “Magnificent 7”, or more colloquially known as “Mag 7,” stocks have looked more like the “Lag 7.” The market’s largest growth names, Apple (AAPL), Microsoft (MSFT), Google (GOOG), Amazon (AMZN), Tesla (TSLA), Meta (META), and Nvidia (NVDA), have trailed badly. Those Mag 7 stocks now carry a bearish story built around one word: capex. As we noted in last weekend’s report, the selling has been sharp. The question is simple. Does it reflect a real problem, or just a narrative investors tell themselves to justify chasing something else?

Start with what the headline hides. This was not the whole group falling together. Microsoft is down roughly 22% this year and just closed its worst month since 2000. Meta is off about 14% over six months. Yet Alphabet is up around 12%, with Apple and Nvidia also higher. The pain landed on the two heaviest capex builders.

So the market is not indiscriminately dumping the Mag 7 stocks. It is discriminating by spending intensity. The hardest builders took the most pain.

That capital did not leave the market. It rotated. Citadel Securities strategist Scott Rubner notes semiconductors now make up nearly a fifth of the S&P 500. That is the highest share on record, and roughly quadruple their 2020 weight.

Retail chased it hard. In June, it traded about $1.9 billion per day in semiconductor options premium, near six times the historical average, mostly in calls. Receivers like Micron ran up more than 200% while the spenders bled. Wedbush’s Dan Ives called it a bifurcated tape; the builders dropped into what he termed the “penalty box.“

Here is the inconsistency worth sitting with. For years, no one complained when these companies returned cash through buybacks. A buyback does nothing for the underlying business. At best, it offsets dilution and hands cash back to sellers, who are mostly corporate insiders. The market applauded it anyway, because it inflated asset prices and the market overall.

“It is a pretty easy task to see whether or not corporate stock buybacks influence stock prices. As we penned last year, the impact of buybacks extends beyond individual companies. Since 2000, net corporate buybacks have accounted for 100% of the equity market’s net asset purchases—a reflection of the diminished participation from pensions, mutual funds, and individual investors:”

Now those same companies are directing cash into capacity, the data centers and chips behind AI, more than $650 billion this year. And the market recoils. That is a psychology problem more than an accounting one. Investors prefer the certainty of a buyback’s return to the deferred payoff of an investment. Howard Marks has long argued that the crowd’s comfort is usually mispriced. Aversion to the build phase is where opportunity often hides.

That said, capex is not automatically good. It creates value only when the return clears the cost of capital. And the AI payoff is genuinely unproven. Ed Yardeni captured the doubt, writing that investors seem to be feeling “AI Fatigue,” questioning whether the spending will ever pay off. That skepticism is fair. The real question is whether the market is pricing it or overreacting.

The current fear has a rhyme. From 2016 through 2020, these same names poured cash into data centers to build out the cloud. Investors fretted about margins and runaway spending. That capex converted into the revenue and margin leadership that has defined the market ever since.

The long-run numbers make the point. Since 2016, Mag 7 revenue has grown by close to 375%, compared with roughly 95% for the S&P 500. Earnings echo it. The group’s 2026 growth is tracked near 38%, versus about 19% for the S&P 493. The gap is narrowing as the index recovers, part of why money rotated away. Even so, the fundamental leadership remains intact. A three-month, flow-driven drawdown does not erase a decade of compounding.

Will the AI build rhyme with the cloud build? That is the honest uncertainty. What the record shows is that the market feared this exact pattern before and was wrong to sell the builders wholesale. We will take up the other side, the risk that heavy spending and slow depreciation are flattering today’s reported earnings, in a separate contrarian piece soon.

This is where the two possibilities meet. If the selloff were only a story, valuations on the Mag 7 stocks would still be stretched. They are not. HSBC strategists Duncan Toms and Max Kettner show the leaders at the low end of their own decade-long forward multiples. Nvidia sits near 20 times forward earnings, close to a 10-year low for the stock. Meta is near 16, Microsoft and Alphabet are near 24. The expensive corner is now the defensive names. Costco, Walmart, and Monster Beverage sit near the top of their ranges.

The cleaner tell is what drove the de-rating. It happened against rising earnings, not falling ones. Meta, Amazon, Microsoft, Nvidia, and Broadcom have all seen trailing multiples fall over the past year because earnings have outpaced share prices. That is not a broken story. It is a repricing.

In the short term, the underperformance is stretched to an extreme that few appreciate. Ned Davis Research’s 21-day rate of change on Mag 7 relative strength, against the S&P 500 ex-Mag 7, has fallen to roughly three standard deviations below average.

That reading has appeared only a handful of times in eleven years. One caution matters. It measures the speed of the decline, not a floor. So the pace looks unsustainable, but a bounce is not guaranteed. Positioning agrees with Goldman and Morgan Stanley’s prime data showing hedge funds near multi-year lows in these names.

Our answer rejects the either/or because both are true in sequence. The first leg was a narrative, capital leaving the spenders to chase semiconductors. But the flows ran far enough that a story-driven decline has repriced the Mag 7 stocks into an opportunity. It is concentrated in the very names that fell the most, Microsoft and Meta.

That is how we are positioned. Last Wednesday, our Sector and Factor Rotation model shifted from a value tilt toward growth. We added mega-cap exposure to the weakness. Net equity exposure barely moved. This is a TACTICAL add, not a verdict that the AI spending debate is over. The exit is defined. If free cash flow and the depreciation catch-up confirm at second-quarter earnings in late July, we sell and move on.

Know the risk on the other side. The semiconductor chase is crowded and heavily levered. Citadel Securities data show leveraged ETF assets at a record high, with semiconductor exposure up about 175% since March. Half of all retail options now expire the same day. One-month equity financing spreads sit near 138 basis points over the risk-free rate. When a trade is that crowded and levered, the unwind tends to be fast. That is one more reason the rotation back toward the spenders can move quickly.

There is a caution on the bull case, too. Nearly every major desk, from BofA and Morgan Stanley to Goldman, JPMorgan, and HSBC, is leaning the same contrarian way. When everyone agrees a trade is contrarian, it stops being contrarian. That crowding is the real risk to our own view. It is why we sized this as a trade with a stop, not a conviction position. As Bob Farrell warned, excesses in one direction tend to invite excesses in the other. The crowd rotated hard into chips. We think the rubber band snaps back toward the Mag 7 stocks, and we will be quickly proven wrong if it does not.