A Major Shift Could Soon Happen in the Mag 7, Here’s What It Means for Your Portfolio

Be honest: if you opened your brokerage app lately and winced, you’re not alone.

For years, the Magnificent Seven, Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta, and Tesla, felt like the only trade that mattered. They gobbled up market returns. They turned retirement accounts into something worth bragging about at dinner parties. They made you feel smart for simply doing nothing.

Then 2026 arrived. And the vibe… shifted.

Every single Mag 7 stock is down double digits from its 52-week high. Microsoft slumped 23% in the first quarter alone, its worst three-month stretch since the financial crisis. The Roundhill Magnificent Seven ETF (MAGS) is underwater year-to-date while the plain-vanilla S&P 493, that’s the rest of the index, the boring companies, is quietly outperforming.

Something is changing beneath the surface. It’s not that these companies are suddenly bad businesses. Far from it. But the story that carried them for three straight years is being rewritten in real time. Let’s walk through what’s actually happening, no jargon, no panic, just the data and what it means.

The Rotation That Wasn’t Supposed to Happen

Here’s a stat that should make you pause: 493 of the 500 stocks in the S&P 500 are outperforming the Magnificent Seven so far this year.

Read that again. Four hundred and ninety-three.

The Defiance Large Cap ex-Mag 7 ETF (XMAG), which is exactly what it sounds like, the S&P 500 minus those seven giants, was actually up 3.28% at one point while the Mag 7 ETF (MAGS) had fallen more than 4%. That’s not a blip. That’s capital quietly walking out the back door while everyone was still humming the old song.

J.P. Morgan’s equity strategy team put it bluntly: "The biggest are no longer the best. In 2023, all seven Mag 7 stocks easily beat the S&P 500; in 2024, six did. Year-to-date in 2026, just two are positive and only one is outperforming."

What changed? Three things, and they’re all connected.

AI’s Big Spending Hangover

The Mag 7 didn’t just dip their toes into AI, they cannonballed into the deep end with a checkbook.

Four of the biggest players, Alphabet, Amazon, Meta, and Microsoft, are expected to spend nearly $700 billion combined on AI infrastructure this year. That’s a roughly 60% surge from 2025 levels.

Let those numbers breathe for a second. Seven hundred billion dollars. To put that in perspective, those four companies together generated $200 billion in free cash flow last year, down from $237 billion the year before. So spending is rocketing up while the cash cushion is thinning. That math makes Wall Street nervous, and rightfully so.

And here’s where it gets personal for shareholders: when a company like Meta raises its full-year capex guidance from a range of $115–$135 billion to $125–$145 billion, the stock doesn’t celebrate, it gets punished. Meta shares slid 10% the day after that revision. Amazon’s proposed $200 billion capex for 2026 is unprecedented, even for a company that practically invented "spend now, profit later."

The flashpoint is clear: capital expenditure is rising faster than free cash flow. The era of surplus cash quietly compounding into buybacks is being redirected into an arms race. And markets, for all their patience with tech narratives, eventually ask the same simple question: when does this spending produce durable earnings?

Microsoft is the poster child of this squeeze. It’s now expecting roughly flat free cash flow for the first time in years, primarily because it’s pouring money into data centers. In fact, Melius Research analyst Ben Reitzes noted he "wouldn’t be surprised if Broadcom generated more free cash flow than Microsoft this year." Let that sink in: a semiconductor supplier might print more cash than one of the world’s most profitable software companies. The cash is flowing out of the hyperscalers and into the infrastructure layer, Nvidia, Broadcom, the picks-and-shovels players.

That’s not a crisis. But it is a fundamental shift in who benefits most from the AI buildout, at least in the near term.

The Power Problem, AI’s Very Physical Limit

This one sounds almost too sci-fi to be real, but stick with me: AI needs electricity. A mind-boggling amount of it.

BlackRock estimates the U.S. will need approximately 148 gigawatts of additional power capacity by the end of the decade just to satisfy data center demand, more than triple the 42 GW data centers consumed in 2025. The International Energy Agency projects that global electricity demand from data centers will more than double by 2030, reaching around 945 terawatt-hours, which is roughly equivalent to the entire current electricity consumption of Japan.

Goldman Sachs has been steadily raising its forecast: global data center power demand is now expected to surge 220% from 2023 levels by 2030.

Why does this matter for your portfolio? Because the Mag 7 are the ones writing the checks to build these power-hungry data centers. And here’s the twist: they don’t control the grid. Utilities move slowly, they’re optimized for reliability and regulatory compliance, not the breakneck pace of AI competition.

This creates a fascinating second-order rotation. Investors aren’t just rotating out of Mag 7 stocks, they’re rotating into energy, utilities, and the infrastructure layer that powers the AI buildout. As one market commentary noted, "in an environment where investors are increasingly uncertain about where the next technological disruption might occur, some capital has rotated toward areas tied to a more fundamental and unavoidable need: power."

The Mag 7 are still driving the AI revolution. They’re just not the only ones, or necessarily the most profitable ones, reaping the rewards right now.

The Earnings Gap Is Closing, And That Changes Everything

For about three years, the Mag 7’s earnings growth utterly dwarfed the rest of the S&P 500. That gap justified their premium valuations. But the ground is shifting.

Morgan Stanley projects Mag 7 net income growth of 25% for 2026, still impressive, more than double the S&P 493’s 11%. But Goldman Sachs sees that gap narrowing to just about 4 percentage points for the full year. And here’s the critical timeline: according to Morgan Stanley, the S&P 493’s earnings growth could catch up to the Mag 7 by the fourth quarter of 2026.

When growth rates converge, valuation premiums compress. It’s not that the Mag 7 become bad investments, they just stop being the only investments. And that’s exactly what we’re seeing: the bull market broadening out beyond a handful of names.

Meanwhile, the Mag 7 currently trade at an average forward P/E of around 28x, remarkably, that’s actually below the Nasdaq 100’s 30.7x. J.P. Morgan’s trading desk went so far as to publicly declare that Mag 7 valuations "might be too cheap." They’re not wrong: these stocks haven’t looked this reasonable on a multiples basis in a while. But reasonable isn’t the same as irresistible, and with free cash flow under pressure, "cheap" can stay cheap until the spending trajectory changes.

What Smart Investors Are Doing Right Now

None of this is a eulogy for the Magnificent Seven. These are still seven of the most profitable, innovative companies on the planet. But the way you own them, and what else you own alongside them, probably needs a rethink.

Here’s the pattern emerging from institutional commentary and fund flows in early 2026:

1. Don’t abandon the Mag 7, right-size them. Wealth managers are deliberately trimming Mag 7 exposure not because the companies are broken, but because the past concentration of returns is unlikely to repeat, and being overweight seven stocks in a broadening market means underweighting 493 others that are finally joining the party. Some advisors are using options overlays to manage downside while staying invested in secular AI growth.

2. Follow the cash flow, not just the narrative. The AI theme hasn’t died, it’s migrated downstream. Broadcom potentially surpassing Microsoft in free cash flow is a signal: infrastructure providers, semiconductor supply chain companies, and energy infrastructure names are capturing more of the value as capex flows through the system. This isn’t a rotation out of AI; it’s a rotation into different parts of AI, industrials, materials, power generation.

3. Diversification is making a comeback. For years, diversifying away from Big Tech was a costly mistake. In 2026, diversification is finally working again. The S&P 400 mid-cap index and S&P 600 small-cap index are outperforming the S&P 500 year-to-date. Value stocks are beating growth by the widest margin since last spring.

4. Watch the Mag 7 individually, not as a monolith. Alphabet rallied nearly 10% over two days after its earnings showed strong cloud growth and a disciplined AI spend narrative, while Meta got hammered on its capex raise. These aren’t seven clones. Different business models, different capex trajectories, different risk profiles. Stock selection within the group matters more than ever.

A final thought: Morgan Stanley’s team has flagged the fourth quarter of 2026 as the moment the S&P 493’s earnings growth could actually catch the Mag 7’s. If that plays out, the "own Mag 7 and nothing else" trade that worked so beautifully from 2023 to 2025 will look increasingly like a relic. The market is already pricing that possibility, you can see it in the ETF flows, the sector rotations, and the quiet outperformance of the boring 493.

What This All Means for You

The Magnificent Seven aren’t collapsing, they’re maturing into a phase where capital discipline, energy infrastructure, and earnings quality matter more than pure growth momentum. That’s a healthy transition, but it’s also a disorienting one if you’re still positioned like it’s 2024.