'Magnificent Seven' stocks are the cheapest in a decade, by one measure

Morgan Stanley Wealth Management says the Magnificent Seven’s valuation premium is down to ~10% versus the other 493 S&P 500 names (lowest in over a decade) while the group still holds a ~45% annual earnings-growth advantage, making some megacaps look like bargains after their rough year. The firm also flags hyperscalers as “downright cheap,” suggesting selective re-entry into Mag-7 AI winners as investors fade semis exposure and as firms move away from costly/energy-heavy tokenmaxxing. Nvidia is cited with a next-12-month current P/E of 18.7 versus a historical 36.9, with Bank of America reiterating a Buy as EPS discount skepticism persists.

Analysis

The market is starting to separate AI exposure into two trades: monetization and workflow control versus raw capex intensity. That favors the hyperscalers with distribution, data, and balance-sheet flexibility (MSFT, GOOGL, AMZN) while leaving the hardware stack more vulnerable to multiple compression if spending growth normalizes. A secondary beneficiary is the fee-earning credit/franchise layer at MS and BAC if debt issuance persists, but that is a short-duration tailwind, not a core earnings driver.

Over the next 1-3 months, the key catalyst is whether capex growth decelerates before revenue proves out. If AI adoption shifts toward lower-cost hybrid inference, the market will likely discount fewer incremental accelerator purchases, which is negative for the broader semiconductor basket even if usage continues rising. If, instead, hyperscaler commentary shows another step-up in GPU demand or backlog, the current rotation away from semis likely reverses quickly.

Contrarian view: the low relative valuation of the Mag 7 is only attractive if earnings quality survives a more capital-intensive AI race. The market may be underpricing the risk that custom ASICs and hybrid workflows commoditize parts of NVDA’s system margin over 6-18 months, but it may also be underestimating NVDA’s near-term pricing power. Falsifiers are straightforward: re-accelerating hyperscaler capex guidance, stronger-than-expected cloud monetization, or another leg higher in SOXX breadth would invalidate the bearish semiconductor rotation thesis.

Key Decisions for Investors

• Enter a 1-3 month pair trade: long MSFT/GOOGL/AMZN basket vs short SOXX or SMH. Prefer entry on any further semiconductor strength or a 2-4% pullback in the Mag 7 basket; target 8-12% relative outperformance if the market rotates toward monetization. Cover if next cloud earnings re-accelerate capex guides.
• Add a tactical long NVDA on weakness, but size it smaller than the hyperscaler basket. The cleanest expression is a 3-6 month call spread or staged equity entry only after post-rally digestion, with the thesis failing if forward commentary points to margin pressure from custom ASIC substitution.
• Avoid initiating fresh long exposure in memory/AI hardware proxies after extended outperformance; use them as hedges against long-hyperscaler exposure until capex data confirms the next leg. This is a 1-2 quarter timing call, not a structural short.
• Set an earnings-season alert on MSFT, GOOGL, and AMZN for capex and cloud margin commentary. If capex stays elevated without corresponding monetization, rotate back into semis; if capex flattens and margins hold, extend the long Mag 7 trade.