JPMorgan Says the Mag-7's Worst De-Rating in a Decade Is Done — That's a Setup, Not a Buy

Mislav Matejka's JPMorgan Equity Strategy team put a number on the trade on Sept. 30: the Magnificent Seven's 12-month forward P/E relative to the broader market now sits near one standard deviation below its historical median — the deepest relative discount in a decade. The bank frames it as a reset that has 'largely run its course,' not a buy rating. In the same window, JPMorgan's institution-focused Market Intel Desk (led by Andrew Tyler) flipped from neutral to tactically bullish on Sept. 29, citing steadier bond markets, softer oil, and the upcoming earnings season.

The two signals are deliberately framed differently. The Matejka note is structural — it concedes that some multiple compression is justified by rising leverage and weaker free-cash-flow profiles driven by AI capex, then argues the reset has gone far enough. The Market Intel Desk's tactical bullish call is shorter-horizon — a positioning, flows and seasonality trade. Both views are constructive, but only the first gives investors a new valuation anchor; the second can reverse inside a week if positioning unwinds.

JPMorgan itself identifies the mechanism that drove the de-rating: capex. The four biggest hyperscalers — Microsoft, Alphabet, Amazon and Meta — are on track to spend roughly $725 billion on capital expenditures in 2026, up 77% from $410 billion in 2025. Bank of America frames the same number more sharply: hyperscaler capex now equals about 96% of operating cash flow, versus 40% in 2023. The reset 'largely running its course' means the market has stopped punishing the spending; it does not mean the spending has stopped.

Read this way, Matejka's note is a compressed-version of the same view JPMorgan's mid-year outlook carried: hyperscalers will keep spending, but valuation now reflects it. The Mag-7 are expected to grow earnings 20% in 2026, almost double the 11% pace for the rest of the S&P 500. For the reset to stay 'largely complete,' those 20% numbers have to keep showing up in prints — and the four-week window starting Oct. 13 is where the market will test that.

JPMorgan's own Q3 release lands on Oct. 13, six trading days after the Sept. 30 note. The market now has to price the message before that print — and the message itself is partially a bet on the prints that follow. The Mag-7 schedule compresses six of seven companies into eight trading days: Tesla on Oct. 21–22, then Meta Platforms, Alphabet and Microsoft on Oct. 28, Amazon and Apple on Oct. 29, and Nvidia capping it on Nov. 17.

For JPMorgan, the setup is unusually clean. Q2 2026 was the most profitable quarter in U.S. banking history — net income of $21.2 billion, EPS of $7.70, IB fees of $3.3 billion (+30% YoY, the highest since 2021), and a record $21.6 billion in Markets revenue with equities up 86%. The Sept. 15 Petno preview told the market to expect Q3 IB and trading to grow 'in the mid-to-high teens' — i.e., the comp isn't a fluke, it's a run-rate. That tailwind is also why the bank's own equity desk is willing to lean bullish into Mag-7 earnings: JPM captures a slice of every underwrite, every block trade, and every options flow that surrounds the Mag-7 prints.

Over the next three to six weeks, JPMorgan's note is a confirmation trade. If Meta, Alphabet and Microsoft guide to capex that simply confirms the $725B aggregate — without surprising to the upside on FCF pressure — and Nvidia prints in line with its $108B guide, the 'reset largely complete' framing holds and the multiple has room to rebuild. The 9-point earnings growth gap between the Mag-7 (20%) and the rest of the S&P 500 (11%) becomes the engine, and the discount at ~1 standard deviation below median looks cheap in hindsight.

Over one to three years, the same note is a structural test. T. Rowe Price estimates the rolling 12-month free cash flow for the hyperscalers could approach zero by early 2027 if capex keeps outrunning operating cash flow. Epoch AI has the cash-capex and operating-cash-flow curves crossing in the third quarter of 2026. If 2027 earnings don't start converting capex into incremental revenue at scale, the 'largely complete' framing ages into a bear-market bounce — and the same 13% drawdown the Mag-7 took from mid-May is on the table again.