A Third of the S&P 500's Profit Growth Is a Memory Price Cycle — and Half of It Is Still an Estimate

The Wall Street Journal reported Oct. 24 that about ten companies — among them Nvidia, Micron, Broadcom and Meta Platforms — account for roughly two-thirds of corporate profit growth, with the two chipmakers alone responsible for 34% of projected S&P 500 third-quarter profit growth. The arithmetic is not a rounding artifact. FactSet's Oct. 23 reading puts blended index earnings growth at 29.5%; its Oct. 16 breakdown showed the same index growing at roughly 20% once Micron and Nvidia are removed. A nine-point gap in the growth rate, from two names.

LSEG's contribution data puts Micron alone at 5.79 points of the index's 30.7% growth rate, with Nvidia and Micron together near 35%. Scale that: Micron added $34.5 billion of year-over-year net income, which implies the index booked roughly $183 billion more profit this quarter than a year ago. That arithmetic puts a third of the market's profit growth inside two stocks — one of them a memory manufacturer whose earnings move with contract DRAM prices, not with share gains.

Micron reported its September quarter on Sept. 30. Nvidia's fiscal third quarter ended on Oct. 25 — today — and results are confirmed for Nov. 17. The guide is $108.0B of revenue, up 89% from $57.0B a year earlier, at a 74.0% gross margin. On the proportions implied by LSEG's contribution figures, that single unreported quarter carries roughly $27 billion of incremental profit, about 15% of everything the index will book this quarter. That is the piece of the 34% that is not yet a fact.

The sensitivity is asymmetric in a way most investors are not tracking. A 10% revenue miss against Nvidia's $108.0B guide costs about $6.7 billion of net income and roughly 1.1 points of the index's quarterly growth rate. That turns one company's guidance into an index-level earnings event, on a day when the market is already paying 18.7 times for growth that is 29.5% only if the estimate holds.

Micron's fiscal Q4 2026 is the most profitable quarter in the company's history, and almost none of it came from selling more bits. Revenue reached $54.23B, up 379% year over year, and GAAP gross margin hit 86.8% — against 39.8% for fiscal 2025 as a whole and 44.7% in the same quarter a year earlier. The driver is price: DRAM average selling prices rose in the high-teens percentage range quarter over quarter, and DRAM alone was $39.8B, 73% of total revenue.

That is the uncomfortable part of the story. The index is collecting a commodity profit stream at a technology multiple: a memory maker's contract-price earnings now carry more operating margin than a chip designer with a software moat. The same operating leverage that produced 1,078% net income growth in a single quarter works in reverse if DRAM prices stop rising, and it reverses far faster than a moat would.

What holds it up is contract, not hope. Micron has signed 26 strategic customer agreements representing more than 35% of revenue through 2030, with customer financial commitments raised to $32 billion, the vast majority of it cash deposits. Three-quarters of that contracted revenue has a defined pricing framework with floor and ceiling bands, the balance resets to market prices periodically, and all of it is take-or-pay with roughly $150 billion of remaining performance obligations. The company has also contracted the vast majority of its calendar 2027 HBM bit supply at significant year-over-year price increases.

The transmission order matters more than the size of any single number. A memory capex pause appears first in Lam Research orders, then in DRAM contract prices two to four quarters later, then in Nvidia's and Broadcom's gross margins, and only then in the index's earnings. Meanwhile the demand side is not weakening: Meta Platforms's depreciation rose 44% year over year to $7.15B in the September quarter, which is what a financed build looks like once it starts rolling through the income statement.

The index's forward 12-month P/E of 18.7 sits below its ten-year average of 19.1 and its five-year average of 19.8, and it has compressed from 19.0 at the end of September. On paper that is a modest discount for a market growing earnings at 29.5%. Priced against a 5.00% 10-year Treasury, a 5.35% forward earnings yield leaves about 35 basis points of equity risk premium for owning a concentrated, cyclical earnings stream — and the previous quarter's growth had the same character, with roughly 19 points of it coming from mark-to-market gains on AI investments at Alphabet and Amazon that Reuters documented as the difference between 54% headline growth and 35% clean.

Near term, the calendar does the work. Nvidia reports Nov. 17 against a $108.0B revenue guide and a 74.0% gross margin. SK hynix reports Oct. 28 and Samsung Electronics Oct. 29, the first two independent reads on HBM4 and DRAM contract pricing since Micron's print. Micron has guided fiscal Q1 2027 to $61.5B of revenue and $38.15 of non-GAAP EPS, with results expected Dec. 23. The pricing that produced this quarter's profit is contracted, not hoped for.

The long-term risk is the mirror image of the near-term comfort. Micron said it will raise fiscal 2027 capital spending above prior plans, with most of the increase aimed at construction for clean-room capacity arriving in late calendar 2028 and beyond. That makes today's shortage the cause of tomorrow's surplus, and the supply response is already funded. TSMC said in October that it is approaching 2027 prudently, and the hyperscalers' depreciation is compounding fast enough — Meta Platforms's up 44% year over year — to eventually compete with the capital spending that funds the next order.

For a long-only investor, the practical consequence is that the S&P 500's 18.7x forward multiple no longer prices diversification — it prices a memory contract cycle running through 2027, with roughly $150 billion of take-or-pay performance obligations underwriting the 2026-27 numbers and a funded construction wave underwriting the 2028 reversal. For a relative-value investor the arithmetic points the other way: ex-Micron and ex-Nvidia, the same index grows earnings at about 20%, and the trade that makes money is not the index but the chain of companies whose spending sets the memory price the index is now valued on.

What would change the view is specific. Micron holding fiscal Q1 2027 revenue at or above the $61.5B guide with gross margin near 86% would extend the cycle into 2028 and make the concentration more durable, not less. A Nvidia guide below $108.0B, or a memory capex pause visible in Lam Research orders, would pull the index's growth rate back toward 20% and its multiple back toward the high teens. Until one of those happens, the market pays 18.7 times for a profit stream contracted through 2027 and whose reversal is already being financed.