Amazon (AMZN) Up 8%, Apple (AAPL) Down 6%: AWS Grew 37% While Apple Guided Gross Margin Down to 47%
Both companies beat on the top and bottom lines on Thursday evening. Amazon closed the regular session at $235.50 and traded up to roughly $252 after hours, a gain of about 7% at the print and 9% at the peak. Apple closed at $333.43 and traded down to roughly $312, a decline of more than 6%. The divergence had almost nothing to do with the quality of the two beats and almost everything to do with a single input cost that both companies discussed at length: DRAM.
Start by discarding Amazon’s headline. Reported EPS of $5.75 against a consensus of $1.82 is not a triple beat, it is a mark to market. Roughly $53.4 billion of pre-tax non-operating gain came from revaluing the Anthropic stake, which flatters net income to $62.6 billion against $18.2 billion a year ago and tells an investor nothing about the operating business. The number that matters is operating income of $27.46 billion, up 43% year over year, against a revenue base of $200.6 billion that grew 20% and cleared the top of management’s own guidance range. Beneath that, AWS revenue of $42.2 billion grew 36.7% against a street modeling 31%, the fastest cloud growth Amazon has printed since 2021 and an acceleration off the 28% posted in Q1. That is the whole story of the after-hours move.
The capex line is where the reaction becomes interesting. Amazon raised its 2026 capital spending outlook to about $220 billion from $200 billion, having already spent roughly $54 billion on property and equipment in the June quarter alone, with trailing free cash flow at negative $7.6 billion. A month ago that combination would have been sold. Alphabet raised its own capex target the prior week and was punished for it. Amazon raised capex by $20 billion and the stock went up 8%, because Jassy paired the increase with a demand statement rather than a capability statement: even at $220 billion the company expects to be short of capacity through 2026 and into 2027, and he described the 2028 demand book as striking. The market has quietly repriced hyperscaler capex from a margin risk to a revenue signal, but only for operators who can show the bookings behind it. That distinction is now the entire hyperscaler trade.
Apple’s beat was also inflated, in the other direction. Revenue of $109.42 billion set a June-quarter record on 16% growth, iPhone revenue of $54.25 billion grew 22%, and Mac revenue of $10.35 billion grew 29% against a street looking for $8.74 billion. EPS of $2.02 included about $0.11 from tariff refunds following the Supreme Court’s February ruling, and gross margin of 50.1% included roughly two points from the same source. Strip the refund and the beat is about two cents. Services revenue of $30.74 billion missed the $31.22 billion consensus, Greater China came in at $18.8 billion against $19.6 billion expected, and iPad revenue fell 6% year over year. Two of the three segments the market watches most closely came up short, and the one that grew fastest, hardware, is the one carrying the cost problem.
The September quarter guide is what broke the stock. Revenue growth of 9% to 11% implies roughly $113 billion at the midpoint against a street at $114.8 billion, and that range already absorbs 2.5 points of foreign exchange headwind. Gross margin is guided to 47% to 48% including about one point of remaining tariff benefit, which puts the underlying figure near 46.5% against an adjusted 48.1% in the June quarter. Parekh was unusually direct about the cause. More than 100% of the sequential margin decline from March to June is explained by memory cost, with other factors partially offsetting, and the step down into September is driven by the same thing. Cook called it a hundred-year flood in memory pricing and said supply constraints will increase significantly across iPhone, Mac and iPad, having already raised prices on Mac and iPad because of it. He has not raised iPhone prices. That is the coiled spring in the September launch.
The same shortage is therefore producing a 37% growth print at one company and a 150 basis point margin compression at the other, which is the cleanest available illustration of where pricing power actually sits in this cycle. AWS rents capacity under contracts that reprice continuously into demand that exceeds supply, so every dollar of memory inflation flows through to the customer and shows up as revenue. Apple sells a fixed-price physical good into a consumer market on an annual launch calendar, so every dollar of memory inflation lands on its own gross margin until the next product cycle allows a price change. Both companies face identical input costs. One converts them into revenue, the other absorbs them as margin.
That reframes the moat question for Apple in a way the last decade has not required. Apple’s competitive advantage has always been assumed to run through supply chain mastery as much as brand, and the demand side of the moat is entirely intact: iPhone growing 22% at a $4.6 trillion market cap, double-digit growth in every geography, switching costs and installed base undiminished. But the supply side has stopped being a source of advantage. Cook noted the DRAM market has three suppliers and said Apple is evaluating all options, which is a remarkable admission from the company that spent twenty years extracting concessions from its vendors. Against a structural oligopoly running at full utilization and prioritizing HBM allocation, Apple’s volume leverage is worth less than it has ever been. Scale gives you first call on capacity, not a lower price. Amazon’s moat, by contrast, is now expressed as the ability to convert capital into contracted revenue faster than competitors can build, with Trainium at a run rate above $20 billion and gigawatt-scale commitments from the largest model labs behind it.
The setups explain the magnitude. Amazon entered the print having fallen seven consecutive sessions for a cumulative 9% loss and up only mid-single digits year to date, with sell-side targets clustered between UBS at $305 and BMO at $360, Goldman and KeyBanc both at $335, and a consensus near $313. Apple entered up roughly 23% on the year, days after briefly crossing a $5 trillion market capitalization as only the second US company to do so. Amazon was priced for a capex disappointment and did not deliver one. Apple was priced for perfection and delivered a supply-constrained guide on the same night its CEO of fifteen years took his final call, with John Ternus assuming the role on September 1.
Base case for Amazon is the stock holding the $250 to $270 range into the September quarter, with AWS growth in the low thirties and operating income landing in the upper half of the $22.5 to $26.5 billion guide. Bull case is $300 to $335 on a second consecutive quarter of AWS acceleration plus visible Trainium revenue disclosure, which would let the street underwrite the $220 billion capex against contracted rather than projected demand. Bear case is a derating of the entire hyperscaler cohort rather than anything Amazon-specific: if depreciation schedules on AI infrastructure come under serious scrutiny, or if a single large model lab restructures a compute commitment, the multiple compression is sector-wide and Amazon trades to $200 regardless of execution.
Base case for Apple is $300 to $330 while the market waits for the September launch and the first data on iPhone pricing. Bull case is $360 or better if Apple raises iPhone prices without volume damage, which would prove the memory cost is passable through to the consumer and reset the margin trajectory in a single quarter. Bear case is $270 on a cohort derating of the megacap complex combined with a second consecutive Services miss, at which point the argument that Apple deserves a software multiple on a hardware cost structure gets much harder to make in an environment where its bill of materials is set by three companies in Asia.
The single figure worth carrying forward is Parekh’s note that the carry-in inventory currently offsetting some of the memory cost is expected to decline beyond September. Apple’s 47% to 48% guide is built on components bought before the flood. December is the first quarter where the margin is set entirely at current prices.