Nvidia's fiscal 2027 second-quarter report delivered a revenue beat and a raised outlook, but the market's focus settled on a single line in the guidance: gross margin is set to drop a full percentage point next quarter, and the company explicitly blamed the rising cost of high-bandwidth memory. That admission reframes the AI capital-expenditure cycle as a transfer of pricing power from the GPU designer to the handful of memory suppliers that can deliver stacked DRAM at volume.
The numbers behind the guide
Revenue reached $96.2 billion, more than double the year-ago quarter and up 18 percent sequentially. The data center segment alone generated $89 billion, exceeding Nvidia's total company revenue from twelve months earlier. Gross margin held at 75 percent, flat quarter over quarter and up 2.6 percentage points year over year. Operating income climbed 19 percent sequentially to $63.7 billion, while net income came in at $59.7 billion, or $2.46 per share. Hyperscale customers contributed $48.7 billion, a 102 percent increase from a year ago. For the third quarter, management guided $108 billion in revenue, plus or minus 2 percent, assuming zero compute sales to China.
Margin compression is a cost story not a demand story
Both GAAP and adjusted gross margin are forecast at 74 percent, plus or minus 50 basis points, a full point below the quarter just reported. The company expects margins to trough between 71 and 72 percent in the fourth quarter before recovering in fiscal 2028. Chief Financial Officer Colette Kress attributed the step-down to component costs, not weakening demand or erosion of Nvidia's own pricing. She disclosed that purchase commitments for critical components surged from $119 billion to $279 billion, driven primarily by memory procurement.
The leverage shifts downstream
High-bandwidth memory has moved from a peripheral input to a co-equal element of the accelerator package. When a buyer of Nvidia's scale and sophistication absorbs a margin hit to secure supply, the economics of the purchase order flow to the vendor. Micron, one of the few manufacturers capable of delivering the stacked memory these platforms require, sits on the receiving end of those commitments. Higher contracted volumes at elevated prices translate directly into higher realized average selling prices for the memory layer.
What to watch next
Micron's next quarterly report will test whether its average selling prices track the step-up implied by Nvidia's commitment jump. The other variable is timing: Nvidia's guide assumes a margin recovery in fiscal 2028, which implies an expectation that memory supply will loosen. If that recovery slips, the margin pressure at Nvidia, and the pricing power at Micron, persists longer than the current forecast assumes.
The structural read
The same hyperscaler capex wave that built Nvidia's franchise is now funding the memory layer beneath it. Leverage in the AI hardware stack has shifted to the supplier with the scarcest capacity. Nvidia's willingness to guide lower margins to lock in HBM volume is the clearest signal yet that the bottleneck has moved downstream.