Nvidia’s Revenue Outlook Beats Consensus as Margin Pressure Builds
Nvidia forecast fiscal third-quarter revenue above Wall Street’s consensus, but its gross-margin guidance fell short of expectations, shifting investor attention from demand to the rising cost of supplying AI systems. Bloomberg Intelligence semiconductor analyst Kunjan Sobhani argues that planned price increases are largely intended to offset higher wafer and memory costs, while Bloomberg Tech’s Ed Ludlow says Nvidia has yet to show how those costs and higher system prices will be divided across the company, server makers and customers.
Revenue guidance beat consensus, but the margin outlook points to rising costs
Nvidia forecast fiscal third-quarter revenue of $105.84 billion to $110.16 billion, versus a $105.15 billion consensus estimate. The range implies a midpoint of roughly $108 billion, though some analyst projections had exceeded $110 billion. Second-quarter revenue was $96.2 billion, ahead of a $92.38 billion estimate.
The more consequential question is whether Nvidia can retain its historic profitability as the cost of building AI systems rises. Kunjan Sobhani pointed to a slight shortfall on gross margin as a likely focus for investors. Nvidia reported adjusted gross margin of 75% for the completed quarter, in line with expectations, but guided to 74%, plus or minus 50 basis points, for fiscal third quarter—73.5% to 74.5%, against the 74.8% consensus cited by Ed Ludlow.
Sobhani said price increases are intended to protect a 75% gross margin rather than necessarily enlarge it. The significant test, he said, will come as the Vera Rubin ramp begins in full during the fourth fiscal quarter. If rising input costs continue to outrun the company’s ability to pass them along, a price increase can limit margin erosion without restoring the prior level.
Ed Ludlow said Bloomberg had reported that server makers were informing customers that systems shipped from January 2027, including Blackwell- and Vera Rubin-based systems, would face price increases of 15% or more. But neither that reported increase nor Nvidia’s revenue guide establishes how much of the higher system price Nvidia will keep.
Ludlow said the earnings release did not explain the immediate margin effect of higher memory costs or the competition to secure fabrication capacity. Nvidia also disclosed that commitments had risen to $279 billion from $119 billion quarter over quarter. The release, Ludlow said, did not explain the contemporaneous impact of those conditions on margin.
The uncertainty also informs the stock’s role in the market, according to Sobhani. Nvidia shares were up about 14% in 2026 after a seven-day losing streak, a more muted performance than in prior periods. Nvidia is “the overall AI trade right now,” he said: investors’ concerns about whether AI spending will continue or accelerate into 2028 are being expressed through the shares. He described the current and following quarters as a “bubble gap,” in which investors remain on the sidelines until they have more clarity about that longer-term demand.
The price increase travels through a chain of costs and intermediaries
The customer buying an AI system generally is not purchasing GPUs directly from Nvidia, Ludlow said. Neocloud providers, hyperscalers, and large software companies operating their own data centers typically buy assembled systems through server makers such as Dell, HPE, and Supermicro. Those companies have traditionally served as the sales channel, even as Nvidia has taken on more of the server design and its underlying content.
That arrangement makes the reported 15%-plus price increase difficult to assign to a single beneficiary. Bloomberg’s reporting was based on server makers telling Nvidia customers that prices were going up. Ludlow’s question was where, and at what point, the increase originates when Nvidia supplies much of the system’s core content while OEMs assemble and sell the finished server.
Kunjan Sobhani framed the answer as a flow through the full bill of materials: foundry production, memory, Nvidia chips, and the assembly work performed by OEMs. Nvidia buys memory from memory suppliers, he said, while wafer costs from TSMC have also risen. But Nvidia has not publicly described how the pricing is allocated. Sobhani’s estimate was that the increase is principally intended to offset higher input costs.
Most of this is going for that input rise from wafer, so to TSMC and the rest to the memory makers.
In Sobhani’s model, higher wafer costs can be passed through the chain, while higher memory costs initially pressure the chipmaker’s economics. A chipmaker seeking to protect its margin could then charge more, leaving OEMs such as Dell to decide whether to raise their own system prices or absorb part of the increase. The same choice exists at other points in the supply chain: any participant can take on some of the cost, but only by accepting pressure on its own economics.
The central issue is therefore not simply whether buyers will pay more for Blackwell or Vera Rubin systems. It is how the added cost is distributed among foundries, memory suppliers, Nvidia, server manufacturers, and end customers. Nvidia’s margin guide suggests that, at least in the near term, higher prices are being asked to carry a growing supply-chain burden.