Nvidia achieved a revenue of 96.2 billion dollars in the second fiscal quarter, double that of a year earlier. The data center business grew by 117 percent to 89 billion dollars. The stock rose six percent a day after the numbers, while the chip manufacturer warned of declining profit margins.
Data Center Division Drives Record Quarter
The data center business remains the most important growth driver and now accounts for 93 percent of total revenue. According to the official statement from Nvidia, revenue in this segment increased by 117 percent over the year to 89 billion dollars. Analysts had expected an average total revenue of around 92.3 billion dollars, which Nvidia significantly exceeded. Meanwhile, the gaming division grew more slowly, reaching 7.2 billion dollars, an increase of 27 percent. The gross margin was 75 percent, both under US GAAP and in the adjusted calculation. The adjusted earnings per share rose to 2.22 dollars, while under US GAAP it was 2.46 dollars per share, in both cases more than double that of the previous year’s quarter. Nvidia also paid out around 26 billion dollars in the form of stock buybacks and dividends to shareholders during the quarter. CEO Jensen Huang explains the leap with a turning point in technology: computing power now delivers productive, profitable results, which is why companies are willing to pay more for it. Demand continues to grow but is being hampered by limited supply.
Forecast Drives Stock Price Up
For the current third fiscal quarter, Nvidia anticipates a revenue of 108 billion dollars, an increase of around twelve percent compared to the recently completed quarter. The company estimates operating costs for the current quarter at around 9.2 billion dollars under US GAAP. The stock closed on August 26, 2026, the day of the announcement, in regular trading down about one and a half percent before the numbers were released. After hours, the price turned and rose by more than four percent at times, as Yahoo Finance reports. The positive reaction continued the following day: the stock jumped six percent in regular trading, and the shares of suppliers AMD and Intel also rose. Nvidia shares had previously fallen in six of the last eight quarters after the earnings report, most recently four times in a row – so this significantly more positive reaction surprises observers. The price fluctuations in chip stocks have often varied independently of their own business figures for weeks, as a summary of the recent AI stock week shows.
Rising Memory Prices Burden Future Costs
The outlook for declining profit margins has a concrete reason: memory chips, especially the high-performance memory HBM used in AI accelerators, are currently among the scarcest components in the entire technology industry. Nvidia expects the gross margin, currently at 75 percent, to drop to 71 to 72 percent in the fourth fiscal quarter before stabilizing at 72 to 73 percent in 2028. The company is trying to offset the higher component costs through its own price increases. According to a report from market researcher TrendForce, Nvidia has already informed some customers that complete AI server systems like Vera Rubin and Grace Blackwell will become more than 15 percent more expensive starting in early 2027. For the same year, analysts expect an increase in HBM contract prices of more than 50 percent compared to 2026. UBS even estimates a rise of 79 percent – independently unverified, as both estimates are based on market forecasts, not manufacturer statements. For companies that purchase or lease their own AI infrastructure, this is likely to result in higher acquisition or cloud costs in the medium term. The billion-dollar investment plans for new data centers, which Nvidia had just announced in August, are also likely to become more expensive as a result.
It will be crucial whether Nvidia can actually pass on the higher memory costs to customers without cloud providers reconsidering their previously accelerated investment plans. The first Vera Rubin systems are expected to be delivered at the new, higher prices as early as early 2027 – a first test of whether the calculation works for Nvidia.
