Nvidia, the leader in artificial intelligence chips, set another record for both quarterly revenue and operating profit, extending its streak of record revenue to 13 consecutive quarters and reaffirming the intensity of global AI infrastructure spending. The company said the memory supply crunch will persist through 2028, a forecast that points to a prolonged stretch of earnings gains and pricing leverage for Samsung Electronics (005930.KS) and SK hynix (000660.KS), the two companies that dominate the global memory market.
Nvidia said on the 26th that revenue and operating profit for the second quarter of fiscal 2027, covering May through July 2026, reached $96.22 billion, or about 133.2 trillion won, and $63.734 billion, or about 88.34 trillion won, respectively. Both figures were up 106% and 124% from a year earlier. Compared with the previous quarter, revenue rose 18% and operating profit gained 19%.
The data center segment, which includes AI chips, generated $89 billion and drove the overall results. Sales channels have broadened beyond the hyperscalers that make up Nvidia's core customer base to include government-led sovereign AI projects, general enterprises and neoclouds such as CoreWeave.
The only constraint on Nvidia's growth is the memory supply shortage. Chief Executive Jensen Huang said AI demand is close to 100% but that the company set next year's revenue growth rate at 70% based on its own supply volume, adding that the growth rate would have been far higher without the supply constraints. Chief Financial Officer Colette Kress said the company is seeing extreme pricing in memory and that the shortage will act as a bottleneck to growth at least through the end of fiscal 2028, or January 2028.
One factor behind rising memory prices is Nvidia's aggressive advance buying. The company's component supply commitments more than doubled to $279 billion, or about 385 trillion won, from $119 billion, or about 164 trillion won, in the previous quarter. Because of the enormous cost of securing components, the company expects its margin rate to bottom out at 71% to 72% in the fourth quarter before recovering to 72% to 73% in the first quarter of next year, when product price increases take effect.
With Nvidia securing high-bandwidth memory volumes ahead of rivals, competition among other big tech companies to lock up HBM is expected to intensify further. Revenue-generating capacity per gigawatt at data centers is projected to rise from $18 billion with Hopper to $25 billion with Blackwell and peak at $40 billion with the next-generation Vera Rubin, suggesting that HBM prices will stay strong for an extended period.
As a result, the three major memory makers — Samsung, SK and Micron — are expected to concentrate production capacity on HBM, setting off a chain of shortages that now extends to standard server DRAM and enterprise solid-state drives.
With a firmly supplier-favorable market in place, medium- and long-term earnings expectations for Samsung Electronics and SK hynix are climbing by the day. According to financial data provider FnGuide, the consensus estimate for Samsung Electronics' operating profit this year is 394.4153 trillion won, up 804.6% from last year, with revenue of 741.1692 trillion won, up 122.2%. SK hynix's operating profit is estimated at 266.264 trillion won, a 464% jump from last year. In 2028, when the supply crunch is projected to peak, Samsung Electronics is forecast to post operating profit of 563.3759 trillion won and SK hynix 414.7592 trillion won.
Nvidia's graphics processing units, meanwhile, are expanding beyond AI computation into semiconductor manufacturing processes. Nvidia said the same day that Samsung Electronics had adopted cuLitho, which runs on the company's GPUs, and improved simulation performance for chip lithography by up to 20 times. By applying GPUs to the process of calculating light diffusion before circuits are etched onto wafers, the technology is seen as contributing to manufacturing optimization.
