Nvidia posted another record quarter as the leader in artificial intelligence chips, prompting analysts to raise their earnings expectations for Samsung Electronics and SK hynix. Expanding investment in AI data centers and a prolonged shortage of high-bandwidth memory (HBM) are expected to sustain profit gains at the South Korean memory makers for some time.
Nvidia said on the 26th that revenue reached $96.22 billion, or about 133.2 trillion won, in the second quarter of fiscal 2027, covering May through July 2026, with operating profit of $63.734 billion, or about 88.34 trillion won. Both figures were records, up 106% and 124% respectively from a year earlier. Revenue set a record for the 13th consecutive quarter.
The data center segment, which includes AI chips, drove the results. Revenue in that segment reached $89 billion and led overall growth. The customer base is broadening beyond established hyperscalers to government-led sovereign AI projects, enterprise AI users and neocloud operators.
The constraint on Nvidia's growth is memory supply rather than demand. Chief Executive Jensen Huang said AI demand is close to 100% but supply constraints are limiting the growth rate. Chief Financial Officer Colette Kress said a memory shortage could remain a bottleneck for growth through the end of fiscal 2028.
The company's remarks pointing to a shortage lasting several more years raise the likelihood of price increases at memory makers, led by HBM.
With the memory shortage extending, earnings forecasts for Samsung Electronics (005930) and SK hynix (000660) are being revised upward. Expansion of the AI server market and rising HBM demand are unfolding at the same time, widening the recovery in memory conditions beyond earlier expectations.
According to financial data provider FnGuide, the consensus estimate for Samsung Electronics' operating profit this year has been steadily raised to 394.4153 trillion won. That represents an increase of more than 800% from last year. Revenue is projected at 741.1692 trillion won, also pointing to sharp growth.
SK hynix, seen as the biggest beneficiary of AI memory demand, is also drawing rapidly improving forecasts. Its estimated operating profit for this year stands at 266.264 trillion won, an increase of more than 400% from last year.
Market participants expect the medium- to long-term earnings capacity of Samsung Electronics and SK hynix to strengthen further if AI server investment continues to expand and the HBM shortage persists. In 2028, when the shortage is expected to peak, operating profit could reach 563.3759 trillion won at Samsung Electronics and 414.7592 trillion won at SK hynix.
Nvidia's aggressive push to secure memory is also adding to upward price pressure. Its component supply commitments have more than doubled to $279 billion from $119 billion. That points to intensifying competition to lock up HBM and other key components in order to expand production of AI accelerators.
Samsung Electronics and SK hynix are accordingly concentrating on expanding HBM production capacity. As output shifts toward HBM, the risk of shortages is growing for conventional server DRAM and enterprise solid-state drives (eSSD), pointing to continued price strength across the memory market.
Nvidia's graphics processing units (GPUs), meanwhile, are extending beyond AI computation into semiconductor manufacturing processes. Nvidia said the same day that Samsung Electronics had adopted cuLitho, which runs on its GPUs, to improve simulation performance for chip lithography by up to 20 times. GPUs are being applied to the step that calculates light diffusion before circuits are etched onto wafers, contributing to manufacturing optimization.
