Samsung, SK Hynix to Benefit as Nvidia AI Server Costs Surge
According to reports, Samsung Electronics and SK hynix enhanced their "bargaining power" within the artificial intelligence supply chain as increasing memory costs influenced prices for server systems utilizing Nvidia technology.
Server makers informed key data center clients, such as Microsoft, Google, and Oracle, that costs for Nvidia-powered AI systems would increase by over 15 percent in numerous instances for deliveries commencing in early 2027. The price changes impacted systems based on Nvidia's upcoming Vera Rubin and existing Grace Blackwell platforms, with the specific extent of the increase differing by chip generation and memory setup.
Memory components became a key factor in the cost pressure. Nvidia's AI accelerators relied on high-bandwidth memory (HBM), whereas AI data centers utilized significant amounts of standard server DRAM. Supply stayed constrained as cloud service providers increased computing capacity and obtained memory supplies via long-term contracts, giving a limited number of memory producers a better negotiating leverage.
According to data from Counterpoint Research, the report indicated that SK hynix held 58 percent of global HBM revenue in the first quarter. In the second quarter, Samsung dominated the overall DRAM market with a 39 percent share, while both South Korean chip manufacturers together represented 65 percent of worldwide revenue.
Also Read: SpaceX, Nvidia Develop AI System for Space
SK hynix gained a significant advantage in HBM linked to AI accelerators, while Samsung held a broader presence in traditional DRAM and increased its stake in advanced HBM offerings. Counterpoint indicated that most of the HBM revenue in the first quarter originated from HBM3E, while HBM4 shipments are anticipated to gain significance in the latter half of the year. Cost challenges also affected the broader DRAM industry.
The report stated that TrendForce expected server DRAM contract prices to increase by another 13 to 18 percent in the third quarter relative to the previous quarter, fueled by ongoing AI-related demand for manufacturing capacity.
Also Read: Sonova Unveils New AI-Powered Hearing Aid
According to the report, South Korean investment firm Hana Securities observed that the announced price hikes for Nvidia-based servers have redirected focus onto memory suppliers, highlighting Samsung and SK hynix as key beneficiaries due to ongoing infrastructure expenditures expected to continue until 2027.
The need for high-performance memory continued to be varied, as firms like Amazon, Microsoft, Google, and Meta created custom AI processors that also needed advanced DRAM and HBM. Nevertheless, rising costs of climbing parts increased total infrastructure expenses.
According to reports, server providers alerted clients about impending price increases of around 17 percent on key Nvidia-based systems, predicting that chip-system expenses for a 1-gigawatt data center might increase by a minimum of $5 billion.
Shin Joong-ho, head of research at LS Securities, noted the limitations posed by increasing component costs on the long-term effects for technology companies.
Also Read: New Zealand Plans Bill to Ban Social Media for Children Under 16
According to reports, Shin stated that Big Tech can handle increased expenses temporarily, but there is a boundary to how much infrastructure investment can increase without impacting returns.
If memory inflation continues to elevate Nvidia system costs, businesses might need to reevaluate the anticipated returns on new AI data center initiatives.