Skip to content
Briefings are running a touch slower this week while we rebuild the foundations.See roadmap
SK Hynix
OrganisationKR

SK Hynix

South Korea's second-largest memory chipmaker; a key global supplier of AI-critical HBM to Nvidia.

Last refreshed: 2 May 2026 · Appears in 1 active topic

Key Question

Will the AI chip boom shield SK Hynix when geopolitical shocks slam Korea's market?

Timeline for SK Hynix

View full timeline →
Common Questions
What is SK Hynix?
SK Hynix is a South Korean semiconductor company and the world's second-largest memory chipmaker, producing DRAM, NAND flash, and high-bandwidth memory (HBM) chips used in AI accelerators. It is part of the SK Group conglomerate.
Why did SK Hynix shares fall 12% in 2026?
During the Iran conflict in 2026, South Korea's KOSPI fell more than 8% in a single session, triggering a circuit breaker. SK Hynix dropped 12.3%, as investors feared geopolitical escalation would curb global technology demand.Source: KOSPI circuit breaker event
Does SK Hynix supply HBM chips to Nvidia?
Yes. SK Hynix is Nvidia's primary supplier of high-bandwidth memory (HBM) for AI chips, including the H100 and Blackwell series used in data centres globally.
What is the difference between SK Hynix and Samsung Electronics in memory chips?
Both are South Korean memory chipmakers; Samsung is the largest globally and SK Hynix the second-largest. SK Hynix has pulled ahead in HBM for AI chips, a segment where Samsung has faced yield challenges.
Is SK Hynix affected by US chip export controls on China?
Yes. SK Hynix operates fabrication plants in Wuxi and Dalian, China. US export controls restrict advanced chip manufacturing equipment for those sites, requiring waivers and limiting their upgrade PATH.

Background

SK Hynix's role in the AI jobs and power story is upstream of everything: as Nvidia's primary supplier of high-bandwidth memory (HBM) chips, its production capacity determines how fast AI data centres can be built, which in turn sets the ceiling on how rapidly AI-driven productivity gains — and AI-driven job displacement — can accelerate. Alphabet's Q1 2026 earnings call, on which CEO Sundar Pichai said Google was 'compute constrained in the near term', implicitly named HBM supply as one of the binding bottlenecks.

SK Hynix operates within a small consortium: TSMC handles the advanced logic dies at the heart of AI accelerators; SK Hynix and Samsung supply the HBM stacks that give those accelerators their memory bandwidth. The three companies form an irreplaceable supply chain for frontier AI hardware, and collective pricing power within that chain is a direct determinant of the unit economics of AI deployment — and therefore of how aggressively companies can substitute AI for human labour at any given cost point.

For the AI jobs story specifically, SK Hynix's supply constraints function as a natural brake on displacement speed: if HBM production cannot keep pace with AI accelerator demand, the rate of AI deployment in enterprise applications is physically limited regardless of software capability or corporate intent. The memory supply chain is not discussed in most AI employment analyses, but it is a genuine governor on the pace at which AI replaces human workers.