SK Hynix's Second-Largest Customer Approaches NVIDIA in H1 Sales
Nashnova编辑部
SK Hynix's second-largest customer contributed ₩17.19 trillion in H1 2026 sales, nearly matching Nvidia's ₩17.61 trillion — the HBM giant is shifting from Nvidia-dependence to a multi-customer revenue base.
How close are the top two customers now?
SK Hynix's H1 2026 report shows its top customer — confirmed as Nvidia by Yonhap — contributed ₩17.61 trillion. The second-largest followed at ₩17.19 trillion, a gap of less than ₩420 billion.
Combined, the two accounted for roughly ₩34.8 trillion, or 26.4% of H1 total revenue of ₩131.9 trillion.
This means → SK Hynix's revenue engine has shifted from single-customer dominance to a twin-pillar structure, with the number-two buyer gaining real weight.
Nvidia is buying more — so why is its share falling?
Nvidia's H1 purchases rose 61.7% year-on-year (from ₩10.89 trillion), yet its share of SK Hynix's total revenue dropped from 27.31% to 13.35%.
In plain terms = Nvidia is spending more, but SK Hynix is growing even faster — total revenue leapt from ₩39.87 trillion to ₩131.9 trillion, more than tripling.
This reflects a revenue surge driven not by one customer but by broad-based demand across multiple buyers scaling up simultaneously.
How far along is the diversification push?
A June SEC filing shows Nvidia accounted for 23.9% of SK Hynix's full-year 2025 revenue; by Q1 2026 that had fallen to 14.8%. The second-largest customer stood at 12.4%.
SK Hynix has completed roughly 10 long-term supply agreements, covering core customers, and is still in talks with other major players.
This means → diversification has moved beyond strategy decks into signed contracts — long-term lock-ins that spread revenue risk and stabilize the base.
Will the Nvidia relationship weaken as a result?
No. The two companies signed a multi-year technology partnership in June, covering next-generation memory development and supply alignment.
In plain terms = SK Hynix's playbook is "make new friends without losing the old one" — expanding into hyperscalers while locking in the next-gen tech roadmap with Nvidia.
The agreement explicitly aligns with Nvidia's infrastructure roadmap, ensuring HBM product cycles stay in sync with Nvidia's chip cadence.
What does the revenue mix reveal?
The U.S. market delivered ₩84.56 trillion in H1 revenue, more than tripling year-on-year, and accounting for 64.1% of the total. China rose from ₩7.37 trillion to ₩32.48 trillion, or 24.6%.
By product, DRAM contributed ₩97.64 trillion and NAND ₩33.53 trillion.
This reflects that HBM — high-bandwidth memory, the core companion to AI chips — remains overwhelmingly a U.S.-driven demand story, though China's growth rate is hard to ignore.
What comes next?
SK Hynix began shipping HBM4 in Q2 and plans full-scale ramp in H2. It also delivered HBM4E samples on a 1c-nanometer process to a major customer in H1.
This means → the key test in H2 is whether hyperscalers sustain and expand their purchase volumes, turning "number-two nearly matches Nvidia" from a single-half snapshot into a durable trend.
If the diversification momentum holds, SK Hynix's valuation narrative shifts from "Nvidia supplier" to "AI memory platform company."
Content is for reference only, not financial advice.