NVIDIA and SK Hynix to Announce Partnership Plan on Monday
N.R. Finch
Nvidia CEO Jensen Huang said Sunday in Seoul that the company's new Vera CPU will use SK Hynix DRAM, extending their partnership from HBM into CPU-level memory — a signal that Nvidia's supply-chain dependency on SK Hynix is shifting from a single component to a full-platform tie.
Why does the Vera CPU matter?
Vera is Nvidia's first standalone data-center CPU, competing directly with Intel Xeon, AMD EPYC, and Amazon Graviton.
This means → Nvidia is no longer just a GPU company; it aims to own both the compute and memory lines inside data centers.
Huang named SK Hynix as Vera's DRAM supplier, extending the relationship beyond HBM — high-bandwidth memory, a high-speed stacked memory built specifically for AI chips — into general-purpose processor storage.
How has SK Hynix's role changed?
Previously SK Hynix mainly supplied HBM to Nvidia; now Vera locks in its DRAM too, significantly deepening the supply-chain bond.
In plain terms = SK Hynix has gone from being Nvidia's "specialty parts vendor" to its full-line memory partner.
SK Group Chairman Chey Tae-won and Huang plan to brief the press Monday morning; the two already dined together with SK Hynix CEO Kwak Noh-jung to confirm the direction.
How long will the chip shortage last?
Huang stated clearly: from wafers to packaging to silicon photonics — a chip technology that moves signals with light instead of electricity — every link in the supply chain is short.
He expects the shortage to persist for "quite a number of years."
This reflects a reality where AI compute demand is growing far faster than semiconductor capacity can expand.
What comes next?
Both sides expect significant business growth from H2 this year through 2027; Huang said they are "preparing for large-scale collaboration."
Whether SK Hynix can scale capacity to match Nvidia's expansion pace will be the key test of this partnership's delivery.
Huang also revealed Nvidia is in talks with telecom operators about the future role of telecom networks in AI deployment — he believes AI will expand outward from centralized data centers.
Content is for reference only, not financial advice.