Nvidia CEO Confirms Vera Chip Will Use SK Hynix Memory, Flags Very Large H2 2026

The Quick Take

• Nvidia CEO Jensen Huang confirmed the Vera chip will use SK Hynix memory, deepening the strategic AI partnership
• Huang signals Nvidia is preparing for a 'very, very large' second half of 2026 and into 2027 for memory demand
• The partnership cements SK Hynix's position as the dominant HBM memory supplier for Nvidia's AI chip roadmap

Nvidia's CEO Jensen Huang has confirmed that the company's upcoming Vera chip will use SK Hynix memory, delivering a direct endorsement of the Korean chipmaker's position as the dominant supplier of high-bandwidth memory for Nvidia's AI semiconductor roadmap. The CEO's statement that Nvidia had a 'very big year' with SK Hynix and is 'preparing for a very, very large second half of the year and next year' represents some of the clearest forward demand guidance Nvidia has provided publicly, signaling that AI chip volumes remain on a steep upward trajectory heading into the second half of 2026.

The market implications for SK Hynix are straightforwardly positive: a confirmed Nvidia partnership for its next-generation chip locks in revenue visibility and validates the company's heavy investment in HBM3e and HBM4 memory production capacity. For Micron, the confirmation cements the competitive gap — while Micron is a qualified Nvidia supplier, SK Hynix commands the dominant share of the most premium high-bandwidth memory contracts. Samsung Electronics' memory division faces the most acute competitive pressure as the industry's leadership hierarchy shifts toward SK Hynix's AI memory specialization at the expense of Samsung's historically dominant overall DRAM market share.

The critical forward signal is SK Hynix's Q3 2026 earnings report, where the Nvidia partnership should begin converting into visible revenue acceleration and margin expansion. Nvidia's next quarterly guidance will confirm whether the 'very, very large H2' is materializing as promised. The macro variable determining whether this partnership continues to compound is the pace of hyperscaler AI capex: if Amazon, Microsoft, Google, and Meta maintain or accelerate data center AI infrastructure spending, Nvidia's chip volumes — and by extension SK Hynix's HBM orders — will sustain at elevated levels through 2027 and create a durable revenue moat for Korea's dominant memory chipmaker.

Synthesized from 1 source.