NVIDIA's Strict Audits: Nitpicking Suppliers for Lucrative AI Chip Deals

Reports have shed light on an unusual strategy employed by NVIDIA in its negotiations with semiconductor suppliers.

According to a South Korean publication, NVIDIA is utilizing a tactic dubbed "proactive pressure" to secure more advantageous terms for its contracts. The market for AI system components is notoriously complex, with major players like NVIDIA and AMD fiercely competing for critical resources, including High Bandwidth Memory (HBM). For several years, this segment has largely favored buyers, allowing large customers to lock in production capacities and optimal pricing.

Journalists indicate that an NVIDIA inspection team recently visited Samsung Electronics' Pyeongtaek manufacturing complex, one of the company's most advanced semiconductor facilities, particularly known for its HBM development. During this visit, the team conducted an internal audit, adopting an exceptionally rigorous approach to reviewing technological processes. Sources close to the matter revealed that NVIDIA engineers raised highly detailed objections, criticizing even minor technical aspects. It was noted that validation requirements were significantly more stringent than during previous inspections.

Such meticulous scrutiny is believed to be a calculated part of NVIDIA's broader negotiation strategy. By identifying even minor vulnerabilities in a supplier's production capabilities, the company gains significant leverage ahead of crucial contract cost discussions. This recent visit is thought to be specifically linked to negotiations for HBM4 memory supplies, intended for the mass production of NVIDIA's upcoming Vera Rubin architecture. A deal of this magnitude would represent a critical milestone for Samsung Electronics, especially considering its previously weaker position in the competitive HBM market.

Sources further suggest that similarly strict audits have been conducted with other key partners, including SK hynix and TSMC. For these major suppliers, such an intense level of control from NVIDIA is reportedly no longer an unexpected occurrence.