NVIDIA CFO Collette Kress blames other firms for suffering in the ongoing memory shortage. In an interview with Tae Kim, Kress shared how her firm had ordered memory in advance as it understood that a price surge was imminent. The surge in memory prices due to high demand from AI chip companies has transformed the industry, with workers at companies such as SK hynix receiving hefty bonuses and those at Samsung protesting.
Others Should Have Ordered Memory A Long Time Ago, Says NVIDIA CFO
The high-performance requirements of AI GPUs have led to significant turmoil in both the high bandwidth memory (HBM) and DDR memory markets. Estimates suggest that the demand for NVIDIA's Rubin AI platform alone is expected to exceed the memory demand of Apple and Samsung combined. Specifically, the Rubin AI chips can require as much as a whopping six billion gigabytes of LPDDR memory in 2027, while Apple is expected to require 2.9 billion gigabytes and Samsung 2.7 billion GBs.
Not only do AI GPUs also require DDR memory in addition to HBM memory, but since both are produced with the same machines, a bump in demand for HBM means that manufacturers allocate capacity away from DDR. The capacity allocation leads to shortages in the DDR market as well, with Chinese firms now believed to be eyeing this opportunity to establish themselves as key players in the market.
NVIDIA Knew Beforehand That Memory Prices Would Go Up, Says Kress
Amidst this supply chain turmoil, NVIDIA has managed to stay ahead of others and ensure a stable supply of the crucial memory chips. The firm's CFO, Collette Kress, discussed her company's approach to the memory shortage in an interview with Tae Kim. Kress outlined that her company had foreseen the disruption in the market.
Discussing how other firms were complaining about the high prices, Kress remarked that many companies "are sitting here going, oh my gosh, the memory price went up." She added that her firm "knew that was going to happen," and it was "something everybody should have, at least we did, ordered a long time ago." Additionally, she also revealed that NVIDIA was working with its memory suppliers to build the chips on order instead of buying products available on the shelf.
According to her:
They’re designing it with us. And then they go, now how much supply do we need? And we’re not just doing it with one. We’re doing it with all three memory suppliers. We say, here’s what we’re building. And then we’ve got to get them all in line and working with us. I don’t see another company doing that.
About the author: Ramish is a seasoned technology writer and editor with more than a decade of experience. He specializes in semiconductor fabrication and market analysis. With a background in finance and supply chain management - via his bachelors in Finance and a micromasters in supply chain management from MIT - Ramish combines financial rigor with deep industry insight to deliver accurate and authoritative coverage.
Follow Wccftech on Google to get more of our news coverage in your feeds.
