One of the biggest bottlenecks in the AI chip market has become high-bandwidth memory.
Reuters reported September 10 that Chinese AI-chip makers including Huawei and Cambricon are raising accelerator prices because high-bandwidth memory has become increasingly scarce. Huawei’s upcoming Ascend 950DT has reportedly been quoted at more than 250,000 yuan, representing increases of roughly 20% to 50% compared with earlier pricing.
China faces additional pressure from export restrictions, but the underlying issue is global. AI accelerators require enormous amounts of high-performance memory, making Micron Technology, Inc. (NASDAQ:MU) one of the key beneficiaries.
Micron Is Selling the Bottleneck
Micron Technology, Inc. is not merely supplying generic DRAM into this market. Its HBM4 is already in high-volume production for NVIDIA Corporation’s (NASDAQ:NVDA) Vera Rubin platform. Micron says the product delivers more than 2.8 TB/s of bandwidth per stack and improved power efficiency.
Micron’s bull case is straightforward. Increasing memory content per accelerator and tight supply give it unusually strong pricing power. The company has also signed long-term agreements designed to reduce the cyclicality that historically plagued memory manufacturers.
The bear case is that memory remains memory. High prices incentivize new capacity, technological substitution and customer efforts to diversify suppliers. The stronger today’s margins become, the larger the incentive for competitors to expand production.
Nvidia Has to Share More of the Economics
NVIDIA Corporation sits on the opposite side of part of this equation. It benefits enormously from AI demand but increasingly has to share system economics with memory, networking, foundry and packaging suppliers. Higher HBM costs can pressure margins or increase system prices.
The bull argument for Nvidia is that customers remain willing to pay because GPU performance is economically valuable enough to justify expensive supporting components. The bear argument is that persistent supply inflation reduces how much of each AI server’s total value Nvidia can retain.
Hedge-fund positioning improved for both stocks during Q2. Micron ownership jumped to 184 funds from 154, while Nvidia increased to 285 from 275. Short interest was approximately 2.7% of Micron’s float and 1.2% of Nvidia’s as of August 14.
The valuation question is essentially whether Micron’s current earnings are cyclical peak earnings or the beginning of a structurally better memory regime. Nvidia’s question is whether margins can remain exceptional as suppliers capture more economics.
Memory scarcity is one of the rare AI trends where Nvidia can simultaneously benefit from stronger end demand and lose a little bargaining power to a supplier.
READ NEXT:
The Trillion-Dollar AI Capex Gap: Why Too Much Hardware Could Be Nvidia’s Trap and Microsoft’s Opportunity and Redditors Are Skeptical of SpaceX’s Orbital Data Centers. Are They Right?
Hedge Fund Resource Center