Morgan Stanley warns that chip prices for memory will stay elevated as AI demand outpaces supply for years to come, since building new manufacturing capacity takes time. Rather than a threat, the investment bank views this as a new phase of sustainable growth for AI infrastructure—hyperscalers will continue spending heavily, favoring suppliers like Micron and Broadcom that have proven competitive advantages and long-term customer relationships.
Summaries like this, in your inbox every morning.
Sign up free →What happened
Morgan Stanley's research team describes 'chipflation' as a phenomenon where memory chip prices rise sharply and remain elevated as demand persistently exceeds supply. The firm notes that demand for GPUs, high-bandwidth memory (HBM), and advanced DRAM should continue to outstrip supply because model sizes keep growing and inference workloads are expanding rapidly.
Why it matters
Hyperscalers building AI infrastructure face a choice between passing higher chip costs to customers or accepting reduced profit margins. However, Morgan Stanley views this as a transition toward a 'durable supply-demand reset' rather than a bearish scenario—data center build-outs and AI-enhanced device development are unlikely to stop, though the pace may begin to plateau. This creates sustained demand for memory suppliers positioned to capture value.
What to watch
Morgan Stanley identifies Micron Technology and Broadcom as well-positioned to benefit. Micron, as a leading producer of HBM and DRAM, sits at the center of AI chip supply chains and benefits from long-term supply agreements. Broadcom's networking switches and custom silicon designs for hyperscalers like Alphabet, Apple, and Meta Platforms offer complementary exposure to the same infrastructure expansion.
Morgan Stanley's analysis reframes a supply crisis as a structural feature of the AI infrastructure build-out rather than a temporary bottleneck. The core insight is that memory chip supply cannot keep pace with demand because each new generation of generative AI models and broader inference deployment requires greater volumes of specialized memory, while building new chip fabrication plants takes years. This creates a multi-year window where prices stay elevated and supply remains constrained.
The investment bank's recommendation to buy AI infrastructure stocks on near-term weakness reflects confidence that hyperscalers will continue their capital expenditures despite higher chip costs. The firm suggests investors should distinguish between the transition phase (where focus shifts from deployment pace to utilization rates and returns on capital invested) and the long-term opportunity. Two stocks emerge as beneficiaries: Micron, protected by long-term supply agreements and its central position in HBM/DRAM supply chains; and Broadcom, which diversifies exposure through networking infrastructure and custom silicon work with major hyperscalers. This positioning implies that the winners in AI infrastructure will not be those most exposed to raw chip price volatility, but rather those with durable customer relationships and specialized capabilities that hyperscalers cannot easily replace.
No discussion yet for this article
Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.
Get Started FreeFree · takes 30 seconds · unsubscribe anytime
1 minute a day. The AI essentials.
200+ sources · Email / LINE / Slack
