Micron surged 16% after crushing earnings estimates, while Nvidia shares moved in the opposite direction — a sign that AI infrastructure investors are becoming more selective about which chipmakers will benefit from the spending boom.
Micron surged 16% after crushing earnings estimates, while Nvidia shares moved in the opposite direction — a sign that AI infrastructure investors are becoming more selective about which chipmakers will benefit from the spending boom.

Micron surged 16% after crushing earnings estimates, while Nvidia shares moved in the opposite direction — a sign that AI infrastructure investors are becoming more selective about which chipmakers will benefit from the spending boom.
Micron Technology Inc. jumped 16% after reporting $41.5 billion in quarterly revenue, beating estimates by nearly $6 billion, while Nvidia Corp. shares fell despite its own record results.
"There is no line of sight to when supply will catch up with demand," Micron Chief Executive Officer Sanjay Mehrotra said, adding that AI memory shortages are expected to persist beyond 2027.
Micron's fiscal third-quarter earnings per share reached $25.11, surpassing the $20.49 consensus. Revenue of $41.5 billion beat the $35.7 billion average estimate. The company forecast fourth-quarter revenue of about $50 billion, well above the $43.2 billion analysts had projected. SanDisk Corp. and SK Hynix Inc. each rose about 13% in sympathy.
The divergent market response signals that investors are becoming more selective about AI infrastructure beneficiaries. Memory makers are capturing the immediate demand surge from AI model training, which requires vast amounts of high-bandwidth memory. Nvidia, meanwhile, faces growing competition from in-house chip designs at major cloud providers and expectations that make each successive earnings beat harder to reward.
Memory Demand Accelerates Beyond 2027
High-bandwidth memory has become the backbone of AI infrastructure, essential for training and running large language models. Micron's guidance implies sequential revenue growth of about 20% in the current quarter. Mehrotra highlighted robotics as a significant long-term growth driver, noting that humanoid robots are expected to require 10 times more memory than advanced driver-assistance vehicles and could drive a sustained, multi-decade demand cycle beginning in the latter part of this decade.
The memory shortage spans the entire HBM supply chain. SK Hynix, which leads the high-bandwidth memory market, is reportedly exploring a US listing that could value the company at around $30 billion. The tight supply conditions have given memory makers unusual pricing power in a historically cyclical industry where oversupply has often crushed margins.
Nvidia's Blackwell Ramp Faces Rising Expectations
Nvidia's data center business continues to grow at triple-digit rates, with its next-generation Blackwell architecture ramping into production on TSMC's 4nm process node (which improves transistor density and power efficiency per watt). But hyperscalers including Amazon.com Inc. and Alphabet Inc. are accelerating custom chip programs, potentially eroding Nvidia's pricing power over time.
The company's premium valuation leaves limited room for upside surprises. Each successive earnings report must clear a higher bar, making the stock more vulnerable to profit-taking even when results are strong. The divergence with Micron's rally suggests the market is beginning to differentiate between AI infrastructure plays that are capturing incremental demand and those that may face margin compression as competition intensifies.
Investment Angle: Micron trades at about 16 times forward earnings, a discount to Nvidia's multiple, reflecting the memory industry's historically cyclical nature. If AI-driven memory demand proves structurally higher — as Mehrotra's "no line of sight" comment suggests — that discount could narrow. For Nvidia, the risk is that each earnings beat gets priced in faster, making the stock more vulnerable to any sign of deceleration. Morgan Stanley's Joseph Moore has maintained his overweight rating on Nvidia, calling competitive threats from custom chips "years away from material revenue impact."
This article is for informational purposes only and does not constitute investment advice.