Micron is going to $4,000 and once you understand what inference actually is, the number stops sounding crazy (Save this…
Milk Road AI Twitter · Milk Road AI (@MilkRoadAI) · 2026-07-01
Milk Road AI makes a bullish investment case for Micron stock reaching $4,000 by arguing that AI inference is fundamentally memory-bandwidth-constrained rather than compute-constrained, making Micron the upstream bottleneck on all AI value generation.
Appears in
Extraction
Topics: ai-inferencehbm-memorysemiconductor-stocksai-infrastructuremicron
Claims
- AI inference decode phase leaves GPUs idle more than 95% of the time waiting for data from memory, making inference memory-bound rather than compute-bound.
- OpenAI and Anthropic alone will need over 100 gigawatts of compute combined by 2030, with AI infrastructure potentially measured in terawatts by 2040.
- Adding more GPUs does not fix inference bottlenecks because GPU utilization is already low; adding memory bandwidth directly reduces token cost and latency.
- Longer context windows compound memory demand, with a 1 million token context requiring dramatically more memory per session than a 10,000 token context.
- Micron trades at 8x forward earnings on projected FY2027 EPS of $112 and has HBM4 ramping at twice the pace of the prior generation, making it the most undervalued AI infrastructure company.
Key quotes
The GPU is not the bottleneck but the memory feeding the GPU is.
Micron is the upstream constraint on how much value every Nvidia GPU can actually generate at inference scale.
Inference is memory. Memory is Micron and the inference ramp has barely started.