Micron Shares Fall 4% After Google's TurboQuant Compression Algorithms Launch
Google's TurboQuant and PolarQuant compression algorithms reduce large language model memory needs by as much as 8x, triggering about a 4% drop in Micron shares. Analyst KC Rajkumar disputes the 8x claim versus current 4-bit models, maintains $700 price target and forecasts sustained memory demand over the next five years.
1. Google Launches New Compression Algorithms
Google introduced TurboQuant and PolarQuant algorithms designed to compress large language model data down from traditional 32-bit formats to as little as 3-bit, aiming to cut memory and hardware requirements for AI inference and vector search applications.
2. Micron Shares React to Memory Shift
Micron shares slid approximately 4% in early trading as investors weighed the impact of lower per-model memory consumption on future DRAM and flash chip demand, raising questions about near-term hardware revenues.
3. Analyst Maintains Bullish Outlook
Lynx Equity Strategies analyst KC Rajkumar argues the claimed 8x improvement benchmarks outdated models, retains a $700 price target and buy rating, and projects continued memory demand growth over the next five years due to supply constraints.