Memory module prices plunged sharply starting last week, driven by mass sell-offs from speculative holders who had stockpiled during the previous price surge. One distributor reported a mainstream 16GB memory module dropped by over $14 in a single day.

Retail sales volumes have fallen more than 60% compared to November, as non-essential buyers pulled back during the prior run-up. The combination of weak demand and concentrated selling triggered a sudden price collapse.

Adding to the pressure, Google’s newly unveiled TurboQuant compression algorithm can reduce key-value cache memory usage in large language models by at least 60%. The news rattled memory stocks, erasing nearly $100 billion in market value from U.S. memory chip companies this week.

Micron took the hardest hit among major players, with shares down 15% since last Friday—a loss of more than $70 billion in market capitalization.

ICgoodFind : Panic selling and a new compression algorithm just sent memory prices into freefall. The supply-demand math for AI memory is suddenly looking different.