A Google AI research paper claiming to reduce AI memory usage by one-sixth has led to a $90 billion market value drop in global memory chip stocks, including Micron and SanDisk. The paper, which introduced the TurboQuant algorithm, was accused of experimental bias by Gao Jianyang, a researcher at ETH Zurich. Gao's open letter alleged that Google's team used unfair testing conditions, favoring their algorithm over Gao's RaBitQ, and failed to properly credit RaBitQ's methodology. The controversy intensified as Gao's letter gained traction, with over 4 million reads on Zhihu and support from the Stanford NLP Group. The market reacted sharply, fearing a structural downgrade in memory chip demand due to TurboQuant's potential to reduce AI inference memory needs. Analysts, however, argued that the impact might be overstated, as TurboQuant targets specific cache types rather than overall memory usage. Google has yet to formally respond to the allegations.