The RAM industry is facing a turning point following two events that promise to transform the landscape of artificial intelligence, a sector that has driven demand over the past year. According to recent data compiled by WccfTech, prices for DDR5 modules have dropped by 30%, marking the first significant reduction since the surge driven by AI demand. Chinese media outlets like UDN report that some retailers are already talking about a "price collapse" and anticipate that this trend is just beginning, reflecting a shift in market mentality from widespread scarcity to a wait-and-see attitude. This decline is particularly noticeable in consumer products, such as 32 GB DDR5 modules, whose price on Amazon has fallen from $490 to $379, a notable reduction although they remain expensive. Amazon Spain is also beginning to show these adjustments. The second-hand market is starting to adapt, and while there is not yet a flood of used modules, distributors are already sensing the imminent change. RAM manufacturers are feeling the impact: shares of Micron, one of the industry leaders, fell by 10% last Monday and have accumulated a 30% loss over the past week. Other companies like SanDisk and Western Digital have also recorded declines of between 7% and 9%, according to CNBC. The main reason for this correction is uncertainty about whether RAM demand will be as high as previously anticipated, following two recent revelations. On one hand, Google introduced TurboQuant, an algorithm capable of reducing memory consumption of AI models by up to one-eighth without sacrificing performance. Although its real impact remains to be seen, analysts predict it will allow for the execution of more complex models with the same or less amount of RAM, which could reduce global demand. RAM is essential for AI, as its capacity determines the efficiency and volume of processed data. Until now, high RAM consumption has forced tech companies to invest billions to secure supply, driving up prices for both consumers and device manufacturers. An example is OpenAI, which in October agreed with Samsung and SK Hynix to purchase 40% of the future global RAM production for its ChatGPT and Sora models. However, the closure of Sora due to high costs and low returns has raised doubts about OpenAI's ability to meet these commitments, especially after canceling the construction of a billion-dollar data center in Texas. Source: elespanol.com