While RAM prices increased up to seven times in the last year due to the demand for artificial intelligence, increasing production was seen as the only solution for the sector. However, the new system introduced by Google has the potential to radically change this equation.
Giant manufacturers are in decline
Following the development, the world’s leading memory manufacturers suffered sharp losses in the stock market. SanDisk shares fell as much as 8 percent, while Micron fell 5 percent. South Korean giants Samsung and SK Hynix also lost 4.7 percent and 6.2 percent, respectively. These two companies control the majority of global DRAM production.
TurboQuant: Reduces memory requirement by 6 times
The new technology called TurboQuant, developed by Google, dramatically reduces memory usage by compressing the way artificial intelligence models work. It is stated that the system can reduce the RAM required for training processes by approximately 6 times.
Thanks to this technology, the data representation of the models can be reduced from 16-32 bits to only 3 bits. This paves the way for a serious transformation in the artificial intelligence infrastructure, which until now has largely depended on hardware needs.
“Google’s DeepSeek moment”
According to Morgan Stanley analysts, this development could create a significant breakthrough in the artificial intelligence and memory industry. Lower memory requirements could significantly reduce the cost of AI applications, enabling them to be used on a larger scale.
Cloudflare CEO Matthew Prince described this technology as “Google’s DeepSeek moment.”
The crisis is not over, it may only ease
Despite this, experts are cautious. Analysts emphasize that technology will not solve the high-speed memory (HBM) problem in the short term and that the problem in RAM supply will not completely disappear.
Some research companies state that this development will only reduce demand pressure, but that the need for artificial intelligence-based memory will continue to grow.
How did the memory crisis arise?
The rapid growth of AI data centers has exploded the demand for RAM in recent years. The industry avoids increasing production aggressively as it has learned from the crisis caused by the supply-demand imbalance experienced during the pandemic period.
For this reason, while prices rise rapidly in response to increasing demand, companies prefer to follow a controlled growth strategy.
