RAM Prices Going Down? Google’s new innovation
The RAM crisis started at the end of 2025, after data centers started buying high bandwidth.
Memory (HBM) in massive quantities, which made the manufacturers shift their production
from standard consumer RAM into making HBM for more profit and guaranteed sales, which
resulted in a consumer RAM shortage and a price hike of more than 270% for some units.
This crisis heavily affected those who are interested in buying or selling consumer RAM, most
PC builders went back to DDR4 RAM rather than DDR5 for the lower price even if it meant using
an older generation product and losing performance. This crisis has also affected GPUs.
manufacturers such as NVIDIA, AMD, and Intel, as GPUs contain VRAM, which is a special type of
RAM is designed for high-bandwidth and parallel processing tasks. The growing demand for AI
computing has significantly increased the pressure on VRAM supply, further contributing to the
overall memory shortage. This forced GPU companies to increase their consumer GPU prices.
and delay future generations’ releases.
But everybody’s hopes were revived after Google released a research paper called
“TurboQuant: Redefining AI efficiency with extreme compression." TurboQuant is a new
compression method for AI models that reduces the usage of memory by six times or more,
speeding up performance up to 8 times, while normal AI models rely on high-dimensional
vectors (huge arrays of numbers), which takes a lot of memory and slows processes.
All this might result in data centers buying and using less memory in the long term. which might
lower the consumer RAM prices, which has already happened with RAM prices on Newegg
Electronics selling websites have lowered by 10-30% in the last few days. But will that solve the
problem? Not currently. While the demand for RAM has been lowered, the negative supply and
Demand imbalance still exists. As data center companies signed long-term contracts that grant
them memory in the future. In contrast, they paid money for more HBM to be made in the
future.
With the current slope of AI advancements, the need for memory will increase, which will
eventually cancel out the TurboQuant advancements and will lead us back to the current point.
unless a new compression method is created. As more data centers are being built, more
Companies are entering the field, and the usage of AI in the average person’s life is increasing.
rapidly.
Memory manufacturers prefer producing HBM as it offers stable demand and a guaranteed
high profit margin; on the other hand, consumer RAM’s demand is not as stable and fluctuates
over the span of the year, and the profit is greatly less than that of HBMs.
Ultimately, while innovations like TurboQuant offer a promising step forward in reducing
memory usage in AI systems; they are not an immediate solution to the current RAM crisis. The
imbalance between supply and demand persists due to long-term contracts, production shifts
toward HBM and the rapid expansion of AI technologies. Although prices may show some
improvement, it is unlikely that we will see DDR5 memory return to prices as low as $95. anytime soon.
- Haytham Ahmed Subhi