2021 will become a year in which the demand for memory will increase considerably, the main reason being the increase in products that use NVMe memory, but the new manufacturing nodes for memory also allow RAM of all kinds with more capacity, faster and / or with less consumption.
What is the Micron 1-Alpha manufacturing process?
Micron just announced that the first batch of DDR4 and LPDDR4 memory chips manufactured under its new manufacturing node, 1-alpha, which supports 8 Gb and 16 Gb densities per memory chip.
Since 2019 until now, Micron have been using their 1Z nm node, but with the new 1-alpha process they can reach an improvement in density of up to 40%, as well as a reduction in consumption of 15% under the same clock speed.
Micron has achieved an additional 10% performance over the previous generation thanks to the combination of the new node and improvements in the internal organization of the RAM.
1-Alpha: a process for all types of RAM
Micron has plans to expand its 1-Alpha process beyond DDR4 and LPDDR4, a transition to which will arrive at the end of the year and which includes not only DDR memories, but also LPDDR including the new LPDDR5 and VRAM memories such as GDDR6 and GDDR6X.
In the case of GDDR6 and GDDR6X, the jump to 1-Alpha is important, since the new node will allow the number of valid chips to increase, reducing the cost and increasing the availability of these memories in the market. Especially 16 Gbps GDDR6 and in the case of GDDR6X this will lead to the appearance of 16 Gb (2 GB) chips.
Will we see a review of the NVIDIA RTX 3090 but with 12 memory chips or an eventual NVIDIA Titan Ampere with an impressive 48GB of VRAM in a few months? Who knows, what is clear is that the high demand for RAM of all kinds makes it necessary for new nodes to manufacture it.
Micron 1-Alpha, to meet the increase in demand for NVMe
Another type of memory that will benefit from node 1-Alpha is NVMe memory, that is, the non-volatile memory used in SSDs and other types of memory based on NAND Flash, this is the real reason for the development of these new types of nodes, since the demand for this type of memory will increase for two reasons:
- GPUs already support access to SSDs connected via PCIe thanks to the new generations of GPUs from AMD and NVIDIA in combination with the DirectStorage API, the games that will ask for an SSD as a minimum requirement will grow and will become one of the key pieces of PCs for Gaming.
- The next-gen video game consoles have only just come out, but at the moment they are in the introduction phase, their numbers are expected to grow and both use NVMe memory for storage thanks to their integrated SSDs.
So the demand for NVMe chips is increasing considerably right now.
Will we see cheaper memory thanks to the new nodes?
One of the reasons companies like Micron have developed new manufacturing nodes is to increase yields, which translates into the number of chips per wafer.
Although the cost per chip decreases the more of them come out per wafer, we cannot fall into the error of confusing cost with price, since the price depends on the value that the market gives to said product, which depends on its marginal utility. . In other words, what we have is an increase in production and a reduction in costs for Micron, but this does not have to translate into a price reduction for its customers and therefore not for the end user.
The 1-Alpha process will compete against Samsung’s processes based on EUV manufacturing nodes and Micron itself intends to deploy the 1-Beta and 1-Ganma nodes that will follow the newly launched 1-Alpha.