|Tidbits from the Cowen conference:|
Color for individual segments for SIMO
SSD controllers - 55% of total continuing sales in Q1, should grow 25% in Q2, and do well going forward. Do not have procurement forecasts from customers for H2.
eMMC/UFS - Big step down in Q1, flat in Q2, and in the second half of 2019 based on the pipeline of design wins growth potential seems good, and it's possible to be flat in 2019 for the full year compared to 2018. For this also, they don't have H2 orders.
SSD solutions - Big step down in Q1, and now waiting for open channel to go into production with 2 Chinese hyperscale customers in H2 2019.
Module makers are becoming more willing to buy NAND in Q2 because prices have stabilized somewhat, and the flash makers are giving them more advantageous terms of supply. Flash makers all have excess inventory, and more supply will come on line in H2 as 96 layer and QLC continue to come online. As a result, forecasts for H2 are uncertain across the board. No one knows if demand will return strongly, or NAND prices will again plunge lower. Without that knowledge no one is planning far out.
What's the total available market for the China hyperscaler enterprise open channel segment?
We think it is customized for specifically what the customers need. Hyperscalers have unique requirements. They want SSDs that can be optimized to their unique data flow. Current enterprise high end SSDs cannot be configured and optimized. That's what Open Channel provides. And they can buy anything from the open channel controller to the entire open channel SSD module/driver software.
This will be the first live commercial deployment of Open Channel SSDs in the world. If it works as advertised, and it can be scaled into a data center and into the customer's storage infrastructure, then it could eventually represent the entirely of a hyperscalers SSD requirements.
No numerical answer on the TAM
eMMC is a mature market and SIMO is managing it as such. Growth is in non-cell phone areas, smart TVs, etc. U UFS was as big as Hynix UFS last year, which is great since MU is relatively new to cell phone memory. UFS 2nd generation chip with MU will go into production in H2 2019. Further, Hynix is actively reducing its exposure to cell phone memory. Mobile as a percent of Hynix sales has gone from 80% to about 66% over the past few years. As Hynix walks away from low end NAND chips for cell phones, the Chinese module makers are entering and picking it up, and they were successful last year in low end non-smartphone segments, and this year they are pushing into smartphone OEMs. Many of these projects are expected to enter production in H2 2019.