I found this April 2020 report on the Phison web site. Phison makes a variety of NAND flash products, including controllers, USB sticks, SSDs, and (I think) resells NAND. The report was (I think) created before the onset of Covid, so I think the 2020 forecasts are on the too high side.
Here are some interesting excerpts with my comments in italics.....
The SSD controller IC market size for independent NAND controller manufacturers is estimated at $500M USD, approximately 40% market share of total SSD controller market size in 2020
Development costs for NAND controller ICs are increasing while the ASP of NAND controllers are decreasing
NAND Controller IC Market Size Analysis
Analyzing the SSD controller IC market serves to better understand all NAND flash storage device markets. According to TRENDFORCE, the 2020 world-wide SSD shipments are forested to be 310 million units
2. It is estimated that the NAND wafer manufacturers who also make their own SSDs will self-develop roughly 60% of their own IC controllers, which equals 186 million units. The remaining 124 million units are the shared available market for the independent NAND controller IC design houses (124 million units is 40% market share of SSD controller out of total 310 million SSD controller ICs). Estimating an average price for a typical SSD controller IC at $4 USD
3, the market size of SSD controller for independent NAND controller IC design houses is estimated at $500M USD (Exhibit 2). From the previous calculation, it is clear that for companies focusing only on NAND controller IC design business, the market size might not be big enough to make a profit or even survive due to the massive R&D investments required.
Interesting to learn that the split of internal versus merchant is 60-40. Samsung is 100% internal, so if we exclude them, I wonder what the numbers would look like? If Samsung has ~33% of SSD market share, then it means internal - external SSD controller market is 50-50. That's good to know, but what I would REALLY like to know is the trend - which direction (internal or external) is market share moving?
Using a 28nm SSD controller IC development as an example, it will roughly cost $22 million USD before the controller IC begins to sell on the market (Exhibit 3)4. When comparing the $22 million USD development cost for one SSD controller, and compare this with the total SSD controller IC market size $500 million USD (40% market share for independent NAND controller IC design houses), it is clear that a pure NAND controller IC business design house will have increasing difficulty making a profit or even surviving in the long term, not to mention while there is pressure to reduce the controller ASP (Exhibit 4).
--
The purpose of the report is to argue that controller development is very expensive and challenging, and that a standalone controller company will likely fail as it needs lots of cash to invest before harvesting the profits of its controller development. It also argues that the controller price is continuously falling due to competition.
SIMO is massively profitable, I think because they are the leading SSD controller company in terms of volume, and SIMO has a loads of cash, and SIMO says prices DO NOT fall, they are stable at $5 with new controllers being more than $5 and older controllers being less than $5 with the average .... for the past 8 years ... sitting there at $5. it makes sense that there is not price competition as SIMO owns the market, and Phison is the only reasonably sized competitor, and Phison probably competes on price. |
