Raw storage (hard disk drives, NAND flash and DRAM memory components) is an essential and intriguing market. Like a ducks legs under water these parts work really hard to keep the ITC market moving – they are not top of mind for most end-users. They are mostly bought by only the very largest organizations, most of whom are suppliers, integrators of distributors. Computers, tablets, phones, cloud services and IoT devices cannot work without them. They are so important that they are also victim to large government legislation restricting the countries to which they can be sold.
My Figure above shows the annual size (in $US billions) of each of the three offering types with a forecast to 2026. It shows that there have been huge periods of growth and decline with peaks of over 20% every four or so years. With only a few manufacturers with deep resources, expertise, relative cheap labour and expensive long-term manufacturing strategies, raw storage is a vast market with many single points of failure. For instance devastating floods in Thailand in 2011 temporarily halted the world’s supply of HDDs. This makes forecasting the market harder than many others; there is no room for conservatism here.
2023 was a low point for this market; there was an annual 25% decline in spending to $146b – you have to go back to 2016 to find a smaller annual total. Sales of all three types of raw storage dropped dramatically. In the full year DRAM sales were the largest at $90b, NAND was in the middle at $34b and hard disk drives at the bottom at $23b. I’m not the only one to predict a return to growth this year and luckily there were greens shoots of recovery in supplier financial reports in the fourth quarte4 of last year.
Despite the fall in market spending in 2023, the market shares of the leading vendors remained relatively consistent. Samsung, also widely known for its smart phones and tablets, retained its top position through being the largest supplier of NAND and second largest of DRAM chips and devices. SK Hynix was in second position overall mainly through its significant lead in the DRAM market. Most leading suppliers are Asian, although none of them are Chinese.; however the two leading suppliers of hard disk drives are both US companies, accounting for almost half of that sub-market.
The massive drop off in sales over the last two years has mad life hard for suppliers. For instance:
- Toshiba de-listed from the Tokyo stock exchange in December 2023 after 74 years,
- In October SK Hynix slashed its prices, partially blaming US export controls for the falling market,
- In 2023 Samsung Electronics cut production significantly as it made big losses from DRAM and NAND production,
- Micron – a rarity as a US-based vendor – found its products restricted for use in infrastructure projects by Chinese regulators,
None of the suppliers has found the 2-year slump in sales easy to manage, making temporary losses in a market that has to be managed on ai long-term basis. It’s reasonable for these suppliers to take short term losses before the market returns to growth; however stock market shareholders get their returns from profits and are likely to be less patient.
The wild variability in quarterly spending puts paid to the general observation that capacity growth is ‘exponential’; in general it is, but in both 2022 and 2023 the total capacity of raw storage shipped was lower than it was the year before.
I calculate the capacities of shipments for each of the three product types; the results of which by quarter since 2013 are shown in the two Figures above – both of which uses a logarithmic scale to accommodate the huge differences between the average capacities of each. The ‘exponential growth’ taken as a given by many in the storage industry applies quite well when looking at the long-term development Figure); however in the short-term the total shipped capacity has been falling each quarter having reached its peak in 2021. Q4 2023 saw positive capacity growth for both NAND and DRAM shipments, while HDD shipped capacity was only 12% down.
As for my forecast:
- If I’m right, 2026 will see capacity shipments reach virtually the same level as in 2021.
- if I’m wrong, it may turn out that market maturity and slow development of new storage products with larger capacities has put an end to the period of ‘exponential growth’.
In any case there are a number of reasons for market optimism such as the current mania for developing Gen AI applications, the much expected (though temporarily delayed) mass roll-out of 5G IoT and the slow return of the smart phone and PC markets to growth.
Not surprisingly the majority of these products is bought by Asian Pacific companies, almost all of whom are themselves suppliers or integrators of raw storage into subsystems into OEM equipment. I show my forecast for the next three years in my Figure above. Of course the destination of the products built is different, with the Americas and EMEA accounting for much larger shares.
If you are a supplier of these products and/or need to know more about my research into them, please contact me.