Ever since Toshiba decided to spin out its memory business in 2017, leading to the creation of Kioxia, there have been rumors of Western Digital (WDC) acquiring it, or of Kioxia acquiring WDC’s flash memory business.
There is a new spate of merger rumors in response to Elliott Management’s proposal last May that WDC should be split into two parts: a flash company and an HDD company.
The SSD Guy blog doesn’t comment on rumors, but I can provide some insight into the challenges of merging Kioxia with WDC, and those insights indicate that such a merger is unlikely to actually happen.
WDC’s Sweetheart Deal
The most powerful argument against a WDC/Kioxia merger is the fact that a combined Kioxia/WDC company would be much less profitable than WDC is without Kioxia. This is largely due to the structure of these two companies’ manufacturing joint venture.
Kioxia and WDC jointly own flash fabs in a 51%/49% ratio, with Kioxia holding the larger share. WDC is allowed to take up to 49% of the wafers from this operation, but can take less if it cannot use all of them. This works to WDC’s advantage. Here’s how:
First, consider what happens when the market is in a shortage. WDC uses its 49% share of wafers to produce NAND flash that it sells in SSDs, flash cards, and USB flash drives. It can sell everything it produces at a high price because prices flatten or may even increase during a shortage. In this scenario both WDC and Toshiba will profit.
That changes when the market becomes oversupplied. Let’s say that the market becomes oversupplied by 1%. WDC can ask for fewer than its 49% of the wafers, leaving Kioxia with any wafers that WDC doesn’t need. With the 1% market oversupply, WDC will probably want 1% fewer wafers, or about 0.5% fewer of the JV fab’s output. Meanwhile, Kioxia won’t simply be dealing with a 1% oversupply, Since the company is selling into a 1% oversupplied market, then its 51% share of the fab’s output will give the company 1% more than it can sell, but the fact that WDC has declined taking 1% of its allocation Kioxia ends up with almost 2% more than it can sell. It’s hit twice as hard by this oversupply as any of its competitors.
If the company doesn’t sell this material then it will need to take inventory charges which will lower profits and may even lead to a loss. Kioxia must find a buyer for these parts.
Since the market is oversupplied all of Kioxia’s competitors have more chips than they can sell. The usual way for Kioxia (or any vendor) to sell its excess product is to take business away from its competition by lowering prices. This will also undermine the company’s profitability.
How does this whole scenario impact profits? Have a look at the chart below, which compares Kioxia’s operating profit margin to WDC’s flash gross profit margin. (Yes, that’s not completely fair, but unfortunately, that’s the only way that these two companies report their profits.)
Note how WDC’s margins never go negative while Kioxia’s often do. This indicates that the lossy part of the business goes to Kioxia while WDC can choose which business it takes to remain profitable.
The most recently disclosed quarter, the fourth quarter of 2022 at the far right side of the chart, shows that quite clearly.
If we compare the two companies’ revenues we get a more complicated story, but it’s still the same one. When market revenues turn down, WDC’s revenues turn down in proportion. Kioxia’s revenues increase, simply because the company suddenly has more product that it needs to get rid of at a low margin.
Every time that the WDC columns turn down, the Kioxia line is a little late to respond, since Kioxia is suddenly given the portion of WDC’s allocation that WDC was unable to sell. This forces Kioxia to drop prices to assure that the excess product doesn’t bloat its inventory.
This phenomenon is spelled out in detail in the Objective Analysis Brief: WDC – Using the Toshiba JV to Improve Profits.
Be honest with yourself: If you were WDC and you were involved in a relationship like this one, would you want to merge with Kioxia?
Does NAND Flash Need to Consolidate?
We can look at this same prospective merger from the perspective of the greater flash market.
Some people say that a WDC/Kioxia merger is a natural phenomenon that must come about as the NAND flash market is consolidating. They will then point to SK hynix’ acquisition of Intel’s flash business as an example.
Although it is not uncommon to find people who firmly believe that the NAND flash business must consolidate, very few can give a sound reason why.
The SSD Guy has been carefully monitoring memory chip markets from the time that there were 28 similarly-sized DRAM makers down to today’s three leaders. The mechanism that led to DRAM’s consolidation hasn’t yet hit NAND flash, although in time it will. (The mechanism is explained in detail in the Objective Analysis Brief Why DRAM Vendors Must Consolidate.) The short explanation is that DRAM market revenues haven’t grown to keep pace with the cost of building new production capacity. This phenomenon is likely to continue to lead to further DRAM consolidation.
NAND flash’s revenues grow faster than those of DRAM, and faster than the cost growth of new capacity. This situation lends itself to the addition of new competitors, rather than to consolidation, which YMTC may soon prove.
The short answer is that DRAM has a dynamic that lends itself to consolidation while NAND flash does not.
Over the next couple of decades NAND flash revenue growth will slow, as is natural for market growth, but the growth of capital costs probably won’t. When average NAND flash revenue growth falls below capital cost growth, then consolidation will become inevitable. We’re not there yet.
Merger Unlikely
Both of these perspectives indicate that a merger between Kioxia and WDC cannot be assumed. There appear to be stronger reasons for the companies to remain apart than for them to merge.
But we expect for people to continue to hypothesize such a merger or even an acquisition. It’s only been a year since strong rumors much like today’s predicted a WDC takeover of Kioxia. The SSD Guy put forward a similar argument then.
Objective Analysis understands market dynamics and the health and business relationships of the participants. This helps our clientele to understand the market and its direction for their strategic planning processes.
If your company would like to rise above the competition by taking advantage of our insight and understanding please contact us to discuss ways that we can help you in your strategic planning processes.
Visit our website, and let us know when you want to explore ways that we can work together to strengthen your business.
This argument has a hole in it. Consider your market decline example. The market declines by 1% so now WDC takes 48.5% of capacity and Kioxia take 50.5%, leaving the plant operating at 99%. Both companies sell the products and separately account for profits on their sales, which have equally adjusted for market conditions -at least, assuming both companies follow the same strategy and have uniform customer demand.
Meanwhile the plant is less profitable since it runs only at 99% capacity, and as 49/51 shareholders, both WDC and Kioxia see very close to the same impact on their valuation and cash flow.
So, how does this turn into a structural disadvantage for Kioxia? The answer must lie either in some other aspect of ownership/obligations not seen in the top line 49/51 arrangement, or in the different sales strategies they follow which have been differently successful.
Thanks for the comment Tanj.
Something that I didn’t say in this post is that fab owners almost never turn down their output. Half of their cost is from capital depreciation, so the fewer parts they produce, the higher the cost of each part. It’s hard to warrant increasing costs when prices are collapsing.
In my example, even though there’s a market for only 99% of the JV’s product, the JV turns out 100% anyway. This drives Kioxia’s revenues to grow at a time when the market is shrinking, with their profit margin suffering as a result.
Kioxia and WDC have elected to take the difficult step of reducing output, but it takes pretty long for that decision to have an effect. The actual output reduction is only hitting this quarter.
Good article Jim with reasonable arguments, but actually the “almost never turn down their output” is not true in this case, as e.g. Kioxia announced significant decrease already in September:
https://www.reuters.com/technology/japans-kioxia-says-cut-wafer-start-production-volume-by-30-2-plants-2022-09-30/
Thanks Mark. That’s a fair point for the current cycle.
If you’re looking at the longer term, though, which I would expect the architects of any merger to do, factory output is not turned down during most down cycles and Kioxia has suffered as a result.
Under the current structure WDC can pass a certain amount of its losses to Kioxia during these times. Were they to merge then WDC would lose this advantage.
That would benefit Kioxia, but certainly not WDC.
Jim