In early February, beleaguered DRAM vendor Elpida Memory Inc. filed for bankruptcy, touching off a flurry of speculation from industry and bids from various microelectronics players. Now that Micron Technology Inc. and Elpida have confirmed that they're in discussions to hammer out the details of a buyout, we thought it was a good time to sit down with Mike Howard, senior principal analyst of DRAM and memory at IHS iSuppli, to get his take.
Memory Designline: What does the deal mean for Micron?
Mike Howard: The deal looks like it's going to be somewhere around $2.5 billion, which when you look at the asset base, is actually a pretty good deal. Micron’s getting roughly 200,000 wafers out of it. They're also getting a lot of technology, which at this point is probably worth as much as the fab. Elpida has a fantastic mobile DRAM product. Micron does have lower-power DRAM but it's a little leaky, it's just not as good and it reflects in their market share—they’re in the low single digits whereas Elpida is high teens and low 20s, depending, so I think [the IP] is going to be as important for Micron as the manufacturing infrastructure.
In a recent quarter, the average selling price for Micron DRAM was almost twice that for Elpida. Why is that, and what can Micron due to improve the picture?
Micron sells into a lot of premium niche segments like networking equipment, automotive, and then they also sell into the PC market. Elpida’s mobile product sells for a premium price but everything else is just bottom-of-the-barrel commodity DRAM. They don’t really sell anything into the server space, and that's what you need to do if you're going to run commodity DRAM. Commodity PC DRAM is the [product whose pricing] can drop 30% to 40% in a quarter when the industry is oversupplied. With embedded DRAM, the pricing tends to slowly ease down as manufacturers reduce cost.
Ultimately I think Micron doesn't want to take all of the [DRAM capacity] that Elpida has and run DRAM with it. I think they would like to maybe run half of it. That brings the industry to a really nice balance between supply and demand. Micron would be able to basically jettison the lousy part of Elpida’s business, get the average price up, and help the entire industry by taking 60,000 or 70,000 wafers out of the mix.
In the aftermath of the buyout, will the market be more rational?
Yes, you’ve got Samsung with 40% to 45% market share, so there’s a clear leader in the industry determining the pace of development, then you’ve got Hynix at a strong number two with a 25% market share, then some sort of number three at 15% to 20% market share. It's a lot easier to anticipate and postulate what your competitors are doing when you only have two guys to watch.
Five or 10 years ago there were eight to 10 companies building fabs. Between 2007 and 2009, the industry lost about $14 billion in value. A lot of Taiwanese companies became increasingly marginalized to the point that they're not on the radar anymore. Qimonda failed, and now with Elpida [failing], it’s starting to feel like things are settling down. We expect a more stable supply market and better pricing to where these guys can actually make an operating income.
What is the long-term prognosis for the DRAM market?
We’re moving into a post-PC era. PCs used to account for 70% of all DRAM demand—now that's down under 50% and we’re heading toward 40% quickly. I think some quarters out in 2015 and 2016, more than 40% of the bits will go into phones and tablets. DRAM in smart phones right now is a bottleneck—[mobile applications] are where PCs were 15 to 20 years ago when it was a significant move to upgrade to 32 MB of DRAM versus 16 MB of DRAM. We’re seeing that same sort of behavior right now with not only tablets but smart phones. With PCs, we are looking at maybe 5%, 7% growth per year for the foreseeable future, but mobile devices are absolutely picking up the slack with over 1 billion smart phones anticipated to ship by 2015.
At the start of May, Elpida bondholders reportedly threatened to block the auction. Do you think the deal could still be in danger at this point?
I think in the minds of the people running the facility, this deal is done. Right now, this is a lot of CFOs and treasurers squabbling over who is paying for what and how much of a haircut these debtors are really going to have to take. There's been a tremendous erosion of value in Elpida. They've got over $5 billion in debt and it's all coming due, and they really don't have any hope of ever paying all of that back, so savvy bondholders recognize when to cut their losses and take what they can get.
Three years ago when Qimonda went out of business, Micron was in discussions to buy all of their assets, but there was enough foot dragging that Micron eventually just stepped away from the deal and the Qimonda shareholders were essentially left with nothing. There is an element of immediacy here—if this deal does not get consummated soon, then I wouldn't be surprised if Micron just walked away. There is an element of immediacy here, I think that saner heads will prevail, though. They will look at that recent lesson in history and they'll get it done. The end of May might be quick but I wouldn't be surprised if we hear something in the next two or three weeks all of the specifics of the deal.
Did you find this article of interest? Then visit the Memory DesignLine
where we update daily with design, technology, product, and news
articles tailored to fit your world. Too busy to go every day? Sign up
for our newsletter to get the week's best items delivered to your inbox.
Just click here
and choose the "Manage Newsletters" tab.