It was only a couple of years ago that Micron Technology took the difficult steps towards becoming a ‘diversified’ IC manufacturer, gradually reducing its dependence on the PC DRAM commodity market. A key part of that strategy was CMOS image sensors and, due to key acquisitions and manufacturing expertise, Micron became the largest image sensor manufacturer. 

Indeed, the strategy had other significant benefits to the company as a whole such as leveraging aged 200mm fabs that couldn’t make memory chips cost effectively. Using these fully depreciated fabs for CMOS image sensors, it could generate higher margins while participating in a market with stable prices.

Other benefits included the ability to serve mobile customers with a group of products that included image sensors and memory, making it a more ‘valued’ supplier.

Many industry observers were sceptical that such diversification would prove successful as no other memory-based manufacturer had successfully diversified without exiting the DRAM market altogether.

Micron’s DRAM market share began to fall several years ago as it reduced capacity spending on PC-based DRAM switching to higher differentiated memory products. Even when it entered the NAND flash market with Intel as a partner, the emphasis was on diversification and another way to sell DRAM, NAND and image sensors into the mobile phone market, for instance.

However, initial rumors that Micron was going to offload its CMOS image sensor business didn’t seem right to me, as this division has been a successful part of the diversification game plan.

The analyst day held by Micron last week changed all that as executives made it clear that it was looking at a number of ways to spin-off the division so that it can make its own decisions and have access to capital to independently grow the business.

That means that Micron can now ‘return’ to being a dedicated memory manufacturer with both DRAM and NAND flash devices at the core of operations going forward.

So I guess the sceptics were right all along!

To play in the memory market you have be dedicated to it, and in times of financial losses due to oversupply, that becomes even more critical. Micron just doesn’t have the bandwidth to be a player in anything other than DRAM and now NAND.

However, it’s no good being a player in these markets anymore. What you need is to be a leader both in technology and manufacturing capacity. Big bucks on both scores is required as well as the ability to be an aggressive first mover company.
Micron’s Steve Appleton sees the memory mess of today as a perfect time to make things happen! ‘Happen’ in this instance means forging ahead with leading-edge process nodes on both DRAM and NAND, boosting capacity to become a major market share player with a solid 30 percent share, and being an aggressive low-cost manufacturer to better weather the price falls compared to its rivals.

The last piece of the strategy is to assist in the further consolidation in the memory market. That didn’t happen according to plan in the last crisis, with Hynix being bailed out by the Korean Government instead of being acquired by Micron.

This time round, Appleton believes that the dedicated DRAM manufacturers will have a tough time surviving without having a significant NAND flash product portfolio as the two types of memory allow major end-users to make purchasing cost savings. These include the Elpida/PSC partnership as well as the Qimonda/Inotera/Nanya alliance. It is these players that Appleton sees as being vulnerable to true mergers/acquisitions without the NAND capability in scale.

However, this may be one of those convenient truths.

Looking at Micron’s capital spending plans for 2008, it is very obvious that the majority of spending is going towards NAND capacity in the second half of the year. Basically, Micron doesn’t need NAND capacity from ‘other’ sources.

This is due to the completion of the new IMFT 300mm fab in Singapore that has an initial capacity capability of 60,000 wafer starts per month and much more after that in further expansion phases. Then it has the 300mm Lehi fab that has not fully ramped and is basically on capacity addition hold. But we can expect more capacity from Lehi when the market picks up.

Micron only has the TECH semiconductor facility, also in Singapore, which is able to expand DRAM wafer capacity in 2008 after its full transition from 200mm to 300mm is completed by the end of the year. Much has already been converted to 300mm but somewhere in the range of an extra 4,000 to 5,000 wafer capacity can be expected.

This means that Micron will need external DRAM capacity if it wants to be a 30 percent market share player sometime down the road. So the bets are on either one of the dedicated DRAM partnerships being a suitor to Micron in an expected round of consolidation!

Finally, it has to be said that Micron is mad to be back in the game of dedicated memory.

But when you have an executive like Appleton who flies stunt planes (with the occasional crash) for relaxation, being mad may be the best strategy!