Flying high at Micron

Battered from the memory's darkest years, a stronger Micron emerges

by John Dodge -- EDN, April 1, 2006

Under CEO Steve Appleton, Micron Technology has been on a 12-year odyssey that is finally showing a payoff.

Micron's share price has nearly doubled since May 2005, and the once troubled memory maker has earned a profit for the past seven out of nine quarters. Confidence in the company's improving prospects manifested itself in November 2005, when Micron and Intel struck a $5 billion deal to create a joint manufacturing operation to produce memory for the burgeoning NAND market.

But it hasn't always been this way under Appleton. In fact, it's hardly been this way for many of his 12 years as Micron CEO. The low point might have been the financially terrifying second quarter of 2003, when Micron posted a net loss of $619.2 million on net sales of $785 million. In 2003 Micron lost $1.27 billion on $3.09 billion in sales and the company bled red ink to the tune of $3.5 billion during the previous three years.

During those difficult times, Micron was continually writing down DRAM inventories, which it was producing below cost. In the company's fiscal 2003 third quarter, Micron's press release announcing results didn't even contain the usual quote from the CEO outlining future prospects. By then the succession of staggering losses suggested that Micron had no prospects. But despite the torrent of red ink, Wall Street, even in 2003, expressed faith in Appleton, given that company shares were trading just below what they are trading for now in better times.

Today Micron finds itself on solid footing under Appleton's enduring leadership. He might well author a book, called How to Succeed in Business and Take a Helluva Long Time to Do It. In hindsight, Appleton's strategy—take Micron down to the nub by getting rid of everything but its core DRAM business and diversify that to protect against long periods of oversupply—seems simple enough. Along the way, the company morphed into an innovator instead of placing all its bets on winning just the commodity DRAM game.

Indeed, Micron, 70 percent of whose business still comes from DRAM, operates in a market chronically wracked by oversupply and declining prices. Just between the second and third quarters last year, Micron says, the average selling price (ASP) of DRAM (DDR and DDR2) plunged 30 percent. On top of that, Micron on several occasions has been charged with DRAM price fixing and is involved in a lengthy patent dispute with Rambus. One Micron employee was found guilty of obstruction of justice in 2003 with respect to the price-fixing allegations.

"DRAM is a tough, tough business," Appleton says. "The ASP has continued to fall, and there's plenty of supply. Our results have held because our ASP exceeds that of the industry."

Appleton's claim is not without merit, according to Doug Freedman, a senior semiconductor analyst with American Technology Research. "Across the board, that is correct. Micron and market leader Samsung enjoy a premium, because their quality is better. Micron needs to be a technology leader to remain competitive," he says.

Indeed, Micron's executive succession plan stresses a background in innovation and R&D. In February, the company elevated former chief technical officer (CTO) and VP of R&D Mark Durcan to the post of COO. One company source confided that speculation that the 44-year-old Durcan is being groomed to succeed Appleton as CEO is on the mark. And smartly, customer types such as Sony adviser and DVD pioneer Dr. Teruaki Aoki and ADC Telecommunications CEO and president Robert Switz were recently added to the Micron board.

With respect to innovation, Appleton says Micron averages about 1,800 new patents a year, although in 2005 the number dropped to 1,561, according to IFI Patent Intelligence. In 2005 Micron ranked sixth in patents worldwide, just 10 ahead of partner Intel and 84 behind No. 5 Samsung, the 1,000-pound gorilla of memory. In 2002, Micron said it was third, with 1,833 patents. 
 

“Appleton is not afraid to let you take risks to put the company in a better position. He’ll let you make difficult decisions and be responsible for them.”
—Mark Durcan, Micron COO
"We're coming up with new ideas and new ways of being cost-competitive," Appleton says. "Sometimes they run in opposition, because the more money you spend on R&D, the higher your R&D charge per part. Sometimes they run in parallel, such as when you call for a new process node that shrinks the die and gets more bits per square centimeter of silicon. R&D can run in both directions on you."

Expand the portfolio

Innovation, however, does not excuse Micron from going all out to continually lower costs to produce commodity DRAM and to come up with higher-margin memory products.

"A long time ago, when we could cherry-pick the market, we supplied only 3 to 5 percent of the marketplace. We focused on high density and high volume, but we did not have a lot of costs with developing other new products. It worked OK as long as the growth profile was pretty strong and periods of undersupply were a little longer than those of oversupply," explains Appleton.

As DRAM oversupply became the rule, Micron started focusing on innovation and development. By 2001 slower growth in DRAM imposed a harsh reality on the Boise, Idaho-based company, forcing Appleton to realize that even with upward of 20 percent of the world DRAM market by then, the commodity memory market would not sustain Micron long-term.

For Appleton, it was déjà vu all over again. He would have to broaden Micron's semiconductor portfolio to include higher-margin, faster-growing products such as NAND and CMOS image sensors. Broadening Micron's semiconductor portfolio would follow his six-to-seven-year effort to spin off an array of businesses outside of semiconductors.

So two years ago, Micron launched a strategy to develop and manufacture NAND flash memory, a flourishing market that has many years to run. And in late 2001, the company entered the CMOS imaging business, when it acquired Photobit. NAND, higher-margin specialty DRAM and CMOS imagers combined accounted for 30 percent and 45 percent of the company's sales in the fourth quarter of 2005 and first quarter of 2006, respectively.

The dark days

To understand where Micron would head in 2001, it's important to review where it had been. When Appleton took over as CEO in 1994, Micron was literally a high-tech conglomerate. The company had independent units selling PCs and contract manufacturing services. There was a Web hosting unit, an equipment company and a display venture. Even a construction company that built everything from fabs to prisons was in the mix. Appleton set out to whittle the company down to just semiconductors.

"By 2001 we were clean. We got rid of everything but semiconductors and stabilized revenues. We never had to take a hit on the critical mass of the company," says Appleton in explaining why trimming the company took as long as it did. "We were basically getting out of something every year. It wasn't like in 2001, when we said, 'Holy smoke, let's kill all that stuff.'"

But just as Appleton finished shedding the nonsemiconductor businesses, Micron plunged headlong into the tech wreck.

"We had a huge recession in semiconductors and a crushing depression in memory. We had a 90 percent drop in ASP over 18 months. It hurt us, but we did not lose sight of continuing to invest in these products. We were getting a lot of criticism, because our results were not very good. Even a year and half ago, some DRAM companies were showing improved results, but ours weren't as good. We continued to invest in technology, knowing that it would take us some time."

Even during the tough times, Micron still made investments. In April 2002, it bought Dominion Semiconductor, the commodity DRAM arm of rival Toshiba. Today it plans to use part of the fab to manufacture NAND. That same year, Micron took an unsuccessful run at acquiring Hynix's memory business. The company has never stopped bolstering its commodity DRAM business, which represented nearly 100 percent of sales at that point.

Although diversification and improved process technology hold the keys to Micron's future, many cost challenges still lie ahead, according to Freedman. For instance, its large United States–based fabs are more expensive to operate than they would be in Asia. "Their costs are higher. You can't write off high labor costs," Freedman says.

“We decided that the best way from a timing element was to partner with (a company) like Micron…”
—Ed Dollar at Intel

But he credits Micron for being an innovator and controlling costs. For instance, Micron was one of the first companies to ship 1-gigabyte DDR2 DRAM, says Freedman. Appleton claims Micron was also the first to ship 1-GB DDRs, 1-GB DDR2s and 2-GB DDR2s. Fab costs are also defrayed with TECH Semiconductor, a 0.11 micron DRAM fab in Singapore, jointly operated by Micron with HP, Canon and the Singapore Economic Development Board.

The cost of DRAM capital equipment is another concern.

"We historically had a model where you spend a billion on a piece of equipment that 3.7 years later was obsolete. You pull it out and sell it for cents on the dollar," says Appleton, adding that one reason for the move into CMOS imagers is that they are produced at far less cost.

Staying ahead in the DRAM game means greater production of wafers with higher densities—an area in which Micron has excelled, according to Freedman. For instance, its award-winning 6F(2) DRAM cell architecture gave the company an edge in 2004 over the commonly used 8F(2) architecture. "One way to offset costs is to be a technology leader," Freedman says.

Last one standing

Some recent developments have gone Micron's way. For instance, No. 4 NAND producer Renesas announced in December 2005 that it would exit the NAND flash memory market north of 8 GB. Undoubtedly, No. 5 NAND producer Micron will move into that spot, with only Hynix, Toshiba and market leader Samsung ahead of it.

Indeed, as the cost of smaller geometries and yet higher volumes in DRAM and eventually NAND drive still more players from the market, Micron will look at scavenging the assets. "Suffice it to say that if there is industry consolidation in the future, we will participate in that to the greatest degree possible. If the opportunity comes along, we're going to take a look at it. Nothing that's imminent sits out there today, though," says Appleton.

If this durable CEO faults himself for any recent miscues, tardiness to the NAND market would be it. Market leader Samsung dwarfs all comers and enjoyed a 15-fold revenue advantage during the third quarter of 2005, according to iSuppli.

"We should have been in the position we are today with NAND four to five years ago. Internally we had a discussion five or six years ago about the benefits or competitiveness of NOR versus NAND and their ability to compete with mass storage. We didn't make the right decision at that time," he says, adding that Micron will soon exit the NOR business.

But failure to jump into the NAND market earlier could hardly be termed fatal. After all, the Intel marketing machine approached Micron to be its NAND partner, and although Samsung declined to comment, the $5 billion joint venture has to make the market leader a bit nervous. Then again, the NAND market is growing fast: iSuppli estimates that it will grow from $10 billion in 2005 to $26 billion by 2009. Apple is so concerned about short supply that last fall it prepaid $1.25 billion for NAND from the top five producers.

"We decided that the best way from a timing element was to partner with (a company) like Micron…(given) the time it would take to develop our own product from a design standpoint," says Ed Dollar, Intel CTO for flash memory. Intel, which made two previous investments in Micron, developed and backed NOR flash but now realizes that NAND almost certainly has stronger growth prospects.

So Intel gets into NAND quickly and relatively cheaply while Micron glides onto the fast track. Durcan explains, "It is one of the watershed events in Micron's history and is going to drive a lot of change. It enables us to get to scale much more quickly than we would have been able to ourselves." Also, Micron bought Lexar Media Inc. on March 8 to strengthen its competence in NAND controller and systems design.

At the same time, NAND will help insulate Micron from the DRAM market's wild swings .

"NAND in combination with DRAM is like a bowl of Jell-O," says Mark DeVoss, flash memory analyst at iSuppli. "Push in the middle, and the sides go up. Push on the sides, and the middle goes up. NAND is the counterweight. If DRAM is out of cycle, NAND could be in cycle. They pretty much use the same equipment and process. And Micron has definitely picked the right partner."

With CMOS imager revenues growing fast but still comparatively small—Micron does not break them out—the combination of NAND and higher-margin non-PC DRAM will carry the company for years to come. Appleton is betting that that will be the right combination for exploiting the red-hot handset market. To reflect the increasing integration of DRAM and NAND, Micron, as part of a February reorganization, merged its systems and mobile division into a single unit that sells all of its memory lines: NAND, DRAM and specialty DRAM.

"Right now, the majority of cell phones take a pseudostatic RAM and combine it with a NOR device. That will continue, but that combination will be replaced by a NAND device and low-power DRAM, or what we call cellular RAM. The market is migrating to a combination of the two devices we make. There aren't many of us that are able to do that," declares Appleton.

Number four memory maker Micron badly wants to move into Toshiba's number two spot and eventually Samsung's number one spot. Is this feasible? Send your thoughts to feedback@eb.reedbusiness.com.

TALE OF TWO YEARS
Happier days are here again

SOURCE: MICRON TECHNOLOGY
2005 Results	Net sales: $4.88 billion
	Net profit: $188 million
The good ol' bad days
2002 Results	Net sales: $2.59 billion
	Net loss: $907 million

FLATTENING DRAM MARKET
(revenues in billions $)

2005	2006	2007	2008	2009
SOURCE: iSUPPLI
$47.3	$51.6	$62.8	$65.5	$67.3

John Dodge is editor-in-chief of ELECTRONIC BUSINESS.