PHOENIX--Contrary to one report issued last week, Korea's Samsung Electronics Co. Ltd. did not surpass Intel Corp. as the world's largest flash-memory supplier in the third quarter of 2003, according to an e-mail newsletter from Semico Research Corp.
The report issued by research firm iSuppli Corp. combines NAND and NOR flash-memory vendors in the same ranking, with Samsung as the No. 1 provider in Q3, surpassing Intel. But in the newsletter, Semico disagreed with the report from rival iSuppli and argued that NAND and NOR should not be combined in the same rankings. So in reality, Samsung and Intel are still the leaders in the separate NAND and NOR markets, respectively, according to the Phoenix-based research firm.
Meanwhile, as reported last week, Samsung increased its flash memory chip revenue 50 percent in Q3, posting sales of $615 million to rank as the largest global vendor, according to an iSuppli report released Thursday (Nov. 20). Samsung, with a 20 percent revenue share of all flash architecture shipments, topped Toshiba Corp., which was in second place with $532 million in revenue in the third quarter and a 17.3 percent market share. Toshiba increased its flash revenues 61 percent over the same period a year ago.
FASL LCC Spansion (the joint venture of Advanced Micro Devices and Fujitsu Ltd.) ranked third in market share with 13.8 percent and $424 million in third quarter revenue. Intel, once the leader in the market, was fourth with a 13.5 percent market share and $416 million revenue. Sharp placed fifth with a 10.4 percent market share and $318 million in revenue (see November 20 story).
Jim Handy, an analyst with Semico, took exception to iSuppli's rankings. "Semico plans to report flash market share for 2003 the same way we did for 2002," Handy wrote in the newsletter. "NAND and NOR will be separated to account for the fact that these are two very different markets. We see Intel continuing to dominate NOR, and we don't expect for this to change until Spansion brand chips replace their Fujitsu and AMD predecessors. Meanwhile, Samsung is quickly ramping their NOR production, so Spansion's day in the sun may be short-lived."
But still, has Samsung taken the first-ranking spot from Intel? "If we look at this in a very simplistic way, the answer would be yes," Handy wrote. "In their quarterly earnings statements Samsung says that they have shipped $693 million of NAND, which is significantly more than Intel's reported $450 million of Wireless Group revenues which include all of their NOR shipments," Handy said.
"In a similar vein, Sun Maid ships more raisins than Sunkist ships oranges. Just as you can argue that both NAND and NOR are flash, you can argue that oranges and raisins are both fruit," he wrote in the report. "The two technologies have very little in common. Semico has steadfastly argued that NAND and NOR are as dissimilar as SRAM and DRAM, and the world seems to finally be coming around to this way of thinking. To take it to the extreme, if we count total memory shipments, then Samsung has been ahead of Intel for years!"
The same argument is true with NAND supplier Toshiba, which is supposedly bigger than Intel. "Add to this the NAND vs. NOR argument stated above, and you can see that Semico doesn't believe this in the slightest," he said.
In fact, it's unlikely that AMD or Spansion is larger than Intel. One of the issues with AMD, Spansion or Fujitsu is double-counting.
"Here's how the rules work in this case: The parts that AMD now ships are marked 'Spansion' only if the chip design was production qualified after the Spansion name was announced," Handy said.
"All other parts shipped by AMD bear and AMD logo until that version of the part is rendered obsolete. Fujitsu follows a similar policy. If we follow the rules as we have in other years then 2003 market share reports will use all three names: AMD, Fujitsu, and Spansion, to account for the revenues of parts marked with each of these three logos. Why is this important? Because the notion that AMD or Spansion is larger than Intel requires the market researcher to put aside this rule in this particular case," he explained.