Monday, January 6, 2020
Taiwan-based Macronix International has developed into the world's largest supplier of ROM chips and second biggest maker of NOR flash memory chips after experiencing ups and downs over the past three decades, such as reaching annual revenues of US$1 billion only 10 years after its establishment, and suffering major losses in the following two decades.
Marking its 30th anniversary in early December 2019, Macronix has mapped out its development roadmaps for the next 20 years, with memory-centric solutions to be its secret weapon in the next 5-10 years, company founder and chairman Miin Wu told Digitimes in a recent interview.
Q: Would you shed light on some impressive achievements during the good days of Macronix's first decade?
A: Macronix got off to a good start after registering as a SEMI member and a dedicated ROM supplier in 1989, with Japan's semiconductor leader Nintendo becoming our first international client, allowing us to earn the "first pot of gold," though with a profit of only 5%.
And then we earned US$16 million through licensing our ROM technology and product lines to Japan steelmaker NKK, who was eager to venture into the semiconductor sector, with the fund used to finance our R&D projects.
Later we borrowed money from Nintendo to build a monthly capacity of 10,000 wafers and further used IPs as collaterals for more loans from the Japanese partner to expand the capacity to 40,000 wafers, paving the way for our annual revenues to expand all the way to hit US$1 billion in 2000.
Q: Macronix suffered headwinds earlier in its second decade, but achieved a major turnaround later. What was the key to the reversal?
A: In 1999, I went to the US for a coronary artery bypass surgery and returned in 2002 to find Macronix suffering annual losses of NT$10 billion (US$333.43 million) for enforcing too many unprofitable R&D projects.
In July 2002, I took the helm again seeking to return Macronix to profitability. We dropped many inappropriate projects and singled out 30 more feasible ones after assessing their costs, market competitions and future development needs. Following four years of efforts, our overall gross profit rates neared 50% in 2006-2007.
Meanwhile, we got licensed technology from an Israeli partner for manufacturing miniaturized ROM chips, with one chip able to support 4-bit memory circuit, compared to the regular norm of one bit for one chip. This helped sharply reduce the delivery lead time to only 1-2 days from 12-14 days and boost Macronix's overall gross profit rates to over 50% in 2008- 2009.
Q: Macronix started to suffer heavy equipment depreciation pressure after purchasing a 12-inch fab from ProMos Technologies in 2010. How do you interpret Macronix's business transformation in the third decade?
A: Macronix entered an expansion stage in the third decade and migrated to 12-inch production. It took five years for the 12-inch fab to start volume production, and heavy equipment depreciation cost sent us operating in the red till 2016. But our business operations have since improved significantly, especially bolstered by successful rollout of 55nm NOR flash and 19nm NAND flash chips.
Our 12-inch fab now has a monthly capacity of 20,000 wafers, equally split for NOR and NAND flash memory chips. We have also completed deployments of SLC (single-level cell) NAND and eMMc (embedded multimedia cards), and we will enter 48-layer 3D NAND production in 2020 before advancing to 96-layer in 2021 and 192-layer in 2022.
Q: Is there any core management practice Macronix has sustained over the years?
A: Besides being the first Taiwanese semiconductor firm to computerize its plant operations, Macronix has also taken the lead in incorporating big data analysis into the industry. I started to engage in big data analysis research 30 years ago, and our in-house-developed sNOVA system has performed well in helping analyze all the aspects concerning memory chips, technologies and plant operations, boosting the product accuracy level from ppm (parts per million) to ppb (parts per billion).
Q: How do you see the growth opportunity 5G will bring for the memory industry?
A: 5G is a market for high-density NOR flash memory. As 5G communications equipment may be installed in diverse outdoor spaces, memory quality counts greatly in determining whether the equipment can operate well. In this aspect, Macronix boasts a distinct advantage in terms of memory quality, backed by the databank it has established over the past three decades. If measured by shipment return rates, our product quality can be 10-50 times better than that of US peers, with even top Japan automotive customers regarding Marconix as their best supplier.
As one of the world's only two suppliers of ROM flash memory chips for 5G equipment, Macronix has established partnerships with 80% of supply chain players. But our ratio of revenues from 5G consumer applications will not be too high as our business focus is not on the consumer sector.
Q: China has just launched a second-phase National IC Industry Investment Fund to strengthen its own semiconductor supply chain. How do you see the development prospect for the country's memory sector?
A: China has a big domestic market and can create its own rules of the game, but Taiwan cannot. Nevertheless, it will still take 20 years for China's semiconductor industry to fully catch up with international competitors. In other words, Taiwanese semiconductor firms can still survive for 20 years more.
China's IC design sector will soon outrace Taiwan's, given HiSilicon's growing competitiveness against MediaTek. And it is just a matter of time that China's chip fabrication sector will catch up as a result of strong government policy and subsidy support.
Q: You have readied the development roadmaps for the next 20 years. Can you reveal your mid-term plans and goals?
A: In terms of revenue ratios for Macronix's three major product lines, ROM has dropped to 30% at present from the past peak of 70-80%, and NOR flash will stay at a higher level than ROM. Now NAND flash will be the major growth driver, with its revenue contribution ratio to pick up significantly after 3D NAND flash chips enter volume production.
Our technology R&D will be focused on two major parts, one for technologies that can be used at the current stage and another for those usable 20 years later.
In the next 20 years, our initial goal is to take good care of main clients and then compete with other major peers. But over the long-term, we will focus more on the development of system solutions. For instance, we will develop memory-centric solutions that can leverage 3D memory chips to store big data and conduct memory computing, so as to sharply reduce power consumption in the data transmission process. The solutions will be rolled out in the next 5-10 years as a very important secret weapon for Macronix.
Copyright © 2019 CST, Inc. All Rights Reserved