Nasdaq-listed Taiwan-based Silicon Motion Technology is a leading supplier of NAND flash controller chips, long quietly driving the upgrades of the global flash memory sector with technology crucial to expanding the solid state drive (SSD) market. Also known as SMI by its marketing brand for NAND controller chips, the company now commands a 30% global controller market share each for eMMC/eMCP embedded memory and client SSD, 25% for flash memory card, and 20% for USB, company president and CEO Wallace Kou told Digitimes in a recent interview.

Kou revealed that SMI is actively proceeding with deployments in the enterprise-grade SSD market and has completed several related reinvestment projects including acquiring China's emerging enterprise SSD supplier Shannon Systems and Taiwan's software-defined storage solutions provider Bigtera. He also shed lights on the prospects of the NAND flash industry.

Q: China is aggressively developing its storage industry supply chain and certainly will not skip the segment of NAND flash controller chips. What strategies will your company adopt to meet potential challenges and tap huge business opportunities in China?

A: SMI started its deployments in the China market 15 years ago as a supplier of NAND flash controller chips, and has since grown together with our China customers and partners. Of course, we have been developing the huge China market in a planned way and with clear visions.

In the industry-spec storage area, for instance, we have tapped into the supply chains of the China auto industry, and after acquiring Shannon Systems in 2015 we have managed to foray into the enterprise SSD sector in China. Through Shannon Systems, we have gradually learned the characteristics of the enterprise SSD market, the real needs of clients, the core of potential issues and how to address the issues, helping to deepen our deployments in the market there.

Q: Acquiring Shannon Systems seems a crucial turning point for SMI's operations. Would you talk more about that company?

A: Shannon Systems is a supplier of enterprise SSDs, and its performance has been better than expected since the acquisition. The deal was completed three years after our first contact with each other, as all the terms matured then, in that Shannon badly needed a NAND flash technology specialist as a cooperative partner and that its customer base could help SMI tap the enterprise SSD market in China.

In formulating strategies for developing the enterprise SSD market, SMI has sensed the need to adopt a brand-new business model: venturing into the hardware system end rather than only offering controller chips for the devices, so as to facilitate easier survival in the market.

In fact, most founders of Shannon Systems used to work at US tech firms such as Marvell and Nvidia, and so we fortunately share highly similar corporate cultures, goals and technological development directions. Prior to the acquisition, Shannonn Systems adopted costly FPGA controller chips for its SSD devices.

Q: In developing the enterprise SSD market, how will SMI handle its coopetition with upstream chipmakers?

A: SMI has maintained close cooperation with major NAND flash suppliers including Intel, Micron, WD, Samsung, SK Hynix, Toshiba, and Yangtze Memory Technology, with the cooperative scope covering SD memory cards, USBs, embedded product lines, client SSDs, automotive electronics and eMMC, among others.

These chipmakers also have their own SSD controller technology, but they are keen to zero in on large-volume products, instead of customized markets or customers requiring small-volume and large-variety production, which are exactly our specialty and opportunity.

SMI maintains customer-supplier ties with NAND flash makers on many product lines. We will not compete with those upstream memory suppliers in the enterprise SSD segment, and will manage to find out our favorable niche-type market position.

Q: SMI has developed rapidly in the enterprise-level storage field. What about the current progress and future goal in this respect?

A: We will kick off volume production of the first controller dedicated to enterprise SSDs in the third quarter of 2018 using 28nm process. In addition, as open-channel SSD will become the mainstream product in the enterprise SSD segment in the next 3-5 years, we will start our deployment in such SSD later this year, with volume production likely to kick off first for large-size China Internet customers.

Q: How will you extend the software-defined storage platform after acquiring Bigtera in 2017?

A: Bigtera specializes in developing software-defined storage solutions, and will be gradually vertically integrated with SMI and Shannon Systems. As Bigtera sustains good cooperation with system operators, it can easily spot problems at the system end or learn customer demand for new functions, which can be passed as feedbacks to SMI for reference in further enhancing its technological development of controller chips.

More and more large-size enterprises need the service of predicting the life cycle of their storage systems, as they have little knowledge about the traits of NAND flash or how the controller chips work in their SSDs. In this respect, SMI can help Bigtera gauge and predict the duration of storage systems at enterprises to materialize more business opportunities.

Software-defined storage technology involves huge business opportunities, particularly in China, where markets are quite regionalized as different provinces and cities have different system integrators (SIs). Bigtera has also set the goal of entering the enterprise market, covering a large range of customers including state-run enterprises, TV stations, telecom operators, financial institutions, and more. In combing its software solutions and hardware systems of SIs, Bigtera can fully understand the exact needs of customers and what controller chips can better meet their needs. This way, SMI can access more enterprise customers and enhance its capabilities of addressing problems.

Q: The combination of deep learning and edge storage is emerging as a new application trend. How do you see this trend?

A: Computing technology was centered on CPU 30 years ago, but the current architecture features memory as the core. Furthermore, instant data processing is increasingly needed, making edge computing a hot topic.

A driverless car, for example, should have data computing done immediately instead of sending the data back to the cloud for processing. Such a car will have at least 60-300 sensors, and the data they receive require immediate edge computing and judgment. In this respect, memory installed should be able to predict and determine advance division of computing labor. Accordingly, memory controllers and AI chips will not be absent in the edge computing storage era.

For one more example, Wi-Fi will soon be applied to open household door locks, and controller chips will play a crucial role in accelerating fast data computing and judgment to determine when the doors can be opened, even if with the availability of AI-based facial recognition technology.

Q: How do you assess the growth potential for the SSD market? Has your company set any goal as to market share?

A: The growth pace of the SSD market will mainly hinge on the price movement trend. In the medium to long term, two trends can be well observed. The first is that after 96-layer QLC (quad-level cell) chips kick off volume production, the capacity of any single chip will expand significantly, creating new room for price reduction. And major NAND flash makers will focus on mass production of 96-layer chips for a long while, as it will take time for them to advance from 96 layers to 128 layers. The second is that the supply scenario will be changed after China's Yangtze Memory Technology starts production of 3D NAND flash memory products. Both trends will serve to drive down NAND flash prices.

On the demand side, traditional HDDs will see their market penetration rate shrink to only 10-15% within five years amid the rapid growth of client SSDs. In terms of enterprise storage market, the demand for high-performance SSDs will also see fast expansion, but NAND flash SDDs cannot replace high-capacity HDDs for cold storage of ever-increasing archive data, as the latter are much cheaper than the former.

At the moment, Samsung remains the top supplier in the client SSD segment, once commanding a global market share of up to 50%. SMI now has a 30% share and sets 40% as its ultimate goal.

In the enterprise SSD market segment, Samsung and Intel are top players, and SMI is just about to enter the market. We hope we can become a significant supplier in the enterprise SSD segment.

In the next 3-5 years, I hope SMI can develop into a more balanced and comprehensive supplier of storage solutions, providing more advanced storage technologies to create more industrial values and contribute more to the upstream NAND chipmakers and the entire industrial chains.