Not everything is about the sexy microprocessor
or GPU in your desktop. Memory is one of the major categories in the
semiconductor world. The memory industry as it currently exists
is dominated by a number of incumbents. The reason for this is because memory is a
brutal and cyclical industry. But it can also serve as a critical starting
point for many semiconductor companies to nurture their fabbing expertise. The industry’s dynamics also make it easier
for new, low cost entrants to sell in the market. One of those new entrants into the market
is Wuhan-based Yangtze Memory Technology Company. The company began as a low cost supplier but
has impressively scaled the value chain in recent years. In this video, I want to look at one of China's
two leading memory companies and one of its flagship semiconductor champions. Memory chips are semiconductors same as the
fast Intel processors inside your computers. But the Intel chips are classified as "logic
chips", meaning that they perform mathematical operations. Memory chips on the other hand store data. There are two types of memory in the market:
DRAM (Dynamic Random Access Memory) and Flash memory. DRAM refers to the 8GB or 16GB in your MacBook. This memory can only store data temporarily
- it vanishes when the power is turned off - but in return you are able to access it
much faster. Flash memory is the type of memory that is
in your USB stick or serves as the faster-access data storage medium in your computer. It is capable of retaining its data even after
the computer's power is turned off. China's memory market is massive. You cannot make a computer or a smartphone
without having both of these types of memory. Well I guess you can make a computer with
a spinning hard drive, but it’s certainly hard to make one without DRAM. Out of the $150 billion of semiconductors
that the country imports each year to make their products, memory represents 35% of that
amount. So it is a huge market. Despite this, domestic memory makers held
just 3% market share in 2018. The reason for this is that memory chips are
a cheap commodity that at the lower end competes on price. It is easier to fab and the product has little
differentiation from other low cost competitors. Scale, technical mastery, and market position
are what matters in this industry. And Chinese memory makers lag far behind. The global NAND flash market is dominated
by a few companies, mostly clustered in Asia. (Same goes for the DRAM market.) The market incumbents are Samsung (South Korea),
Kioxia (Japan), Western Digital (US), Micron (US), and SK Hynix (South Korea). The DRAM market is almost literally just Samsung,
SK Hynix and Micron. Competing against these giants, Yangzte Memory
is the most advanced and promising domestic memory company in China. Their specialty in producing NAND memory is
impressive. In my video about Tsinghua Unigroup, I mentioned
Yangtze Memory in passing. Here we are going to flesh out the story a
little more - a story that actually intersects with another video of mine. The company that is now Yangtze Memory begins
with SMIC, currently China's leading independent semiconductor foundry. Isn’t that cool? It’s like a Marvel movie universe! Everything’s connected! Anyway, I go into it more in my SMIC explainer
video, but SMIC's operating style throughout its wild early years was to strike deals with
local city governments in China so to get state funding for its expensive fabs. Wuhan had been party to one of these deals. In order to get results fast, SMIC set up
Wuhan's foundry to build memory. But the thing about the memory market that
I already mentioned is that though it can be relatively easy to enter, for that same
reason it is also a brutal market short on profits. SMIC's Wuhan factories endured years of losses. Uncomfortable about these seemingly unending
operating losses, Wuhan terminated the SMIC partnership and established their own independent
foundry: Wuhan Xinxin Semiconductor (XMC). In 2008, an American company called Spansion,
a flash memory joint venture between AMD and Fujitsu, decided to outsource some of its
foundry work to Wuhan Xinxin. In 2014, the two companies expanded their
partnership and allied to develop 3D NAND. 3D NAND is an advanced version of memory chips. Older generations of memory are called "planar
NAND". They are flat, 2D structures. 3D NAND on the other hand looks like skyscrapers
and can have many dozens of such layers. It allows for much more storage in the same
amount of space. Samsung first commercialized this technology
back in 2013. 3D NAND is difficult to fab at scale, but
memory firms need to do it in order to produce products that can compete at the higher rungs
of the market. Achieving it is indicative of a firm with
potential in the market. Likely due to several promising advances in
the 3D NAND field, in July 2016, semiconductor holding company Tsinghua Unigroup purchased
a majority stake in Wuhan Xinxin Semiconductor and established a holding company to govern
the company. This holding company was named in reference
to its location on the Yangtze River: Yangtze Memory. Tsinghua has been interested in the semiconductor
and memory market for a long time. The company is the owner of Unisoc, one of
China's largest fabless companies. They had also unsuccessfully attempted to
purchase shares in American memory giants Micron Technology and Western Digital. When that failed, the company turned inwards. Then-Tsinghua CEO Zhao Weiguo vowed to invest
$100 billion over the next 10 years to turn Yangtze Memory into a world-class domestic
memory chip company that can compete with the likes of Samsung. Yangtze Memory's ability to be competitive
in the memory market depends on it scaling the technological sophistication ladder. And more than just announcing a breakthrough
on a press release, Yangtze's manufacturing yields need to be good too or else the company's
cost structures cannot compete in the global market. In 2019, Yangtze Memory announced the mass
production of a type of 3D NAND with 64 layers using a 20nm process. This is a critical breakthrough that instantly
puts the company in the same realm as the market leaders. Furthermore, they could produce 60,000 wafers
each month by the end of 2020. This would be close to Intel's estimated 85,000
wafers per month capacity, but is far behind Samsung's 470,000 monthly wafers. Intel later sold their NAND flash business
to SK Hynix for $9 billion in late 2020. In April 2020, they announced their next big
step: That they have developed a 128-layer 3D NAND memory chip. Again they expect mass production to begin
sometime in the following year. If and when that happens, then Yangtze is
going to bring a near-leading edge product into the NAND market. Some of Yangtze's product will go to manufacturing
partners. Makers and sellers of laptops and the like. But the company is also selling its SSD products
directly to consumers under the Zhitai brand. Are people buying Yangtze's products? We do not know. The company is not public so they don’t
need to report how consumer sales are doing. But considering that NAND prices hit a down
cycle in 2019 and COVID has been throwing things into a twirl throughout 2020, it is
likely that sales are starting off slow. Which is fine. Much of Yangtze's great memory advances are
on the NAND side. When it comes to DRAM, the temporary memory
inside your computer, Yangtze is still behind the current incumbents. The company is competing with various other
Chinese companies in the space for this, but nothing solid yet. DRAM is harder to manufacture than NAND due
to several required extra steps, but not impossible. We should probably look to anticipate some
announcement in the coming years. The government has strongly gotten behind
Yangtze Memory and in April 2018 President Xi visited the Wuhan factory. He gave a speech there, in which he said that
in order to achieve China's 200-year goals, some major core technologies must overcome
difficulties on their own. Such a visit shows official approval at the
highest levels of political leadership. The current flash memory incumbents are bracing
for the anticipated entry of Yangtze and other Chinese memory national champions into the
global markets. Within 1-2 years, the company is probably
going to surpass Micron in terms of production capacity, flooding the market with this new
supply. Furthermore, Yangtze's announced goal of eventually
getting to 300,000 wafers a month adds 10-20% to current supply and will cause disruptions,
assuming the product is good. It would not be the first time it has happened. Chinese companies entered the LED and solar
cell markets and crashed margins for the incumbents. In the case of the European solar incumbent,
margins fell from 25% to zero in just three years. Even before Yangtze’s entrance, the NAND
market was pretty competitive, with five major players. Despite this, the incumbents have enjoyed
high margins, preventing the industry from consolidating. The recent sale of Intel’s NAND memory business
though is indicative of the future. The memory world is bracing for Yangtze Memory's
rising presence.