In a milestone for China’s semiconductor industry, Yangtze Memory Technologies (YMTC) announced last week that it has developed a 128-layer NAND flash memory chip (128L) in-house. The company expects mass production to start sometime between the end of 2020 and mid-2021, a spokesperson for YMTC told TechNode. 

The Wuhan-based firm hit this milestone while fighting to continue production during the lockdown of its home city.

Read more: What industry can’t stop? Semiconductors 

As wafers hit surface area limits, space on them is like downtown real estate: it comes at a premium. Layering circuits allows chipmakers to fit more memory into the same space—building up instead of out. 128L puts YMTC on the cutting edge of flash memory, but scaling up to mass production to match its competitors will be challenging.

Flash memory is used in products from “entry-level” USB and memory cards, to more complicated solid-state hard drives. YMTC’s current generation of 64L memory has its foot on the lowest rung of this ladder.

Samsung, Micron and SK Hynix hit the initial production milestone in 2019, and started selling their in-class chips in early 2020. YMTC’s product could compete with them, but comes six months to a year behind the competition. 

It is an important step on China’s path to semiconductor independence, but the fact that YMTC has managed to stack 128 layers of circuits on a wafer won’t necessarily make it a big player in the global semiconductor industry, experts said. There are several hurdles that YMTC needs to jump through in order to compete with incumbents in quality, scale, and price. 

Analysts said that YMTC’s previous NAND chip was hardly a wild success. “Because YMTC has just begun selling 64L NAND products, and because of the impact from COVID-19, the actual sales figure remains low at this point,” Avril Wu, a semiconductor analyst at Taiwanese market research firm TrendForce, told TechNode. 

YMTC has not released any information as to how many units of the 64-layer memory chip it has sold, and did not reply to TechNode’s request for data. TrendForce expects YMTC to account for 8% of the global flash memory market in 2021.

The timing is right for YMTC to launch the 128L chip, as innovation from some competitors is likely to slow. The price of NAND Flash fell by an average of 46% in 2019, leading to losses, conservative capital expenditures, and record-low output growth expectations, TrendForce said.

Manufacturing difficulties

For YMTC to compete with international peers, memory chip production standards will matter as much as design. One measure used in the industry to gauge quality is yield: the proportion of chips on a wafer that work properly. 

In the best case, YMTC will become a big player in the global flash memory chip game. In the worst case, Randall said, its clients won’t evolve past China.

“YMTC lags behind other mainstream memory manufacturers in terms of yield and product stability,” Wu said. She added that “it is actively bridging this competitive gap.” More than a matter of design, yield is affected by the production process. Companies refine the manufacturing process as engineers gain know-how in making a particular design. 

Market analyst Wu said the “primary hurdle” is the procurement of manufacturing equipment. The billion dollar machines that are used to produce chips are made by few companies in the US and Europe, like Dutch ASML and American LAM Research. They take months to produce and have long waiting lists, which is why usually there is a months-long lag between announcing a product and bringing it to market.

“In the future, if the US government prohibits European and US equipment suppliers from shipping to YMTC, it will negatively affect the company’s capacity expansion schedules,” Wu said.

The issue of experience and know-how is important for scaling production as well, James Lewis, Senior Vice President and Director of the Technology Policy Program at US think tank Center for Strategic and International studies, told TechNode. 

“It’s not foreign sources for semiconductor manufacturing equipment that is the obstacle,” he said. 

The 64L’s yield was reportedly “good enough,” said Stewart Randall who heads the electronics and embedded software department at Intralink, a consultancy that provides market entry services to China, told TechNode. This is a positive sign for the 128L’s yield, but its production is harder. “Let’s see how the 128L does,” he said.

With a little help from a friend 

In all likelihood YMTC will manage to scale up capacity and mass produce its 128L flash chip in 2021, analysts said. But the scale at which this production happens is crucial to the economies of scale that allow companies to offer competitive prices. Given the lack of know-how and equipment, it will take time for YMTC to match the offers of incumbents in price and quality. 

But the Wuhan memory chip-maker has a powerful friend holding its hand. It is funded by government-backed conglomerate Tsinghua Unigroup. Beijing’s Big Fund, focused on promoting the development of homegrown semiconductors, raised $29 billion last summer. Tsinghua Unigroup received the most state funding out of all semiconductor players in the world between 2014 and 2018 in a December report published by the Organization for Economic Construction and Development.

As a strategic company that isn’t listed, YMTC and Tsinghua Unigroup don’t need to churn profits the same way that its competitors do. The state-owned company is likely willing to bankroll losses in order to help a Chinese semiconductor company establish itself in the market

“Neither financial nor human resource factors are issues for YMTC,” Wu said. Backed by Tsinghua Unigroup, all it needs is time to make a dent in the flash memory market. TrendForce said that YMTC’s mass production of 128L is likely to drive down prices for the industry overall. 

Beijing has other ways to help YMTC, but it must strike the right balance. “The temptation will be for the Chinese government to press companies to give YMTC preference, but this works only if the chips are competitive in price and performance,” Lewis said. 

Foreign companies could relocate rather than buy an inferior product from YMTC, Lewis said. This policy is “a bit touchy now, as the government doesn’t want to encourage foreign companies in China to leave” during the Covid-19 pandemic, he said. 

YMTC’s clientele is overwhelmingly made up of Chinese companies, but it also works with Phison Electronics, a Taiwanese company that packs flash memory chips into controllers for USBs, memory cards, and SSDs, sources told TechNode.

A big step for YMTC, a small step for China

The development of the 128L flash memory chip is an accomplishment. Founded in 2016, the company has managed to come head-to-head with decades-old players like Samsung on one  crucial aspect of semiconductor design: stacking circuits on a wafer. It is big news for the Wuhan-based firm, but it is only a small step in China’s efforts to achieve self-reliance in semiconductor production and manufacturing. 

The 128L wafer will allow YMTC to up its game, from memory cards and USBs into solid-state drives for computers. The fact that it has developed its own chip architecture, called Xtacking, bodes well for future intellectual property conflicts, Wu said.

But memory chips are some of the easiest integrated circuits to produce, the “low end of semiconductor technology,” Lewis said. Chinese companies have yet to make significant progress in designing more advanced chips, like graphic processing units, and rely on western companies. China’s semiconductor industry might soon be supplying memory components to the globe, but it will continue to import all the other chips that computers are made of from the rest of the world. 

Eliza Gkritsi

Eliza is TechNode's community listening reporter at the Shanghai office. She acts as a link between the editorial team and TechNode Squared members. She previously worked as a reporter for WikiTribune...