Billions of Dollars Later, China Chipmakers Gain Ground but Still Trail Global Leaders
3 to 5 years behind
I’m not usually one to jump on China’s “flavor of the day” high-tech bandwagon, as I do think most of the fluff that gets tossed around lacks substance and is manufactured simply to win public favor. But all the chatter about microchips — and China’s deficiencies and aspirations in the area — finally spurred me to look for some substance on this very topical high-tech matter.
These chips first came under the spotlight last year when Washington banned China’s No. 2 telecom equipment maker ZTE from buying such U.S.-made high-tech components as punishment for violating sanctions against Iran. It came back into focus in May when sector leader Huawei got slapped with a similar ban, even though many suspect that move’s timing was designed to give Washington more leverage in its year-old trade war with China.
The topic was back in headlines this week when Unigroup, one of China’s top aspiring chipmakers, announced its formation of an internal group focused on the manufacture of DRAM, one of two types of memory chips. Techies will know the other main kind of memory chip is NAND, and that the global market for both types is currently dominated by the likes of South Korea’s Samsung and Hynix, as well as U.S. giant Micron.
Observers said the brief announcement reflects the broader memory aspirations of Unigroup, which co-founded the promising NAND-making operation in Yangtze Memory Technologies back in 2016 with a couple of chip-oriented investment funds.
A lot has been said about China’s chip-making ability in a very general way without too much detail. Many point to HiSilicon, the chip-making unit of Huawei, as an exemplary case of how China could quickly play catch-up in the space. But there’s seldom much additional color on what that means. With all the verbiage flying around, I decided to try to identify exactly who are the Chinese players to watch, and how far they trail their foreign peers.
I’ll step back a moment and take a look at China’s broader chip development over the last two decades before delving into specifics. When I first arrived in the region in 2002 to cover the sector, the industry was still in its infancy.
A few homegrown names like SMIC, Grace Semiconductor and Huahong NEC captured most of the headlines back then, mostly focused on the “contract manufacturing” task of making chips for bigger name companies. Most of the big foreign names like Intel and Micron also had manufacturing presences here, though those were largely confined to low-end chip test and assembly work.
Most of those early efforts didn’t really go far, and it wasn’t until around five years ago that Beijing finally decided enough was enough and it wanted to develop its own industry. Anyone who has spent any time here knows that when something like that happens in China, watch out. Sure enough, billions of dollars suddenly became available for such investment from government-linked sources, led by the $22 billion China Integrated Circuit Industry Investment Fund set up in 2014.
As with everything in China when it comes to emerging high-tech sectors, the flood of government investment has yielded patchy results so far.
The big problem is that whenever such massive funds become available, everyone and their mothers suddenly become experts in the field and start clamoring for handouts. The result is usually huge amounts of wasted spending for facilities that are poorly planned and ineffective.
In this case the analysts I talked to, one at Gartner and the other at TrendForce, both agreed that China is still at least three to five years behind global leaders when it comes to memory chips. The picture is a little better for telecom chips that power smartphones and wireless devices, with both research houses particularly praiseworthy of Huawei’s HiSilicon chip unit.
Let’s begin with memory, which is a huge sector due to the presence of DRAM and NAND chips in quite a large array of devices in all shapes and sizes. As I’ve said already, Yangtze Memory got kudos from both analysts for its relatively rapid progress in NAND in the three years since its founding. Still, Gartner analyst Sheng Linghai told me the company is still at least three years behind the world’s most cutting-edge NAND makers.
For DRAM the picture is even bleaker, and China is probably at least five years behind the rest of the world. TrendForce couldn’t point to any viable Chinese DRAM-makers at the moment. Gartner’s Sheng said ChangXin Memory Technologies, based in East China’s Anhui province, could be one company to watch. But even he wasn’t particularly impressed just yet.
The pair were more impressed with HiSilicon for its telecom chips that act as the central “brains” for smartphones and other mobile devices. Both Gartner and TrendForce said the company’s chips are already not far from those made by Qualcomm, the undisputed global leader for such high-end chips.
TrendForce added that HiSilicon had already passed Taiwan’s MediaTek, often considered a lower-end runner-up to Qualcomm. Both also pointed to Unigroup’s Unisoc, formerly known as Spreadtrum, as a company to watch. But even in that case, there was some feeling the company had failed to reach its potential despite earlier big hopes.
At the end of the day, China really does seem to have produced a relatively thin crop of potential global challengers when one considers the huge sums spent. What’s more, the most promising player, HiSilicon, is really coming from the private sector if we believe that Huawei is a private company. But all that said, it really only takes one or two companies to dominate specific kinds of microchips, which the world has already seen from the small field of players now controlling the space.
There are problems with your assessment, the recent ban had me looking at the Kirin chipset (HiSilcon) and in terms of grunt speed Qualcomm has an edge, but in terms of actual output Kirin slaughters it, because of the optimization of software and the way neural networks are used. The developments in 5G and bluetooth also step through, while the dual GPS is a long awaited development in its own right
The move to Hongmeng OS is the most important aspect in this and cannot be removed from the chip design. Nor can the link between the camera arras and the Neural networks on the chip. The chips run cool, and with a significant less power because of it, the distributed cores combining low power slower RISC cores doing housekeeping rests high speed hot cores. It is impressively smart in its macro design, the problem is micro design where it is reliant on ARM which has now proved untrustworthy.
Here the core design aspects have to mastered and made into production ready template tools for actual burns. China once Taiwan is included is sitting well to catch up, and despite the best efforts of the US to create frictions between the too the economic links between the mainland and province are increasing (hence the US campaign in Hong Kong, attempting to make political integration less appealing).
When it comes to capital investment, China has plenty of problems, but far less than several years ago, scammers are plentiful, but so are worker owned enterprise, and when experts come together, they tend to form the later. This is an extra element that cannot be neglected, productive capital in the west is almost a joke, but in China national investment with all the problem it has had and will have, has proved extremely effective in making China the largest actual economy in the world (not one whose books are inflated with money printing and counting debt as asset).
However your estimate of being 2-3 years behind before leading the industry is I think spot on. China has the ability, and Taiwan should jump now. “If we believe that Huawei is a private company” It is not a private company, it is incorporated under Chinese Law and is worker owned, that cannot be compared to only aiming at the quarterly profit margin of a ‘public company’ in the west.
China’s semiconductor fab capacity to reach 20% worldwide share in 2020
don`t underestimate the Chinese industrial capacity and speed……since the hardware is done…..what remains is now only the software…..the chips are soon the highest level in its industriy…..the two tech world will be soon only one only……the Chinese way….with OS systems far more sophisticated than the american…..adding to the declice of the empire….
fact is…..the ignorance of the US industrial elites in the US is the driving engine in this development…..the idiocy of its politicians is perfecting it….
“act is…..the ignorance of the US industrial elites in the US is the
driving engine in this development…..the idiocy of its politicians is