Chinese Chip Designers Can’t Meet Mandated Goals Without US Technology
If you haven’t heard, the United States and China are locked in a trade war. One of the major issues in play on the United States side of the equation is the problem of forced partnerships and IP theft that US companies often report as the price of doing business in China. A great deal of sophisticated technology manufacturing is done in China — but according to chip designers themselves, hitting Beijing’s aggressive goals for local production will be impossible if the US starts refusing to cooperate.
There are alternatives in China, but the gap in technology is too big,” an executive from one of China’s leading artificial intelligence chipmakers, which relies on U.S. technology for chip design, told Nikkei. “If we lose access to U.S. software or can no longer receive updates, our chip development will run into a dead end.”
His thoughts were echoed by executives from NextVPU, an AI firm started by ex-AMD staff, and by others, including a representative from the Semiconductor Manufacturing International Co, or SMIC, mainland China’s leading, most-advanced foundry. “We would use whatever chip equipment and materials we have locally if their performances were good enough,” said a manager from Semiconductor Manufacturing International Co., China’s top contract chipmaker. “But we still need [American] equipment, materials, IPs and chip design software. It’s not likely for any of the chipmakers in the world… to get rid of American vendors soon.”
This graph from IC Insights makes it clear just how lopsided the market for chip design firms truly is.
SMIC, for example, is currently focused on ramping early 14nm production, with plans to build a $10B foundry to focus on 14nm for local customers. Assuming that facility is online and fully spun up by 2022, that puts Chinese IC production roughly five years behind the rest of the world as far as node leadership and seven to eight years behind as far as volume production. But according to reports, China’s overall self-sufficiency rate in semiconductor production is only about 15 percent, due to the globalized, fragmented nature of the world supply chain.
The Made in China 2025 plan calls for 40 percent of Chinese semiconductors to be built in China by 2020 and 70 percent by 2025. According to experts, the current expected achievable target is 20.5 percent by 2023. According to Nikkei, one of the barriers China faces is that its own developers are more reluctant to rely on locally produced technology or to pay the higher prices associated with smaller-scale production from local fabs. That’s another piece of this puzzle — companies don’t perceive the Chinese product as being equivalent, so they’re less likely to purchase it.
All of these preferences and positions matter, so long as China has some route of buying US-built semiconductors. Should those routes close altogether, it would be a different story. Left with no choice but to rely on its own expertise for semiconductor design, China would undoubtedly lag badly, probably for years. But such a move by the United States would also spur dramatic investments from the Chinese government, which would likely feel it had no choice but to spend money to replicate those pieces of the technology puzzle it could no longer purchase elsewhere.
This is the central risk of the entire affair, at least from a commercial business perspective. US companies that spend billions developing new technology do not want to hand it over to Chinese firms as part of the cost of doing business. At the same time, however, they do not want to be locked out of these markets altogether. Freezing the Chinese out of the global semiconductor market by choking their access to American technology will not result in the Chinese returning to the days of vacuum tubes (though I wouldn’t mind seeing the world’s first and only vacuum tube-powered cellular phone). It could, however, kick off a series of geopolitical event that led to other countries agreeing to move away from US technology firms and sources if they came to regard us as untrustworthy, uncertain partners.
Historically, it’s extremely difficult to keep technology away from dedicated nations that wish to acquire it. The US was unable to prevent other countries from inventing the atomic bomb after WW2. Firms that enjoy strong market conditions and near-monopolies on software distribution may enjoy those conditions for decades, but even in these sclerotic markets, change eventually rears its head. Microsoft and Intel defined the PC industry from the late 1980s through the late 2000s, but now share space with a variety of computing devices based on other operating systems and running different CPU architectures. Stuxnet — one of the more sophisticated malware efforts ever created — successfully delayed Iran’s nuclear ambitions, but it didn’t singlehandedly prevent them.
Ironically, the impact of the export control decisions the United States is making right now is as unknowable as the conditions that govern other long-term bets that semiconductor manufacturers make. As Robert Palmer, former Digital CEO once reportedly said: Designing microprocessors is like playing Russian roulette. You put a gun to your head, pull the trigger, and find out four years later if you blew your brains out.
It’s still possible, of course, that the US and China will agree to resolve their trade disputes and Chinese companies will regain access to US technology. Should this fail to occur, however, we’ll be in Russian roulette territory as far as the long-term consequences. No country can afford to be completely cut off from technology and semiconductor markets, and that means China would have little recourse but to build its own knowledge base and parts, possibly as part of a broad international coalition it might create with partners afraid of being treated similarly by the United States.
There’s no doubt that refusing to work with the Chinese would dramatically slow Chinese technological development — for a while. Whether that works out to a long-term advantage for the United States — that’s the question, isn’t it? Because part of the argument in play here is that the Chinese will recognize that they can’t currently live without access to US semiconductor technology and will, therefore, be appropriately incentivized to play ball. But it’s also possible that the Chinese, even if forced back to the table for the moment, lay their own long-term plans to move away from the United States’ technological dominance altogether. None of that means the US’ stated goals in the ongoing trade war are less important or valuable, but it definitely speaks to how complex resolving problems like this can be.
Feature image by SMIC.
- US Bans AMD’s Chinese Joint Venture From Developing, Selling Hardware
- Report: Tariffs Could Raise Laptop, Phone Prices 22 Percent
- Western Digital Freezes Out Huawei, Which Braces for Massive Sales Drop