China’s semiconductor quest is likely to fail, leaving rapprochement with US the only way out
- China does not have a good track record when it comes to developing its own chip industry, even with the help of foreign technology partners
- Country’s chip makers must invent new ways to design and make semiconductors of equal or better performance than products from TSMC, Samsung and Intel
One of the biggest battles in the ongoing US China tech war is over semiconductors – the enabling technology behind everything from smartphones to earth-orbiting satellites.
However, efforts by the US and China to satisfy their respective national security concerns will likely end in failure.
China must reduce its near-total dependency on American chip tech, but it does not have a good track record when it comes to developing its own chip industry – even with the help of foreign technology partners.
Its first attempts date back to the 1980s when European electronics companies were invited to set up joint venture wafer fabs in Shanghai. A second try in the 1990s also failed, partly because China’s top down bureaucracy was too rigid for the fast moving semiconductor industry.
And while SMIC is considered China’s national chip champion today, it is entirely dependent on foreign technology. What chance is there of achieving self-sufficiency when starting from scratch and with no foreign technical help?
Some have compared today’s push for semiconductor self-sufficiency to Mao Zedong’s “two bombs, one satellite” program that began in the mid 1950s. If it could be done then, why not now?
Then, as now, foreign technical help was needed. In the case of the A-bomb, the Soviet Union sent several thousand of its nuclear scientists and engineers to China. Their help was so critical that when the Russians pulled out in 1960, Mao arranged for song-and-dance girls to ply the Russians with alcohol so their scientific notebooks could be copied secretly before they left, according to Mao biographers Jung Chang and Jon Halliday.
For China, fully decoupling from US semiconductor technology will require a herculean effort over many years, if not decades, but it is already off to a shaky start. Thousands of Chinese companies – most with zero experience in the technology – have jumped on the chip bandwagon in recent months, eager to tap government funding or hoping to get rich from an IPO on the new STAR market.
Eventually this bubble will burst and the serious players can begin the real work. But they must invent new ways to design and manufacture semiconductors of equal or better performance to the best circuits coming out of overseas players such as TSMC, Intel and Samsung.
China’s biggest challenge won’t be money – it will be finding qualified people. Poaching a few hundred semiconductor engineers from Taiwan might be enough to ramp up a new chip fab in China, but that is far short of what will be required to completely rebuild a semiconductor ecosystem.
Finding “a handful of geniuses” – an approach China has used in past national programs – probably will not work either. A completely new mindset may be required to come up with an alternative to the existing silicon-based chipmaking process. Materials like graphene and gallium nitride are often cited as potential replacements for silicon.
For its part, the US needs to bring semiconductor production back home but even with rare bipartisan support, the dysfunctional federal government could be a major hindrance. There is also the question of who could lead a revival in US wafer fabs.
The last time the US faced such a crisis was in the 1980s when Japan came close to taking the American semiconductor industry away, as it had done earlier in consumer electronics. The response was SEMATECH, a government funded consortium of chip makers and their suppliers, but cooperation was hobbled by a reluctance to share information among rivals.
It took the leadership of industry legend Robert Noyce – co-inventor of the IC, Intel co-founder and SEMATECH’s first chairman – to convince all sides to work together for the good of the country. Nobody can fill Noyce’s shoes today.
Intel, for decades the US national chip champion, has now fallen behind TSMC as the global chip technology leader – a major wake up call that in part prompted the bipartisan call for chip funding.
In May, TSMC agreed to build a new wafer fab in the US state of Arizona, supported by federal and state subsidies. While that is a significant first step, it will not be producing wafers in volume any time soon and as a foreign-owned operation still may not ease national security concerns.
There is a cheaper alternative for the US – but it is also risky. That is, take steps to “secure” supply from TSMC’s wafer fabs in Taiwan, which already serve the needs of the major “fabless” US companies like Qualcomm, Nvidia and AMD, as well as consumer product giants like Apple.
There are already moves in this direction, with a bipartisan group of senators writing to the US Trade Representative Robert Lighthizer asking him to start negotiating a trade deal with Taiwan, citing the need to reduce reliance on mainland China and diversify supply chains.
Whether such a move would also require an end to Washington’s policy of ambiguity in the Taiwan Strait is unclear. But efforts to bring Taiwan closer into the US orbit are complicated by China’s territorial claims on the island. Miscalculations on either side could even lead to a war.
If China overcomes the odds and succeeds in its quest for chip self-sufficiency, it will shift the global balance of power. If it fails, which is the more likely outcome, there is only one alternative: rapprochement with the US.
Craig Addison is a production editor on the Post’s tech desk in Hong Kong. From 2002 to 2009 he worked for SEMI, the Silicon Valley-based trade group representing semiconductor equipment and materials companies.