Hi and welcome to this special open to all February 12, 2021 episode of Peter Lee’s China Threat Report.
It’s all about the geostrategics of semiconductors.
It’s titled: Taiwan’s Silicon Shield Collides with its Silicon Lance.
This episode is open to all because I benefited greatly from knowledgeable commentary on twitter and the web while I was working on this story. You know who you are, even if I don’t, and I wish to say thank you and also invite your thoughts, comments, and corrections.
So, on with the show.
There’s a story about a Chinese arms dealer from the second century BC.
He was touting his catalog of weapons, including a lance that was invincible and a shield that was impenetrable.
A skeptical customer asked, well, what happens if somebody using your invincible lance attacks somebody with your impenetrable shield? Who wins?
The arms dealer didn’t have an answer, and the expression “lance and shield”—“mao dun” “矛盾” in Chinese became the common term for “contradiction”.
Well, Taiwan is facing a contradiction between its “Silicon Shield” and its “Silicon Lance”.
Taiwan, as is now widely known, is wildly successful in a key link in the electronics supply chain: the silicon foundry.
The silicon foundry doesn’t design chips but it makes them for customers who either can’t or don’t want to make them themselves. The experience, specialized knowledge, and tens of billions in investment to manufacture chips at the bleeding edge is immense and as a result the commercial foundry business has become more and more concentrated.
Today, half of the world’s foundry chip capacity is controlled by one company: the Taiwan Semiconductor Manufacturing Corporation or TSMC.
It supplies household names like Apple. Key players like AMD, Qualcomm, NVIDIA, and Intel.
And, until last year, it provided chips to the flagship telecom company of the People’s Republic of China, Huawei.
Twenty years ago, Taiwan’s semiconductor industry was characterized as the island’s “Silicon Shield”: a technical and commercial resource valuable enough to the corporations and governments of the West that they would be inclined to defend Taiwan against the People’s Republic of China.
But in 2018, with the election of the independence-minded DPP government in Taiwan and the triumph of China ultra-hawks in the Trump administration, the Silicon Shield was refashioned as a Silicon Lance: an offensive weapon to attack the PRC.
The first rumbling was felt with the announcement of an export ban on US sales to Huawei’s little brother competitor, ZTE, for violations of the Iran embargo.
The Taiwan government thereupon notified Taiwan chip suppliers, Mediatek and Nanyang, that their sales to ZTE would be subjected to a license requirement. This, as it turns out, was a rehearsal for the big one: a ban on TSMC chips sales to Huawei.
Another notable techwar incident was Taiwan and the US cooperating to nail Fujian Jinhua, the PRC’s great hope for domestically produced memory chips, for colluding in the theft of some old chip designs from the US chip manufacturer Micron.
The partner to the alleged theft was Taiwan’s second largest foundry, UMC, which was Fujian Jinhua’s technical partner.
Taiwan law enforcement assisted the FBI by wiretapping and raiding UMC to collect evidence. The Trump administration pitched in with a ludicrous accusation that Fujian Jinhua’s production of low-end memory chips for domestic consumption was a threat to US national security.
Thereupon, US sales of equipment and materials were banned, Fujian Jinhua and its $5.6 billion investment went bust, and 200 UMC engineers left the mainland to return to Taiwan.
What Taiwan got out of it, presumably, was the chilling effect on UMC and a signal that cooperation with the mainland on semiconductors, whether via corporate partnerships or poaching of engineers, would receive the hostile attention of the Taiwan government, backed up by its big brother, Uncle Sam.
What the United States, or at least China hawks in the United States, got out of it was the closing of the Taiwan to mainland semiconductor technology drain, a drain that seemed to be growing under the previous KMT administration, and one that threatened to undercut US efforts to suppress the development of the PRC semiconductor industry.
In other words, in coordination with the United States’ tech embargo and decoupling agenda, Taiwan shut down the Silicon Expressway to China and plugged a hole in the Silicon Shield.
Then came the big one, Taiwan supporting a US ban on chipsets to Huawei by TSMC, and that indeed came down in 2020.
As a result, Huawei lost its position as the leading manufacturer of smart phone handsets in the world and, indeed is on track to lose its smart phone business entirely.
Huawei sold off its low end smartphone line to a group of investors, and the new company, Honor, apparently is not subject to US bans, is using US chips, and will soon gain access to Google’s Android operating system.
Perhaps there is hope that the Biden administration will relax its ban on sales to Huawei for the other handset business. But Reuters claimed—and Huawei supremo Ren Zhengfei denied—that Huawei is in discussions to sell off its flagship phone lines as well to a Shanghai consortium.
So Taiwan had a pretty good two-year run as America’s Silicon Lance.
It reduced the Taiwan chip knowhow brain drain to China; it further set back the PRC’s faltering efforts at import replacement for semiconductors by a few years (and a few billion dollars); and it did a good job of blocking Huawei’s rise to global telecoms pre-eminence by gutting its handset business while the United States went after Huawei’s carrier infrastructure business around the world.
But a funny thing happened to the Silicon Shield while Taiwan was eagerly wielding the Silicon Lance.
Let’s say, the Silicon Shield got too big and heavy for Taiwan to carry.
The Taiwan semiconductor blockade of the PRC had a number of consequences, some of them foreseen and some not foreseen.
The big foreseen consequence was China funneling enormous investments into its own semiconductor manufacturing industry to reduce its reliance on Taiwan.
In the general press targeting the low information crowd, China hawks sometimes spin this as “China thirsts for world domination in semiconductor manufacturing and must be stopped.”
The exact opposite is true. China buys a lot of chips and desperately hopes to buy fewer chips, not export them.
I can assure you, you don’t have to worry about the PRC grinding TSMC—or Intel, or Samsung, or any first-line semiconductor manufacturer—into the dust.
Despite some sizable investments, the PRC’s efforts in semiconductors are low-end and floundering, thanks in good part to US hostility to equipment and technology transfer to China at the high-performance end of the semiconductor manufacturing industry. a hostility that predates the Trump administration’s high profile assault.
Cratering Fujian Jinhua was just one element of America’s war on the PRC semiconductor industry.
Trump took it to the next level, of course, by targeting the PRC’s “wannabe TSMC”, that’s Semiconductor Manufacturing International Corporation or SMIC, for its alleged military links.
In order to shut SMIC out of the high performance semiconductor game, the US prevailed upon the Dutch equipment manufacturer, ASML, not to ship SMIC one of its most advanced extreme ultraviolet lithography machines, which is indispensable to make the most itty bitty and cutting edge semiconductors.
Even though SMIC had contracted for the machine and paid for it.
And just before leaving office, Trump put SMIC on the entities list, which requires US export licensing for shipment of the most modern US semiconductor equipment.
But the U.S. embargo isn’t the only factor.
For China to have a hope of becoming a major player in all semiconductors, including the highest end stuff, would require the active support of the current cutting edge semiconductor companies in Taiwan, the United States, and/or South Korea…or their engineers, like Fujian Jinhua tried with UMC.
A few years ago, the PRC was creeping into the higher end semiconductor space thanks to its usual combination of big spending, talent poaching, IP theft (alleged), licensing, and leveraging its immense domestic market to play divide and rule and lure or spook foreign semicon companies to set up local manufacturing joint ventures.
Burgeoning Western fears about China’s economic rise—and its efforts to move into the higher-value-added product space—have made it possible for China hawks to basically slam that window shut.
China triumphalists may object, but in my opinion China simply does not have the game—the money, the human capital, or access to technical assistance and equipment—to catch up to TSMC on the high end in the foreseeable future.
When the TSMC ban hit, there was a flurry of reports on the PRC making tens of billions of dollars of investments into domestic semiconductor capacity.
Take the announcement that the PRC government’s National Integrated Circuit Industry Investment Fund had another 204 billion yuan—that’s $32 billion US dollars—to invest in domestic capacity.
Sounds like serious money, doesn’t it?
Then consider TSMC’s capital budget for 2021: $28 billion dollars. Samsung, the other big foundry operation, will put in about $22 billion in 2021. Intel: another $14 billion.
That’s over $60 billion dollars expanding capacity and extending the technology lead for three companies that are already working in the 3 nanometer space while China is launching a kinda sorta 7 nanometer chip that’s really a 10 nanometer chip with benefits.
What I call “resistance chipmaking”—that is to say the PRC manufacturing high-performance chips that the West doesn’t want it to make, is going to lag years if not a decade behind the West.
China knows it, TSMC knows it, America knows it, so the specter of China Inc. rising to the challenge of the Taiwan embargo and conquering the Western semiconductor manufacturing world is, shall we say, a paper tiger.
In economic terms, every yuan spent trying to develop a high end Chinese domestic semiconductor industry in the teeth of Western resistance is a yuan wasted: money that could have gone into other segments—artificial intelligence, chip design, panda eugenics, whatever—with a hope of a positive return.
Probably the best use of China’s money is importing as many fancy chips as it can from people in the West who can make them quicker, cheaper, and better.
China will remain a huge market for foreign chip and component suppliers for the foreseeable future.
China imports $300 billion a year (yes, $300 billion US dollars a year, you heard that right; it’s China’s single biggest import item and represents 60% of global chip consumption)…yup, $300 billion in computer chips of varying function and complexity.
Which is actually good news for China because, as the recent squealing from the US semiconductor industry to the Biden administration indicates, the tech industry’s lust for China sales is one of the bigger reason why economic engagement with China is trumping the China hawks’ decoupling strategy.
And Chinese strategists and economic planners know that. Or they should.
But that doesn’t mean TSMC is safe, or that the Silicon Shield is intact.
Just the opposite, in fact.
Weaponizing TSMC as the Silicon Lance—in other words using TSMC as a geopolitical tool against the PRC, in other words, instead of a responsibly managed global good-- had major and apparently unforeseen and unintended economic and geopolitical consequences that are weakening the Silicon Shield.
The world’s reliance on Taiwan as a sole source supplier of high end tech stuff is attracting unfavorable attention now thanks to the Great Chip Drought, a shortage of chips impacting industrial users around the world and apparently putting the global economic recovery at risk.
Today’s chip drought is being sold as a Covid-related “nobody could have foreseen” thing but the understory appears to be that the Taiwan chip foundries were happy to indulge their biggest and most profitable customers: the telecoms handset manufacturers…
…including, ironically, of course, Huawei…
…at pretty much the expense of everybody else.
The US government telegraphed its ban on TSMC sales to Huawei early enough that Huawei went on a buying binge in 2020 to stockpile chips and try to keep its businesses afloat.
China was expected to import $300 billion dollars worth of chips in 2020; instead it imported $380 billion worth.
It looks like TSMC also tipped off its most important customer, Apple, that Huawei orders were tying up production; so Apple went on its own stockpiling binge, and tied up even more of TSMC’s capacity.
Who got the short end of the stick in terms of accessing TSMC product? Well, it looks like pretty much everybody else but most importantly auto manufacturers.
The 2020 chip drought is expected to shut down production lines and cut auto industry revenues by $60 billion, or 10%. Ford and GM each expect earnings to drop by $2 billion.
The one outlier: Toyota which says it has four months worth of chips and you can wonder if this is another example of superb Toyota foresight, or perhaps a sign that Taiwan gave the most important manufacturer in its most important Asian partnership a little heads up.
This fiasco has, of course, given pretty much every country with an auto industry the idea that, instead of relying on a couple of foundries in Taiwan that basically want to make chips for Apple above anything and anyone else, semiconductor customers should strive for national self-sufficiency in semiconductors.
Taiwan did not help its case in Europe, I think, by seeming to dangle a quid pro quo of European covid vaccines in return for Taiwan bumping European car manufacturers up the TSMC priority list.
Chip shortages with a side order of vaccine extortion? Pinch me, I’m dreaming!
The EU as a group and Germany have targeted localized semiconductor manufacturing capability in response to the chip drought’s impact on the vital automobile industry.
Not good for Taiwan, of course, or the “Silicon Shield” strategy.
One Taiwanese analyst highlighted the whole conundrum in a rather desperately optimistic way, saying:
"For Taiwan to maintain its 'silicon shield,' it needs to persuade the European countries and the U.S. that keeping TSMC in Taiwan is the best option,".
Good luck with that, as they say. Judging by this February 11 report, things are going exactly the other way. The headline reads:
TSMC—and Taiwan, I expect—are not enthusiastic. Per the piece
A TSMC spokeswoman said the company considered a variety of factors, including customers’ needs, when determining the location of a new factory. There were no concrete plans for cooperation with the EU at present, though the company was open to the possibility…”
It’s not just those China-appeasing Eurocrats.
Another headache for Taiwan is that even its key allies, the United States and Japan, appear to be prioritizing semiconductor self sufficiency over maintaining Taiwan’s precious Silicon Shield.
Last year, TSMC rather grudgingly promised to put a $12 billion plant in Arizona, supposedly to ensure the supply of strategic semiconductors to the US military; and the Japanese government has reportedly offered tens of billions of dollars ofsupport to try to lure TSMC to build a plant in Japan, and so far has only been able to score an R&D facility.
It will be interesting to see if Taiwan and TSMC are able to sustain their enthusiasm for the Arizona project now that Trump is gone and Joe Biden, Taiwan’s second choice, occupies the White House.
The tone of a recent Reuters headline implies that Taiwan is going to play it cool with Team Biden, at least for now:
I’m guessing Team USA disagrees, since President Biden is preparing an executive order outlining America’s efforts to get out from behind the chip shortage 8-ball.
TSMC will also have to deal with the fact that its leading foundry competitor, Samsung, is planning a $17 billion chip plant in Texas, assuming it can extort the necessary concessions from the local government.
The US Semiconductor Manufacturing Association is also responding to the chip crisis by pushing for government assistance to boost US domestic production, currently languishing at 12% of global capacity.
This would seem to be bad news for the Silicon Shield.
However, TSMC can regard the spectacle of the world begging for chips and scrambling to line up manufacturing alternatives with a certain equanimity.
After all, the rest of the world is kinda like China when it comes to scraping together the time and money and personnel and political will needed to match TSMC’s enormous first mover advantage.
Intel, the company that is supposed to be America’s chipmaking champion, apparently totally f*cked up its move into the 7 nanometer space and is talking about outsourcing some of its high end production to, you guessed it, TSMC.
With competitors playing catchup, the most dangerous near to medium term enemy for TSMC Taiwan is probably TSMC USA: the possibility that for economic, political, and security reasons the United States will insist on screwing Taiwan by moving TSMC’s core industrial capabilities to the USA.
You know, just like we did with the Japanese auto industry back in the 1980s.
But maybe not.
Today, with the China containment drive, Taiwan’s well-funded outreach inside the Washington Beltway, and the Pentagon’s desperate desire to resurrect Taiwan as America’s unsinkable aircraft carrier off the coast of China, I wonder how much political traction a demand to reshore the semiconductor industry to its ancestral home in America will gain.
One would guess that the timeline for TSMC to feel the competitive heat—with help from some strategic footdragging by Team Taiwan’s friends in defense ministries around the world—is 7 to 10 years.
Which means ten years of Taiwan as a vital global interest and ten years for the world to acquiesce to Taiwan independence, a Taiwan mutual defense treaty with the US and Japan, a NATO tieup, and so on and so forth…
…before the PRC’s conventional military advantage in the Taiwan Straits becomes unassailable.
All good for Team China Hawk planners.
Ironically, the People’s Republic of China may hold the trump card.
Recent events have focused attention on the vulnerability of the global economy to disruptions in the supply of semiconductors from Taiwan; it looks like it would be in China’s interest to highlight and feed those anxieties.
The cutoff of TSMC chips has reduced the PRC’s incentives for strategic restraint vis a vis Taiwan; and a heightening of tensions across the Taiwan Straits may fuel international demands that the Taiwan semiconductor foundries, as a geostrategic and economic asset of global importance, be moved off the island to safer ground.
In which case, TSMC ceases to be Taiwan’s Silicon Lance or Silicon Shield.
Instead, it becomes a Silicon Bullseye.
That’s all for this episode. Thanks for listening, reading, and supporting Peter Lee’s China Threat Report. Stay well, stay safe…and see you next time.