The Coming Chip Wars

A version of this article appeared in War on the Rocks.

 

Controlling advanced chip manufacturing in the 21st century may well prove to be like controlling the oil supply in the 20th. The country that controls this manufacturing can throttle the military and economic power of others.

The United States just did this to China by limiting Huawei’s ability to outsource its in-house chip designs for manufacture by Taiwan Semiconductor Manufacturing Company (TSMC), a Taiwanese chip foundry. If negotiations fail, China may respond and escalate, via one of many agile strategic responses short of war, perhaps succeeding in coercing the foundry to stop making chips for American companies – turning the tables on the United States.

Short of war, there would be no obvious way to get those foundries back. Without them, the U.S. defense and consumer electronics industries will be set back at least five years — and because China has its own advanced chip foundries, it could become the world leader in technology for the next decade or more.

Here’s why.  And how they may do it.

And why the world just got a lot more dangerous.


There are two types of companies in the chip industry.

  1. Companies like Intel, Samsung, SK Hynix and Micron design and make their own products (microprocessors and memory chips) in factories that they own
  2. There are also foundries, which fabricate chips designed by consumer and military customers; TSMC in Taiwan is the largest of these in the world

The chips that TSMC makes are found in almost everything: smartphones (i.e. Apple iPhones), high-performance computing platforms, PC’s, tablets, servers, base stations and game consoles, Internet-connected devices like smart wearables, digital consumer electronics, cars, and almost every weapon system built in the 21st century. Around 60% of the chips TSMC makes are for American companies.

Background
In 2012, a bipartisan committee of the U.S. House of Representatives investigated whether the Chinese company Huawei had put backdoors into its equipment that enabled it to spy on data therein. The committee found that Huawei could not or would not explain its relationship with the Chinese government and did not comply with U.S. laws, The report recommended that no government or contractor systems include Huawei systems. In 2019, the U.S. Department of Commerce’s Bureau of Industry and Security added Huawei to its Entity List, effectively limiting the sale or transfer of American technology to the company, (though a series of licenses have been granted to waive the restrictions in some cases.)

This month, the Commerce Department required overseas semiconductor firms that use American technology and equipment to apply for a license before selling to Huawei. The order was targeted at TSMC, which is Huawei’s main supplier of advanced chips; without these, Huawei will be at a competitive disadvantage against Apple or Samsung in the smartphone industry, and against Cisco and others in the market for network equipment. (Some analysts have pointed out the order has potential loopholes.) Next up, it’s likely Washington will prohibit sales to China of the equipment used to make chips, which comes from companies like Applied Materials, KLA and Lam.

TSMC was forced to choose sides and picked the U.S. – For Now
In May 2020 TSMC announced it was going to build a $ 12 billion foundry in Arizona to make some of its most advanced chips. Foundries take at least three years to build and the most expensive factories on earth. Construction on TSMC’s facility is planned to start in 2021, but actual chip production will not start until 2024.

But while the TSMC announcement is welcome, if and when the Arizona foundry is built, it will only be able to make about a quarter of the chip production of TMSC’s largest semiconductor fabrication plants and would amount to just 3 percent of the manufacturing capability that TSMC currently operates in Taiwan. There they have four major manufacturing sites, called GigaFabs, each of which have 6 or 7 fabs producing thirteen million wafers a year. Compare that to the quarter million wafers they intend to produce in the U.S. in 2024. So if the United State lost TSMC in China, one new American plant would not make up the difference in capacity.

China’s Semiconductor Industry
A decade ago, China recognized that its initial success as the world’s low-cost factory was going to run its course. As the cost of Chinese labor increased, other countries like Vietnam could fill that role. As a result, China needed to build more advanced and sophisticated products on par with the United States. However, most of these products required custom chips — and China lacked the domestic manufacturing capability to make them. China uses 61 percent of the world’s chips in products for both its domestic and export markets, importing around $ 310 billion worth in 2018. China recognized that its inability to manufacture the most advanced chips was a strategic Achilles Heel.

China devised two plans to solve these problems. The first, the Made in China 2025 plan, is the country’s roadmap and financing vehicle to update China’s manufacturing base from making low-tech products to rapidly developing ten high-tech industries including electric cars, next-generation computing, telecommunications, robotics, artificial intelligence, and advanced chips. The goal is to reduce China’s dependence on foreign technology and promote Chinese high-tech companies globally. In addition, to encourage Chinese high-tech companies to go public in China rather than the United States, the Chinese government set up its own version of the Nasdaq called the STAR market (Shanghai Stock Exchange Science and Technology Innovation Board).

China’s second plan is the National Integrated Circuit Plan, China’s roadmap for building an indigenous semiconductor industry and accelerating chip manufacturing. The goal is to meet its local chip demand by 2030.

Make no mistake, these are not government pronouncements that don’t end up going anywhere. This is a massive national effort. China is spending over a hundred billion dollars to become a world leader in developing their semiconductor industry. The China Integrated Circuit Industry Investment Fund or Big Fundhas raised $ 51 billion – $ 22 billion in 2014 and another $ 29 billion in 2019. China has used the capital to start 70+ projects in the semiconductor industry (such as building fabs and foundries, acquiring foreign companies, and starting joint ventures) and have gone from zero to making 16% of the world’s chips, though today their quality is low. Going forward, China plans to start investing in chip design software, advanced materials, and semiconductor manufacturing equipment.

How Do the Chinese View Our Actions?
China believes that this is their century and sees American actions as designed to hold China back from its proper place in the world. Given the importance of controlling the supply of advanced chip manufacturing, China would be forced to respond if the United States cut off their access to this supply.

The question is whether China will view the action against Huawei as sanctions against a single company or a portent of further action against China’s access to advanced chips.

What Has China Learned From Our Prior Actions?
In the 21st century the U.S. has blinked even when its own interests were at stake. From the perspective of some China policymakers, America is exhausted from endless wars in Iraq and Afghanistan and will not fight again. They see that the United States is divided politically, distracted by the COVID-19 pandemic and unlikely to risk American lives for something as abstract as a chip factory.

Paper protests
When China has acted aggressively over the past couple of decades, it has seen that the American response has largely been paper protests. In 2012 China occupied the Scarborough Shoal and took control of it from the Philippines. As China was not ready to militarily confront the U.S. at the time, in hindsight the U.S. could have parked a carrier strike group over those shoals and likely prevented their plans for military construction. Instead, Washington blinked and did nothing but send a nasty note.

Today, the Spratly Islands have new Chinese bases bristling with surface-to-air missiles, cruise missiles and fighter jets, which has changed the calculus for a war in the western Pacific. Any attempt by the United States to control the air space in the area will face serious opposition and heavy losses. What was previously an uncontested American “lake” is now contested by China.

Up until this week Hong Kong, while part of China, was a democracy with guarantees of freedom of speech, assembly and the press. China recently tore up that agreement and is preparing to impose the same draconian limits on speech, assembly and press that muzzle the rest of China. There’s not much the U.S. can do other than express concerns and perhaps remove Hong Kong’s special trade status. But China doesn’t care. They’ve already factored the American response into their move and decided it was worth it, with the cynical calculation that any U.S. response will make Hong Kong poorer, and that any business Hong Kong loses will mostly end up in other parts of China. And a poorer Hong Kong will be punishment to its citizens for standing up for the rights they had been promised.

The day after China’s move on Hong Kong, Chinese Premier Li Keqiang left out the word “peaceful” in referring to Beijing’s desire to “reunify” with Chinese-claimed Taiwan, an apparent policy change.

The lack of an effective American response to these events has shown Chinese leadership the unwillingness of America to forcefully engage in Asian affairs. This will embolden China’s next move.

China’s Goals and Options
To respond to the United States cutting off Huawei’s access to Taiwan’s most advanced chip foundries, the Chinese government is likely thinking through their next moves. Their planning starts with they want to accomplish. It may look something like this in the preferred order.

  1. Return to the Status Quo – Restore Huawei’s Access to TSMC fabs to secure a steady supply of chips
  2. Don’t let the restrictions escalate
  3. Turn the Tables – Convince TSMC/Taiwan to allow China to have sole access to TSMC
  4. Kick Over the Table – Ensure that the TSMC fabs can’t be used by anyone

China’s Options
So how would China achieve these goals?

China may wish to avoid any escalation perhaps by accepting the American restrictions as they currently are with a promise that they will go no further.  This return to the status quo, with a restoration of Huawei’s access to TSMC’s foundry, may simply require negotiating some form of trade deal or agreeing to restrictions on the sale of Huawei networking gear (34% of their revenue). This kind of deal would let the Huawei consumer and enterprise businesses (66% of their revenue) survive and thrive. However, it requires the Chinese to back down. And they may have decided that the Rubicon has been crossed.

If China doesn’t negotiate but retaliates, the danger is that the United States ups the ante further by prohibiting TSMC from working with more Chinese firms, and/or bans the sale of the equipment used to build chips to any company in China. Such escalation may lead China to perceive that the U.S. actions are not a dispute about Huawei, but a salvo in a wider economic war.

If it gets to that point, China’s plans no longer are how to negotiate with the U.S. but how to force TSMC to do its bidding. And as TSMC is in Taiwan, in what China claims is a province of China, things can get interesting.

The most obvious option is to simply carry out the threat the Chinese government has made since 1949: that there is only one China, and Taiwan is a rebellious province, and that they will reunify China, by force if necessary. An invasion or blockade of Taiwan would give Chinese hardliners a reason to try out all their new military equipment, while distracting the masses from the pandemic economic downturn. This option has the highest risk of provoking an American military response, and while possible it’s extremely unlikely. While these more aggressive scenarios might seem implausible, China’s behavior has become more aggressive and more risk-tolerant as the COVID-19 pandemic, which began in Wuhan, roils the world.

China can achieve their immediate goals of 3 and 4 above and weaken Taiwan without an outright invasion.

One option is a major disinformation campaign against TSMC and the United States that would make current influence campaigns emanating from China pale in comparison. This would emphasize that the U.S. is the aggressor, illegally waging economic war against China. It would announce that since Taiwan is a province of China, China has the right to restrict TSMC sales to the U.S. and that China ill enforce an embargo of any TSMC sales to American-affiliated companies.

This could be coupled with an equally massive disinformation campaign to the Taiwanese people, pointing out to them that the United States won’t go to war over a semiconductor company, and that China’s requestsare fair and reasonable. (How effective a disinformation campaign would be is up for debate, given that Chinese campaigns in Taiwan’s January elections did not result in the election of China’s preferred candidate.) China could offer a no-invasion pledge in exchange, while reminding the Taiwanese government what they already know: regardless of promises the United States can’t defend them. Even if the United States attempted to intervene, there is a serious debate unfolding about how useful legacy American platforms – especially carriers – would be in a shooting war with China.

There’s a high probability Taiwan will still refuse despite all of this, so China would then ratchet up the pressure.

China might then start some type of trade war with Taiwan to ensure access, following the playbook Beijing used to coerce Korea over Terminal High Altitude Area Defense (THAAD) or Australia over its recent decision to lead a call for investigating the origins of the novel coronavirus. On the more extreme end, these Taiwanese chip foundries might be subject to an aggressive campaign of sabotage.

Finally, they could nationalize TSMC’s two less advanced fabs in mainland China. Next, if there’s no agreement, China could launch a precision guided missile strike against one of the older, less advanced TMSC fabs in Taiwan to send a message they’re serious.  They could announce they’ll destroy one foundry each week until TMSC agrees to sell only to China. Even if they destroy all the TSMC foundries in Taiwan it will still be a net win for China. It’s highly unlikely Taiwan would go to war with China over this. The end result would be that U.S. military and consumer technology would have no advanced foundries, but China would.

What Would the United States Do?
Would the United States go to war with China over chips? The loss of TSMC would mean we’d be rapidly scrambling to find alternate sources. We could turn to Intel to restart their foundry business or turn to Samsung or even Global Foundries. But the transition and recovery would take at least three to five years if not more and tens of billions of dollars.  In the meantime, we’d have second-tier status in technology.

The outcome could depend on the timing of Chinese actions.

When Might China Take Action?

An October Surprise – Before the 2020 election
The current U.S  administration may not want to start a war over a chip factory before the 2020 presidential election, but it is unpredictable enough that a campaign season focused on China policy could change the calculus.

After the 2020 election
If the presidency changes hands, the incoming administration might de-escalate and reverse original restrictions, but a lot can happen between now and January 2021.

A Trump administration in its second term and no longer worrying about reelection might reverse the ruling in exchange for a better trade deal.

Downside: Lots of economic uncertainty for the next seven months exacerbating China’s pandemic recovery. More immediate action might be required.

Lessons Learned

  • The dispute over Huawei’s access to TSMC has highlighted how vulnerable American industry is to the loss of its sole supply of advanced chips.
  • If the matter cannot be solved by negotiation, China may perceive the restrictions as economic warfare and rapidly escalate, potentially threatening Taiwan
  • It is not at all clear that Washington has thought through the consequences of its actions here, or that the current administration has considered chip supply as part of a wider supply chain security and national industrial policy.
  • Given that China has more positive options than the United States, it is surely time for those in charge to consider where this might lead

Steve Blank

Trillions are at stake in the retirement wars, and Vise nets $14.5M from Sequoia to manage it

The retirement wars are heating up.

As millions of baby boomers leave their jobs in the coming years and transition into retirement, there is a huge competition for who will manage their savings. On one hand are traditional wealth managers, firms like Edward Jones, who either employ full-time human financial advisors or empower independent contractors to help clients plan through their finances. On the other side has been the rise of “roboadvisors” like Wealthfront that use algorithms and simple financial products like ETFs to advise people at lower cost.

VCs have been bullish on roboadvisors — startups like Wealthfront and Personal Capital have each raised more than $ 200 million according to Crunchbase — but there has been less investment activity trying to help the financial advisors themselves. After all, aren’t all these folks supposed to be automated away by algorithms?

Vise (from “advise”) is taking a bit of a contrarian bet: its founders Samir Vasavada and Runik Mehrotra believe that humans — augmented with the right AI tools — can prove even more adept at handling the financial affairs of their clients than an app.

The company debuted at TechCrunch Disrupt SF last year, and we wrote up an in-depth profile of its journey from self-funded startup to our stage. Well, according to the founders, it just so happens they met Sequoia at the firm’s Disrupt happy hour, and one thing led to another and Vise is now announcing a $ 14.5 million Series A term sheet led by Sequoia partner Shaun Maguire.

Previous investors including Keith Rabois through Founders Fund and Ben Ling at Bling Capital filled out the round, and the startup’s total fundraise haul is now at $ 16 million.

For the founders, the main goal for Vise has been to build a new product using the best practices from the AI and machine learning worlds and converge on a platform that helps independent financial advisors come up with their own ideas to communicate to clients. “Our big thesis was, we want to think about things that are different in this industry — we don’t want to build a product that’s the same as how every other product has been built in the space,” Vasavada said. “We want to build a radically different product, and the way in which we do that is bringing in a diverse team.” That’s included everyone from product folks at notable Silicon Valley companies, AI researchers, and financial services experts.

Vise’s platform. Photo courtesy of Vise.

Financial advisors already rely on a suite of software from CRMs to investment analysis platforms to perform their jobs, but those tools have rarely been integrated into one place. That’s made the existing market for software here quite fragmented. “Number one is it’s too bloated. There’s just too many tools and they don’t do enough and don’t provide much value add. It’s expensive. It’s hard to manage. And the most important thing is it is not at all personalized to the advisor or personalized to the client,” Vasavada said.

Instead, Vise aims to be a one-stop shop for all the needs in the daily workflow of an investment advisor. That includes determining different investment options in a clean interface, personalizing those options for individual clients, and even helping guide investment advisors through the talking points on why certain investment decisions make sense compared to others given a client’s context.

Vise founders Runik Mehrotra (L) and Samir Vasavada (R). Photo via Vise.

In their views, Vasavada and Mehrotra see the wealth advisory market dividing into several buckets, with independent wealth advisors who target $ 500,000 to $ 2 million in assets per client as the sweet spot for Vise. Those customers have more specific needs and require more personalization than clients with less assets and so are ill-served by roboadvisors, while at the same time, major institutional players find them too small to handle given the fee structures they have at their scale.

Ultimately, Vise is a pure B2B play, and the founders want to maintain that focus into the future. They believe that wealth advisors have special knowledge of their clients needs and the relationships to match, which Vise can’t compete with.

In addition to Sequoia, Founders Fund, and Bling, Human Capital, Lachy Groom, Steve Chen, and Jon Xu joined the round according to the company.

Startups – TechCrunch

Things to do in Pittsburgh this week, from Star Wars trivia to Fantastic Fungi – NEXTpittsburgh

Things to do in Pittsburgh this week, from Star Wars trivia to Fantastic Fungi  NEXTpittsburgh
“startups when:1d” – Google News

These 15 companies could be acquisition targets for Amazon, Microsoft, and Google as the coronavirus crisis ratchets up the cloud wars and crushes tech valuations – Business Insider

These 15 companies could be acquisition targets for Amazon, Microsoft, and Google as the coronavirus crisis ratchets up the cloud wars and crushes tech valuations  Business Insider
“startups when:1d” – Google News