Interview

«The US and China Have Now Moved into a Full-Blown Economic War»

Jonathan Goldberg, Partner at Snowcloud Capital and Founder of D2D Advisory, talks about the recent escalation in the tech standoff between the US and China, explains what it means for the Chinese semiconductor industry, what strategic role Taiwan plays, and what it all implies for the outlook on the semiconductor sector.

Christoph Gisiger
Drucken

The situation in the semiconductor sector is tense. On October 7, the U.S. government drastically tightened sanctions against China’s semiconductor industry. Stocks of equipment suppliers ASML, Applied Materials, Lam Research and KLA are under particular pressure, with losses ranging from 14% to 19%.

«With the news this week of the US Government’s latest rounds of restrictions on China’s semiconductor industry it is hard to escape the conclusion that the two countries have now moved beyond a trade war into a full-blown economic war», says Jonathan Goldberg.

The industry expert knows what he is talking about. Mr. Goldberg is a partner at the venture capital firm Snowcloud Capital in Silicon Valley and the founder of D2D Advisory, a consulting service specializing in the electronics and semiconductor industry. He also has a deep knowledge of China, where he lived for about ten years. His blog «Digits to Dollars» is a must read for anyone interested in the most important developments in the chip sector.

In this in-depth interview with The Market NZZ, which has been edited for clarity, Mr. Goldberg says what he thinks of the recent escalation in the tech standoff between the U.S. and China, what it means for the Chinese semiconductor industry, what strategic role Taiwan plays, and what it all means for the outlook in the semiconductor sector.

«If it’s a choice between losing Taiwan and leveling TSMC, China will have no problem leveling TSMC»: Jonathan Goldberg.

«If it’s a choice between losing Taiwan and leveling TSMC, China will have no problem leveling TSMC»: Jonathan Goldberg.

Source: JG

Mr. Goldberg, how do you read the most recent US sanctions against China’s semiconductor industry?

There is so much in there: More Chinese companies on the entity list, more things that are banned. The US is basically prohibiting China to buy or build semi cap equipment for advanced nodes, or access the software needed to use other people’s leading-edge processes. At the broadest interpretation, the sanctions shut off China’s semiconductor industry from advanced semiconductor processes.

What does this mean for the tech standoff between the US and China in general?

It’s clear that the primary objective is to limit the ability of the Peoples’ Liberation Army to obtain advanced technologies. That makes sense from a national security standpoint: China is an adversary, and the US government wants to curtail its capability. But there is a very close relationship between the commercial sector and the military in China. In fact, many countries are like that. The US government is collectively one of the biggest customers for companies like Intel. So if the goal is to shut down the PLA’s ability to get chips, how do you do that without wrecking the broader Chinese semiconductor industry that’s 90% commercially used? That’s very hard to do. Semiconductors are a big industry in China, there are 2000 chip companies in China. Therefore, it’s going to be very tricky, unless it’s very broad and everybody gets cut off – and that is a very provocative, large escalation of this conflict.

You know China’s tech sector from years of experience. How do you assess the US government’s restrictions from a historical perspective?

It feels like these rules were thought up during the US-Soviet Cold War. During that era, the US and the Soviet Union operated largely siloed economies for seventy years, with very limited inter-dependencies. By contrast, the US’s and China’s economy are co-dependent, and the latest restrictions do not seem to fully conform to, or even recognize, that reality.

What are the direct consequences of the sanctions for companies in the chip sector?

I’m sure hundreds of lawyers are reviewing them as we speak. It’s a lot of complexity, and these changes are so big that it’s going to be a few months before we really see how things end up. I have a lot of questions around foreign companies with plants in China like SK Hynix and Samsung. In the last few days, the US government has been giving out waivers to let them keep doing business in China, but it’s unclear what that allows. Equipment companies have already started to pull their employees out of the country. That’s probably just in the Chinese owned fabs, but if I ran one of those companies, I would be worried about what business I can do in China. It’s also going to be interesting to see which Chinese companies will be allowed to use TSMC’s foundry services going forward. And, there is a big question around software and IP which is a big part of semiconductors today.

What does this mean for SMIC, China’s largest and most advanced semiconductor foundry?

That’s the big question. SMIC is able to mass produce 14 or 16 nanometer nodes, but it’s going to be very hard for them to get past that, because to get really past 14 nm you need ASML’s extreme ultraviolet lithography technology. It’s very clear that China is not going to get these machines from ASML. That begs the question: Can China build its own EUV machines? Well, part of the sanctions limits the components needed to build EUV systems. I’m sure that in some academic center in China there is somebody who can build a very limited EUV system. But that’s not commercial. Maybe they can do a wafer a day, but TSMC does hundreds of thousands of wafers a week.

How will China respond to the sanctions?

The restrictions are so big and the consequences potentially so large, they will have to respond somehow; even if takes a long time for that response to become apparent. I certainly think that this is going to accelerate, strengthen and reinforce China’s ambitions to become self-sufficient in semiconductors. But it’s not as if they were lacking in will before October 7 when the new sanctions were announced. So this is just going to reinforce what they are already believing.

Does China’s chip industry have any chance at all of catching up to the leading edge?

I do not think they can do it. It will take them ten years at best. They have already spent $100 billion or $200 billion over the last ten years, and they haven’t narrowed the gap to TSMC. Much of what TSMC and ASML do revolves around process and management. Chinese foundries have been poaching TSMC employees for years, but there is no easy way to transfer all the soft skills that underpin semiconductor manufacturing. But who knows. I have been going to China for a long time, and one thing I learned is to never underestimate their ability to catch up to an existing technology. Then again, EUV is very challenging. It’s super hard to do and highly expensive. Even competent, advanced companies struggle with it. Look at Intel. They were once the leader, and they completely stumbled around EUV implementation.

YMTC, China’s largest manufacturer of memory chips, has now also ended up on the entity list. Why?

YMTC is a big target in these new sanctions. They have their own fabrication plant for NAND memory chips which is somewhat different, because you can do advanced NAND chips without EUV. In the past, we’ve always heard that YMTC was sort of «good enough». They didn’t have the best product, it wasn’t the fastest, most reliable, or densest, but it was «good enough», and it was cheap.

And today?

A few months ago, reports came out that Apple would put YMTC’s NAND chips in the iPhone. That’s a big deal because Apple cares a lot about NAND, it’s one of the biggest components in the iPhone. YMTC essentially built a plant just for Apple. On top of that, we started to see specifications for what YMTC is producing. And, it turns out it’s not just «good enough», it’s probably the best product out there, and it’s cheap and they’re throwing capacity around. So it’s really tricky, because it’s not clear how this is a military technology. YMTC doesn’t seem to be a big supplier to the PLA. Yet they are explicitly a target in the latest sanctions which certainly makes China think that this is all about shutting down the whole industry, that this is economic, not about the military.

Apple is now reportedly putting plans for this partnership with YMTC on hold. But here’s another question: What role do Taiwan and TSMC play in the crackdown on China?

I want to be careful, because it’s not a semiconductor question, it’s a geopolitical question. It’s clear that TSMC is the leader by far. Samsung is second and can do some of these advanced processes, but TSMC is significantly ahead. It’s an incredibly capable firm, and they’re stuck in the middle: They have to walk this fine line between China and the US. Obviously, they are under a lot of pressure to open fabs in other parts of the world, but they are not going to do that. The Taiwanese government is going to be very unwilling to let TSMC open major plants outside of Taiwan. TSMC is building a plant in Arizona, but it’s a tiny plant, and by the time it is opened it’s going to be a couple of years behind of what they can do technologically in Taiwan. There is a perception that the US has to care about Taiwan because of TSMC. The US government made it very clear that TSMC is a strategic asset for the US economy and the US military, so they have to care about what happens in Taiwan.

What does TSMC mean to China?

I don’t think the Chinese government really cares about TSMC. Of course, Chinese semiconductor companies want to have access to TSMC, but if it’s a choice between losing Taiwan and leveling TSMC, China will have no problem leveling TSMC. If we want to take the glass-is-totally-empty viewpoint, and if China’s government starts to feel that it is totally cut off from TSMC’s manufacturing, the idea of cutting everyone else off to level the playing field makes a grim amount of sense.

That sounds quite alarming.

We can speculate widely on risk case scenarios, but Taiwan matters to China in a way that’s different from the US. For the US, Taiwan is an economic and military strategic asset, for China it’s about national identity, it’s a very powerful symbol. I really hope it doesn’t come to war, but from an industry perspective we all have to keep in our mind that there is a non-zero chance that we lose access to TSMC.

For this reason, Western chip companies like Intel are receiving government support to build new factories. Do these subsidies make sense?

We should look at that from a commercial perspective and from a national security perspective. From a commercial perspective, Intel is in a bad place. They have stumbled, and they have fallen off the curve of Moore’s law. Typically, when companies do that, they don’t come back. But given enough time and money, they will catch up - probably. Yet, it’s very challenging, and it’s an immense amount of money. Intel is talking about raising $60 billion to $100 billion, and their CEO is going all over the world looking for every subsidy they can get. Also, time is not on their side because TSMC is not standing still. Intel said they will catch up and exceed TSMC by 2025, but there is no guarantee of success.

What are Intel’s chances of catching up technologically with TSMC?

We will see. Personally, I would not be surprised if they run out of money before they can catch up. So they might just split the company in two: Spin-off the foundry business and keep the design stuff. But the problem is when you spin off the foundries, whoever owns that business then has to spend $100 billion to catch up.

From a strategic perspective, how important is Intel to the US and to the West in general?

From a national security standpoint, Intel is incredibly important. I can tell there is a debate within the US government about how much support they want to give to Intel. The US government doesn’t like to subsidize companies directly, because philosophically that’s not what we like to do. But there are a lot of people inside the US government who say: We need Intel, we need that asset. It feels a little bit like in the Nineties when there were all these shipyards in the US that have been building nuclear submarines and aircraft carriers. Most of them had to shut down, but the government kept one or two in business just in case. So they are going to have to throw money at Intel to keep it in the game if worse comes to worse. It will be like the Manhattan Project: throw a bunch of money at it, and they will get something in time.

What role does Samsung play in this kind of strategic considerations?

Samsung is tricky. It’s the only other company capable of producing chips at advanced processes. But they’re not as good as TSMC, and they’re much harder to work with. TSMC is very focused on just one thing: They make other people’s chips. Their whole focus is on maintaining that process, and they have an organization that reflects that. Intel and Samsung are different. There are a lot of things that are going on inside Intel. There is the design side, they have a lot of subsidiaries. And Samsung of course is a chaebol that does everything.

What does this mean regarding the race for the most advanced chips?

EUV is so expensive, just one machine alone is $300 million. And then you have the operating cost. It’s like you need a power plant for all the electricity these things need. As a result, it tends to draw the attention of other corporate people: You have non-engineers forced to get involved in engineering decisions. And as a finance guy, I can say that you shouldn’t have finance guys making engineering decisions. That’s clearly what happened at Intel, and it looks like that’s what happened at Samsung too: The chaebol got involved and made decisions which slowed them down. So Samsung is there, but it’s not as clean a story as TSMC which is executing very well.

How will the sanctions affect equipment manufacturers such as ASML, Applied Materials, Lam Research and KLA?

The honest answer is we don’t know entirely yet. For all of these companies, China is their number two market, and it’s growing much faster than anywhere else. If China gets really cut off, it’s a big chunk of revenue they would lose. For instance, a big part of YMTC’s process relies heavily on KLA, and KLA just had to pull all of its people out of the YMTC plant. It’s a big revenue hit to them. The good news is that those companies have a very healthy backlog of orders. Also, what we’ve seen with all those past sanctions, the biggest force working against US government restrictions has been US companies, lobbying against them, finding loopholes. They are all complying to the letter of the law, but at any opportunity to sell they will sell.

And what are the consequences for companies like Apple, Nvidia, AMD and Qualcomm, which rely on Asian manufacturers to manufacture their chips?

Apple will still be able to get everything they need from TSMC. At this point, it also doesn’t look like they will have problems getting their phones assembled in China. In the case of Qualcomm, things are more complicated. The Chinese smartphone makers Oppo, Xiaomi and Vivo are all big customers. Presumably, Qualcomm is going to be able to sell them chips, but if you really stick by the rules, this is pretty high-performing stuff. You probably wouldn’t use Qualcomm’s mobile chips for a super computer, but you could theoretically. Same thing with Intel, AMD and Nvidia. All these companies sell a lot of CPUs and GPUs into China. So it’s just a mess.

Against this backdrop, what would you advise investors when it comes to investments in semiconductor stocks?

In the last two or three years, we’ve had this period where semiconductors were just through the roof. But those of us who have been doing this for a long time kept saying: Let’s not forget that semiconductors are cyclical. We knew that gravity was going to set in eventually. That being said, I think this is not going to be the worst cycle unless the recession gets really awful. There is still a lot of underlying demand, and in each segment of chips it’s a very different story.

What does that mean specifically?

There is overcapacity in laptops and phones, and we’re going through that pain now. We will probably have data center demand tick down at some point soon. But automotive and industrial are still constrained, so we probably won’t see excess inventory in those categories until the middle of next year. As a result, it will feel like this is a long cycle because right when one segment is recovering, another one will start to feel pain.

What signs should one look out for that the cycle is recovering?

The thing to look for is stabilizing inventory numbers. Six months ago, we’ve already started to see rising inventories in handsets. People ignored it, but we knew that inventories were becoming a problem. So inventory levels at the fabless companies are something to keep an eye on. Once those start to taper off, once they start to inch down again, that’s when I would be looking to get in. But you’ll have plenty of time. We have to have a few more companies write down some inventory. We have a couple of quarters before everything gets flushed through.

Which semiconductor stocks do you favor?

I wouldn’t buy it today, but Nvidia is always interesting. They’re firing on all cylinders, and they seem to be in a good position since there is an immense amount of demand there. I also like ON Semiconductor. The company has done a very good job of rationalizing its product portfolio, and you’ve seen an immense improvement of its margins as a result, and I think there’s more for them there. Another one is Lattice Semiconductor. The stock is always expensive, but they seem to be executing very well. There is a growing need for FPGA chips, and both of their biggest competitors are somewhat stuck: Altera is stuck inside of Intel, and it hasn’t been going very well. Xilinx is now getting integrated inside of AMD, and that’s going to distract them. So it’s a good time to be Lattice.

To conclude, let’s take a brief look at the big trends for the future. What is the medium to long term outlook for the chip sector?

We’ve been touching some pretty dark topics. But at my heart, I’m fundamentally very optimistic about the industry. Gartner, IDC and other research firms predict that in the next ten years growth in hardware, electronics and semis will be 60% of technology revenue growth. In other words, hardware is forecasted to grow more than software. Over the previous ten years, or over the past five years certainly, venture funding has been 90% software. I think the pendulum will start to move into the other direction, and there is a huge opportunity to invest in electronics and semis. With the slowing of Moore’s law, it’s going to require a new set of companies to find new solutions, and there are a lot of interesting startups emerging on the electronics and hardware side - and that’s a big opportunity.

Jonathan Goldberg

Jonathan Goldberg is an investing partner at the venture firm Snowcloud Capital and the CEO and founder of D2D Advisory, a strategic advisory firm. He also serves as CFO and Director of Nodle, a decentralized wireless network for the Internet of Things. Prior, he was Senior Director of Strategy of Qualcomm’s Data Center Group and Senior Director of Corporate Development and Investor Relations at Peregrine Semiconductor. He lived and worked in China for over ten years and has deep ties to the electronics industry in Asia. Mr. Goldberg has a CFA, an MBA from the University of Chicago Booth School of Business and a BA from University of California, Berkeley. He is the author of the Digits to Dollars newsletter and the book «A Practical Guide to IPO’s: How to Take Your Company Public».