SEMICON China: Me and 70,000 of My Closest Friends

If you want to know why I made the trek from California to Shanghai to attend SEMICON China, it is well summed up in the press release that SEMI put out on opening day:

SEMICON China 2018 opens today at SNIEC in Shanghai with a record 70,000 visitors expected with 3,600 booths from more than 1,000 exhibitors bringing the latest innovations in the electronics manufacturing supply chain. SEMICON China kicks off as China’s fab construction spending is forecast to grow from a record high $6.2 billion in 2017 to $6.8 billion in 2018, accounting for over 50 percent of worldwide construction spending. China is expected to significantly expand investments in wafer fabrication equipment this year to become the top-spending region in the world.

Opening the Show

A procession of the people who have made the Chinese semiconductor industry and SEMICON China what it is today gave a perspective. They ranged from Lung Chu, the head of SEMI China, to Ajit Manoja, the CEO of SEMI globally, to Wang Ning, the head of CECC, and the head of the Shanghai chamber of commerce.

SEMICON China has become the biggest and most influential industry exhibition in the world. This year is the sixth consecutive year that SEMICON China has been the biggest SEMICON (it is three times the size of the San Francisco-based SEMICON West). It is also the 30th year of SEMICON China. To give a comparison, when SEMICON China started, there was 60,000 square feet of exhibition space (6000m²). This year, there is 730,000 square feet. As it says above, there are over 3,500 booths from over 1000 enterprises (I don't quite know how that counting works, since that means lots of companies have more than one booth). Also running is FPD China (FPD stands for flat panel display).

The challenge, as Wang Ning pointed out, is that China consumes one-third of the world's semiconductor output but only 7% of that is domestically produced. The cost of semiconductor imports has now reached twice the cost of crude oil imports. It is the biggest user of foreign exchange reserves. Bottom line is that China needs to improve its global competitiveness.

[He didn't point it out, but I will. Much of those imported semiconductors are re-exported inside finished goods, which has to be the biggest contributor to foreign exchange reserves. You can't say that about much of the oil, although it is a feedstock for plastic.]

When Ajit Manoja, the head of SEMI, presented, he pointed out that in 2016 China started construction of 26 volume production fabs and by 2019 they will be ready to produce. By 2020, China will be #1 in global investment in the semiconductor equipment market (I'm actually surprised they are not already). However, China cannot be successful on its own since there are interdependencies everywhere. Ajit pointed out that we need to grow this industry beyond its current $400M. Of course in 2017, he said:

Some of the growth is due to memory prices. But we have always been relying on something to give us growth, and now there are ten or fifteen things. I'm optimistic we can have a CAGR of 6% taking semiconductor to $1T by 2030.

The mantra of SEMICON China is "connect, collaborate, innovate". But, as Ajit said, we need to add "prosper and grow."

The head of the Chinese Semiconductor Industry Association came on. I'm not sure of his name since he wasn't on the agenda and I didn't seem to write it down. He was there to take away the punch bowl:

It takes two to three years to build a new fab, and another two to three years to ramp it to full volume. So to build a profitable plant takes a long time. I expressed my viewpoint last year and I'll say it again. For China, the semiconductor industry is rising, but has not quite risen yet. It will take time.

Keynotes

After all the introductions came the true industry keynotes. I will cover some of these next week. I'll not bother with the ones that were thinly veiled pitches for company. The keynoters are all in the above picture.

Huawei

The wrapup for the opening day was Chu Qing, the Chief Strategy Officer for Huawei. I'll try and give a flavor of his presentation, and pull out one or two points that I thought were interesting. He spoke in Chinese, very fast, and the simultaneous translators couldn't really keep up. Plus he leaped around from electronics, to biology, to philosophy, which made it harder still. Yes, he mentioned Plato and the shadows on the wall of the cave. He mentioned emperor penguins and how you need to get into the group to survive. He was also the only person whose slides were only in Chinese. Most other people had both Chinese and English on their slides, a few had only English. Paul Dempsey was sitting next to me, and he recorded the whole thing in Chinese. His wife is Chinese and he's going to get her to tell him what it all truly meant.

The above slide was his title slide, which gives you an idea of how surreal his presentation was. As you can see, it was titled 终结？还是元年 which, between Google and my own Chinese, means "The End? It's still the beginning". I'm not sure if there is any significance attached to the title being written in two font sizes (the small two characters mean "or").

He said he had three stories. The first story is about artificial intelligence (AI). He sits in a lot of meetings at customers on AI and they are all about bus bandwidth, and getting the IC power down, and stuff like that. But that's engineering, not philosophy. He prefers to look at the mythical future.

Then we had a diversion into René Descartes (that's him in the above picture), bionics, Plato's cave and the shadows, and how the human brain is constructed for intelligence and divided into different functional zones. He's not a big believer in big data, since that's basically just recording the past and trying to use it to predict the future. Like Moore's Law:

Many people regard Moore's Law as an objective rule, but it is just a commercial strategy.

Next diversion was into Hilbert Spaces, which extends vector algebra into infinite dimensions. He wrapped up this section talking about what the translator (who was Australian, to make the whole thing more surreal) called "mechanism theory", although I suspect it is something else. Future innovations should be truly new thoughts and challenge the basic philosophy. AI is about more than just engineering and memory bandwidth.

The second story was about internet of things (IoT). There is a huge gap between IoT expectations and reality, especially in the networking. There is a network for the car, the home, all with different business models and technologies. It's a pseudo-concept. IoT should be the largest network ever. That's where the value is. The first generation was Bell and Ericsson who invented the phone. Now the high valuations are no longer in banking, they are in our industry. First AT&T, then IBM/Microsoft, more recently Google and Apple. So the big question is who will be the big guy in the IoT era. Maybe someone we don't suspect.

The third story was the IC industry. The life or death of the industry. 

The above picture needs a little decoder ring. The fish at the front is Nokia. The next one (Q) is Qualcomm. Then Broadcom (although that bit isn't happenening, it seems). Who comes next?

We are in an era of M&A that will shape the whole industry. This is where the penguins come in. An emperor penguin can survive easily in the center of the flock. The ones at the outside not so much. Isolated they will die. So the IC industry has to huddle together to survive.

Will it consolidate completely into a single...no, not a penguin, a tree. The Chinese on the above picture means (at least according to me) "The last IC company?". The industry is going to sort of stop at 3/5nm and then it all becomes software. Anyone can build anything but almost no companies can afford the cost of a full die development, which might measure in billions. But schoolkids will be able to program 3nm chips.

There has been lots of consolidation. The slide above shows some of the names that have disappeared (although if I was Hock Tan I'd take exception to Avago being on the list, since the only reason it disappeared is that it swallowed Broadcom and took on its name).

The big question for each semiconductor company is whether we'll be inside or outside the wall, in the middle of the penguins or isolated. Summer is approaching. In a clear dig at the recent blocking of the Qualcomm acquisition by CFIUS and the Trump administration, he ended with:

If you try and eliminate M&A, your local companies may just get eliminated faster.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.