5 Takeaways From Semicon

The notable topics were the upturn, China, EUV, fan-out and R&D.

popularity

At the recent Semicon West trade show in San Francisco, there were a multitude of presentations on a number of subjects. The event, sponsored by SEMI, had presentations on the outlook for ICs, equipment and packaging.

Clearly, though, the show is much smaller with fewer attendees, as compared to past years. Most of the big companies no longer have booths. Hardly any have press events or media briefings.

It’s still a great place to network, however. At the event, there were over 600 exhibiting companies. Official attendee numbers are not in yet, but SEMI expects +/- 5% over last year’s numbers.

Meanwhile, in no particular order, here are my five takeaways from Semicon West:

1) When is the downturn?
Both the chip and equipment businesses are booming. On the chip side, market research houses have raised their forecasts. In November, for example, the World Semiconductor Trade Statistics (WSTS) group predicted that the IC market would grow a modest 3% in 2017 over 2016. Then, in July, the WSTS raised its forecast and said the semiconductor market would reach $378 billion in 2017, up 11.5% over 2016.

The new forecast is attributed to a surge in the memory market. In fact, there are shortages of both NAND and DRAM. Analog, mixed-signal chips, RF and other products are also hot.

Just how long will the party last? No one knows for sure. But in 2018, the WSTS sees the market cooling down as the IC market is projected to grow by 2.7%.

One indicator is NAND. NAND, especially 3D NAND, is hot and driving the memory market. It is also fueling the fab equipment boom. For about a year, there has been a shortage of NAND for two reasons. First, OEMs are gobbling up parts. Second, NAND vendors are taking longer than expected to convert from planar NAND to 3D NAND, according to Jim Handy, an analyst with Objective Analysis.

Handy describes the situation as a “NAND stall.” “(A NAND stall is) the inability to move from one technology to the next, whether it’s moving from node to node or whether it’s moving from planar to 3D NAND in this case,” he said. “It’s anybody’s guess when this 3D NAND monster will be wrestled to the ground. 3D NAND has proven to be far more difficult than anybody ever thought it would be.”

In any case, the question is clear—How long will the shortages last? “We are expecting to see shortages until the middle of 2018,” he said.

Then, at that point, the “NAND stall” could subside. This, in turn, would cause an oversupply and falling ASPs. “Now, we are in the middle of a NAND stall. Usually, at the end of these stalls, we switch pretty quickly from a shortage to an oversupply. That sudden switch causes the prices to collapse at a pretty alarming rate,” he said.

2) Rumor mill: China may push out CapEx
SEMI released its mid-year forecast at Semicon West. SEMI reported that worldwide sales of new semiconductor manufacturing equipment are projected to increase by 19.8% to a total of $49.4 billion in 2017, marking the first time that the semiconductor equipment market has exceeded the market high of $47.7 billion set in 2000. In 2018, 7.7% growth is expected, resulting in another record-breaking year ─ totaling $53.2 billion for the global semiconductor equipment market.

In 2017, South Korea will be the largest equipment market for the first time. Then, in 2018, SEMI forecasts that equipment sales in China will climb 61.4% to a total of $11 billion, following 5.9% growth in 2017. China is forecasted to become the second largest market in 2018.

Today, China has nearly two dozen fabs in construction or on the drawing board. China is targeting to boost its domestic production in both memory and logic. China has several 3D NAND fabs in the works. But the nation is behind in technology by a wide margin. The country has produced a 32-layer 3D NAND chip with questionable yields.

In comparison, the multinationals are moving from 48- to 64-layer devices with 96-layer products in R&D. China could introduce 32-layer chips, but the ASPs would be low. They might have to sell them at a loss. China could wait and develop more cost-competitive 48- or 64-layer chips, but this would take time.

All told, don’t expect China to ramp up 3D NAND in a big way for some time. “They actually have stated that their goal is to be in volume production by 2020,” Handy said.

At the same time, China is also building a number of 200mm fabs. Yet, there is a severe shortage of 200mm equipment, meaning China is having difficulties equipping these fabs.

So the capital spending boom in China may emerge in 2018. But some experts say it could happen later than sooner.

3) EUV resist power play
At Semicon, ASML made headlines when it announced that it finally reached its target goal for the power source in extreme ultraviolet (EUV) lithography. It achieved 250 Watts of power with sustained operations.

That wasn’t the big news, however. For some time, the other stumbling block for EUV is the resists. They can’t meet spec.

Then, during Semicon, EUV resist developer Inpria said it secured $23.5 million in Series B funding. The funding was led by Samsung Ventures, Air Liquide, Applied Materials’ venture arm and Intel Capital.

Another investor was Inpria’s competitor—JSR. JSR is also developing EUV resists. Perhaps JSR is hedging its bets. Or the VCs are looking for an exit strategy. To be sure, something is happening behind the scenes.

But clearly, it takes a ton of resources to play in the EUV game. Inpria has done good work. It just needs more help.

4) Fan-out disillusionment?
Last year, fan-out wafer-level packaging reached a milestone when Apple adopted the technology for its iPhone 7. For the iPhone 7, TSMC fabricates Apple’s A10 application processor. Based on a 16nm finFET process, Apple’s A10 is housed in TSMC’s InFO (Integrated Fan-Out).

The other OSATs are also developing high-density fan-out in an effort to get a piece of the action. Yet, they aren’t having much luck. Other smartphone OEMs aren’t exactly biting. Fan-out is still too expensive compared to the current solutions.

At Semicon, one OSAT described the scenario much like Gartner’s “hype cycle” chart. When a promising technology enters the market, the industry falls into a state of exuberance. Then, if the technology doesn’t take off as planned, it falls into a disillusionment stage. High-density fan-out, according to one OSAT, is in the ”disillusionment stage.”

It’s promising, but it isn’t taking off with the exception of TSMC. And TSMC is expected to get Apple’s iPhone 8 business, including the fan-out deal. So is it back to the drawing board for the OSATs?

5) Make U.S. R&D great again
At Semicon, two R&D organizations–Imec and Leti–always have separate and impressive events. Both organizations are filling a big need in the industry—R&D. Imec of Belgium and Leti of France are not only doing R&D on the leading-edge, but also on IoT and a host of other technologies.

So what’s wrong with that picture? Not long ago, U.S.-based Sematech filled the void and was the big player in the R&D game. But in 2015, Sematech was folded into the State University of New York Polytechnic Institute (SUNY Poly), a high-tech university system in New York.

SUNY Poly does fine work. Yet, there is a big void in terms of having a R&D organization in the United States. Clearly, there is R&D taking place in the U.S. Semiconductor Research Corp. (SRC) is one example. Then, Intel, GlobalFoundries, Micron and others have R&D units.

Is that enough? And does there need to be a separate R&D lab, much like Bell Labs in the grand old days?

Late last year, I sent an e-mail to president-elect Donald Trump and his organization. I explained what the U.S. needs is a new and viable R&D organization to fill a big void. It doesn’t have to focus exclusively on the leading-edge. There are other viable technologies to work on, such as 5G, AI, IoT and others. And there is no shortage of brain power here.

I haven’t heard back from the Trump organization—yet. Perhaps my thinking is wrong. Maybe R&D is strong in the U.S. and the government should stay out of the way. But I assume I will need to wait for a reply from our government as I continue to attend more events and learn about technology from the other R&D organizations.