Revenue in the first half of 2017 was a record $675 million, up 16 percent on the first half of 2016 and being driven by strong demand with additional manufacturing capacity coming on-stream in the United States and Japan. Gross and operating profit for the first half of 2017 were a record of $176 million and $110 million, respectively, increased as compared to $134 million and $71 million, respectively, in the first half of 2016.

So eeNews Europe asked Ellwanger for an update on Tower’s progress; a follow up to an interview we conducted a couple of years ago (see CEO interview: What’s next after Tower’s turn-around?)

All good so far considering Tower had annual sales of about $100 million with a loss of twice that size in 2005, the year that Ellwanger was appointed, and reached $1.25 billion in 2016. “Yes we’ve had 10 or 11 years of growth. The growth percentage has been high but we were coming from a low base,” said Ellwanger. In fact, Tower’s compound annual growth rate is above 25 percent and the period covers a number weak years for the semiconductor sector as a whole.

But has that been allowed because the competition from mainstream foundries against the specialists has diminished in recent years?

It was the case that larger pure-play foundries had a plan to redeploy their older wafer fabs to making More-than-Moore processes as they were bringing up new wafer fabs to do leading-edge digital IC manufacturing. But perhaps the demands of developing those very expensive, deep nanometer processes kept those big foundries occupied and gave smaller foundries, such as Tower, a chance to grow.

“There are several foundries that do deep digital following Moore’s Law and TSMC is at the leading edge. They also have a very good analog offering,” commented Ellwanger. But Ellwanger also points out that the number of More-than-Moore manufacturing process offerings is extremely broad and that Tower rarely sees TSMC as a competitor in the market.

Next: Three sectors

“We deal in three major technology sectors,” said Ellwanger. “Power discretes and PMICs [power management ICs] are about 30 percent of revenue. RF components for front-ends, power amplifiers, low-noise amplifiers and RF infrastructure are about 30 percent, and then there’s sensors. That is CMOS image sensors but specialised such as industrial vision, studio cameras, medical and x-ray using stitched die. We also have some magnetics sensors and MEMS activities and that’s growing,” he added.

“Overall when you are enjoying greater than 20 percent organic growth you are doing ok,” he said. “We rarely compete with TSMC. Within the area we choose to compete it is a level playing field whereas below 10nm requires a tremendous amount of capital. For us 0.18-micron BCD is state of the art. This means we don’t need so much capital. Our differentiation can be maintained through the calibre of people we employ.”

Ellwanger added: “That organic growth is at above the rate of the foundry market is the best health indicator and that you are in the right markets.”

Although many people speak of IoT it is a highly fragmented area of opportunity. One of the more concrete examples is the electrification and electronification of the automobile to create a communicating vehicle. Is that something that is driving Tower’s growth?

“Well it is very difficult for us to know what percentage of our power discretes goes into automotive. For PMICs I have a good feel, especially for lithium-ion batteries and we are gaining customers. RF for automotive is again a good growth segment for us. Denso used us for a 24GHz automotive radar. Lidar with cameras and MEMS is also promising. We don’t have volume production at present but solid-state lidar is definitely in demand. It is a focus and a direction for the company,” Ellwanger said.

“Automotive sensors are a tremendous opportunity with the need for sensor fusion across areas radar based on SiGe, lidar and CMOS image sensors.” In 2016 about 20 percent of Tower’s revenue was due to the automotive sector.

But the semiconductor and electronic industries are fast-moving environments. What technologies does Tower need to be bringing in?

Next: Technology is necessary but not sufficient

“It is very important to have technology leadership. It is necessary but not sufficient to achieve success. That technology leadership must be within growth market and aligned with customers in multiple generations of their roadmaps,” said Ellwanger.

There are some instances where customers have brought processes to Tower and provided long-term contracts and asked Tower to take over manufacturing but increasingly Tower has to engage in leading-edge research and then go out and seek customers, Ellwanger said. But he added that it is definitely NOT a case of build it and they will come. “You find a partner or multiple partners to develop engineering based on that research.”

Seeking more detailed response eeNews Europe asked Ellwanger if Tower moving into compound semiconductors, such as gallium nitride, to complement its position in silicon-germanium?

Ellwanger is cagey. His first point is that the silicon-germanium process is good for just about all RF applications and is cost efficient. “SiGe at 65nm is probably more than sufficient for most frequencies and has the advantage for allowing integrated chips.”

“There are different materials we are looking at and maybe they’ll work out and maybe they won’t. For compound semiconductors outside our served market then acquisition or partnership may be the way to go. There are things we are evaluating,” Ellwanger said.

But Ellwanger said that in terms of year-on-year market growth sensors is probably the biggest opportunity with a 25 to 30 percent CAGR.

The acquisition model can provide both technology and a customer, Ellwanger said. Otherwise it can be three years of technology development followed by three years of looking for a customer.

Ellwanger gives the example of discrete power where Tower took over base power production flows from Vishay, Fairchild and International Rectifier (now Infineon) and then provided incremental improvements specific to those customers. The alternative example is the PMIC design flow where Tower now works with three of the top 6 or 7 PMIC providers. Tower does this with a PMIC platform that it developed and that multiple fabless customers wanted to access. And in the RF area Tower works with manufacturing companies but who are effectively fabless in the area where Tower serves them.

So what about technology introductions in the power domain where silicon-carbide and gallium nitride are much talked about? “GaN is something we believe is needed. Right now GaN for power is very small part of the market but at some point it will take off. We have been asked about it,” he said without saying more.

Next: More MEMS?

MEMS as an area where Tower is active. It is also an area where there are multiple manufacturing approaches to a fragmented set of applications. It has been argued for many years that MEMS market development has been hampered by its reputation for requiring a dedicated manufacturing process for almost every product. Is this a suitable case for Tower to get more involved?

“We have Cavendish Kinetics as a customer with their tunable antenna RF switch. We do make inertial sensors and speakers. But we are very selective. It has to be good performance, with a very good upside in terms of market and not too interfering with the running of a high-volume wafer fab,” said Ellwanger. He points out that the product-specific nature of most MEMS processes and the small size of MEMS makes wafer counts low and repeatedly resetting equipment to run a few wafers can be disruptive to running other processes in a large fab. “You really need a dedicated MEMS line. You would like to have that; a small line; a pilot line. Then you could transfer [a process] later to the high-volume line.”

Tower could spend its own money speculatively building such a pilot line. But Tower usually prefers to spend money alongside that of a customer who will take output.

“We have an annex at Newport Beach wafer fab built with FLIR Systems. We are willing to do a lot with key customers,” said Ellwanger. Tower makes MEMS at both Migdal Haemek and Newport Beach. However, he said there is nothing to report at present.

Along with additional processes comes the need for additional production. At present Tower seems will placed with 150mm and 200mm wafer fabs in Israel and 200mm wafer fabs in Newport Beach, California and San Antonio, Texas. By way of the TowerJazz Panasonic Semiconductor Company (TPSCo) in which Tower is the 51 percent shareholder it has control of three more fabs; 200mm fabs in Arai and Tonami Japan and a 300mm wafer fab in Ouzo, Japan (see Tower buys three wafer fabs for $8 million).

Next: China and India

And then there is the recently confirmed news that Tower is in partnership with Tacoma, Chinese investors and Chinese regional authorities to build a 200mm wafer fab in Nanjing, China. Tower has received an initial payment of $18 million and according to reports there is a longer-term plan for a 300mm wafer fab in the same industrial park (see Tower confirms Chinese fab project).

Meanwhile a project in India that Tower was involved with over several years appears to have fallen apart (see Lead partner pulls out of India fab plan). So does this mean that India is now out of the running as a semiconductor manufacturing location for Tower and other chips companies?

Ellwanger said: “We were strongly involved in India for the building of a 300mm wafer fab. The government slowed it to a three-phase process and with that slow-down the project came a point where the original lead partner could no longer meet the financial commitment.”

He continued: “The main learning isn’t about India. It was about not having a binding MOU [memorandum of understanding]. In the case of China there was a binding MOU signed in July 2016 and we have now received initial money this year.”

“There can be slowness in any country with legislators and bureaucrats but is India off the table? Absolutely not. At some point, there will be a semiconductor industry in India. Prime Minister Modi is an amazing person. We always have irons in the fire and India is a big value proposition for us,” said Ellwanger.

Related links and articles:

News articles:

CEO interview: What’s next after Tower’s turn-around?

Tower confirms Chinese fab project

Lead partner pulls out of India fab plan

VC wants to reboot Indian wafer fab plans

Tower buys three wafer fabs for $8 million

Tower releases 65nm CMOS plus power process

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles