Anadigics: 25 Years On A Roller Coaster
Anadigics has packed an awful lot into its first 25 years: it has experienced the highs of pioneering 4-inch GaAs production and leading high-volume manufacturing of power amplifiers for handsets; but it has also suffered from the lows of dealing with unsustainable losses and losing market share to superior chip technology. Richard Stevenson tells the company’s story.
In the mid 1980s the world was gripped by the Cold War. USSR and its communist allies were pointing scores of nuclear missiles at targets in the West, which had a similar arsenal in place for its adversaries. Both of these Superpowers hoped that the devastating consequences of a nuclear war would ensure peace. But this could not be guaranteed, so Ronald Reagan’s Administration tried to develop a system that intercepted nuclear missiles in mid-flight. This project, which had an unofficial nickname Star Wars, focused on developing laser systems that could be attached to satellites and shoot down nuclear missiles.
If Star Wars was going to be a success, it would require the construction of novel systems incorporating a rafter of cutting edge technologies. For example, many GaAs ICs would be needed to provide amplification and manipulation of the signals used in different types of radar systems operating in various bands of the microwave spectrum. A successful program could consume many of these chips, and several start-ups specializing in GaAs IC technologies were launched to try and cash in on this emerging market.
In amongst this group was Anadigics, an East-coast startup that has kicked on to become arguably the most revolutionary GaAs chipmaker of the last 25 years. Its trailblazing efforts have included leading the move to 4-inch GaAs manufacturing; becoming the first high-volume manufacturer of power amplifiers for handsets; and being one of the first manufacturers to move to more efficient, more reliable HBTs, by switching the transistor design from an AlGaAs to a GaInP emitter. It’s a terrific list of achievements that could never have been foreseen by its three founders, Ron Rosenzweig, George Gilbert and Charles Huang, when they formed the company in Warren, NJ, in 1985.
Founding a start-up was not a new challenge for Rosenzweig and Gilbert – back in 1968 they co-founded Microwave Semiconductor Corporation (MSC). “We were designing and manufacturing state-of-the-art RF and microwave transistors that were mainly used in amplification and oscillation," recalls Rosenzweig, who was the company’s CEO.
MSC, which mainly targeted defense and telecommunication markets, had some III-V expertise and in 1976 it started a line of GaAs power transistors and power amplifiers. These were discrete devices that were subsequently attached to ceramic substrates.
In the late 1970s MSC was bought by Siemens. Rosenzweig and Gilbert were given five year contracts as part of the deal. Towards the end of these contracts their entrepreneurial spirit got the better of them, and these two guys in their late 40s wondered how they would see out their working life. “The encore turned out to be Anadigics," says Rosenweig.
The focus for the start-up was the manufacture of GaAs integrated circuits for defense companies contributing to the Star Wars program. “The business plan revolved around making modest quantities of high-performance niche products with selling prices in the $25 to $100 range," explains Rosenzweig, Anadigics’ first CEO. If the company could ship tens of thousands of these components each year, then it had a good chance of turning in a profit.
Winning funding to kick-start Anadigics was relatively easy. In the late 1960s, when Rosenzweig and Gilbert co-founded MSC, venture capitalists were incredibly rare, but this time round there were plenty to appeal to. And what’s more, this twosome had a fine track record of running a company. Within three years of its launch they had taken MSC into the public domain, and its strong performance on the NASDAQ throughout the 1970s meant that its sale netted a good return for all its investors.
Rosenzweig knew that GaAs expertise was crucial to the company’s success. He could have tried to poach former colleagues from MSC with this skill set, but he didn’t want to risk upsetting his former employer. “Instead, we made a strategic decision to go out of our normal circle and get the best and brightest that was possible."
This quest turned up Charles Huang, head of the RF device division of silicon-valley start-up Avantek, a microwave component manufacturer. Huang could also bring a different, West-coast culture to the team, and Rosenzweig was delighted when this GaAs expert agreed to co-found Anadigics.
One of Huang’s great strengths was his deep knowledge of metal semiconductor field effect transistors (MESFETs), the universal technology of the time. “In 1985 the HBT was still a laboratory curiosity, being funded by DARPA for various development applications," explains Rosenzweig. MESFET manufacture had the great advantage of not requiring any epitaxial steps, and was based around the implantation of ions into a semiinsulating substrate.
Piles of cash
The three founders found that they had plenty of cash to play with. They raised $8 million of first round funding in 1985, and quickly followed that up with a further $15-20 million. Today many start-ups would use this funding for development of chip designs, and outsource their manufacturing, at least initially. However, back then it was a different world, and Anadigics fitted into the philosophy of the time. “We created a vertically integrated, real-menhave- fabs company," recollects Rosenzweig, who had a further $8 million at his disposal to buy the capital equipment for the 3-inch GaAs fab.
But despite this tremendous investment, they failed to generate a profit. This was partly due to unforeseen events in Europe that were completely outside of their control. Led by the Solidarity movement in Poland, communist countries were turning to democracy, and the US government felt that it made sense to drop investment in the Star Wars program given the new world order. Sales to defense companies also received a massive blow due to cuts in the defense budget. The proportion of gross domestic product for defense was trimmed by 25 percent between 1986 and 1991, and further cuts occurred throughout the 1990s.
Anadigics clearly needed to target new markets with different products, which would have to be far cheaper. Fortunately it had already carried out some of the groundwork. “Although we had built up our business plan around Star Wars, we did not want to only be a defense company," explains Rosenzweig. From the outset the Warren outfit had also been investigating opportunities in the satellite communication market; in the cable TV market, where GaAs could be used in tuners; and in fiber optic telecommunication, where GaAs chips could be used to make laser drivers and receivers. The only application that Anadigics did not consider venturing into was GaAs ICs for logic circuits in super computers, a market that was already cornered by the likes of Vitesse and Gigabit Logic.
Although there were plenty of promising opportunities for GaAs ICs, Anadigics struggled to find a high-volume contract. At this point the company was only bringing in sales of a few million each year, nowhere near enough to generate the cash to pay for the wages of 50-60 staff, a high R&D bill, and the scores of 3-inch, $1000 GaAs substrates that were being consumed. This state of affairs resulted in a burn rate of $8 million a year. “We were grossly unsuccessful, but no different from anyone else at that moment in the field," says Rosenzweig.
By 1988 Anadigics was running out of money fast. It appealed to investors, and raised another $10 million that kept bankruptcy at bay for a couple more years. But when that additional investment ran out it was clear that the company needed a new strategy. Rosenzweig’s plan was to find a strategic partner, and he found two allies in Europe: Thompson CSF and Philips. Although no business came directly from eiter of these relationships, the faith of these European partners was incredibly helpful to Anadigics, because it encouraged existing investors to part with more of their cash.
One problem still remained: uncovering the killer application. Again, Europe played a hand in Anadigics’ fortunes, with British Sky Broadcasting and Sky independently launching satellite TV. “We came across a company called Continental Microwave, which had put in a contract for one million low-noise converters," explains Rosenzweig. Continental’s CEO believed that this product needed to employ GaAs in the manufacture of low-noise block converters that take a set of frequencies in the Ku-band (11-12 GHz), amplify them, down-convert them to an intermediate frequency of 1 GHz, and then send the resultant signal to the set-top box.
Anadigics priced the parts at about $5 and won the order for them. To meet the subsequent demand, everyone within the company dropped what they were doing and started working on this project. “Suddenly there was the first high-volume GaAs IC market of any substance," recollects Rosenzweig.
Within three years Anadigics was churning out 5 million ICs. This required further capital equipment, but to put the volumes in perspective, this throughput is a tiny fraction compared to the 200-300 million ICs produced by the company today. Back then the process required a lot of manual labor, and yields varied substantially. “You had good yields and bad yields - it was not a very controllable process," reminisces Rosenzweig.
The $15 million annual sales generated by shipping 5 million GaAs ICs put the company on a far better financial footing. It helped spur annual revenues to $20-25 million, and made the company a viable entity that was close to turning in a profit. Winning additional financing was also far easier, and the company had a couple of additional rounds of investment.
The mobile revolution
Not long after this a product came along that changed the GaAs IC industry forever: the handset. Initially this wasn’t the portable, lightweight device of today that fits easily into a pocket or handbag, but a car phone running off the automobile’s battery. In this context efficiency hardly mattered, and silicon LDMOS and bipolar technologies were capable of meeting the output power requirements. However, handset manufacturers were well aware of the potential of GaAs power amplifiers, which promised to offer far higher efficiencies that would draw less power from the battery, leading to increased talk time.
Anadigics decided to try and team up with the cell phone manufacturers. The market leader of the time was Motorola, but they already had their own internal GaAs capability. So Anadigics scoured the globe trying to find a firm that would invest in the Warren outfit and pay its development costs for GaAs power amplifiers. “We wound up with Ericsson being a strategic development partner. That was very important. They didn’t fund us, but they gave us the specs," recollects Rosenzweig.
When Anadigics started shipping the PAs in 1995 it entered a new era. Annual sales grew to $25-30 million, and the company was now profitable. The latter achievement had major implications – it allowed Anadigics to go public and raise a further $24 million.
“Part of the reason for raising that money was to use the proceeds to expand the fab, so that we could handle much more production," explains Rosenzweig. “Around 1997 we realized that production was not economically viable on 3-inch wafers, and we expanded to 4-inch. We were the first company to do that."
By then Anadigics’ sales were dominated by GaAs Pas and revenue had rocketed to $75 million. “We were supplying Nokia with PAs, we had virtually all of Ericsson’s business, and we helped Qualcomm launch the PCS industry in the US," recounts Rosenzweig. Introduction of PCS technology was a major breakthrough, because it started the era of digital cellular communication.
Anadigics was now in the incredibly enviable position of being the global leader in a rapidly expanding market, and to increase its chip production and cut manufacturing costs it built a 6-inch GaAs fab that came on-line in July 1999. This helped the company to continue to ramp up its GaAs IC shipments and realize record sales of $51 million in the third fiscal quarter of 2000.
But the glory days didn’t last long. During the late 1990s handset manufacturers started migrating from batteries operating at 6 V to 4.5 V and finally 3V versions. The HBT was better suited to these lower voltages, and it also had another major plus point, higher efficiency. “Once HBTs were proven to meet the price points that the handsets required – a $1-2 power amplifier – the game was over for the MESFET in the handset market," admits Rosenzweig.
Anadigics’ demise was quick. In 2000 its sales were unscathed by the HBT and were worth $172.3 million, but by 2001 - the year the dot.com bubble burst – MESFET shipments for PAs had collapsed, dragging down total annual revenue to just $84.7 million. Everyone could see this coming, and shortly after it happened Bami Bastani - who took over as CEO from Rosenzweig in October 1998 - faced a tough decision: should Anadigics exit the handset business, or become a HBT company?
He chose the latter, but he was determined that Anadigics would remain a trendsetter, rather than becoming a crowd follower. To this end the company was not going to make a HBT with an AlGaAs emitter, which was rapidly becoming the incumbent technology, but a product with an InGaP emitter that would combine superior reliability with better high-temperature operation and improved linearity over battery lifetime.
By the fall of 1999 the company had developed a GaInP HBT process on its 6-inch line, and few months later it was shipping InGaP HBT samples to potential customers. But it took an awfully long time to convince them to buy this type of product in volume, and Anadigics went through some very tough times during the beginning of the last decade, shedding employees and recording a loss for every fiscal year between 2001 and 2006. But by 2007 Anadigics’ high quality product was making significant inroads into the 3G handset market, and the company was netting quarterly sales of over $80 million.
It then committed the cardinal sin: Failing to meet customer orders. Bastani resigned shortly after the news broke, and Anadigics started searching for a new CEO.
Company chairman Gilles Delfassy took hold of the reigns as an interim measure. He also contacted a good friend of his, Mario Rivas, to see if he might be interested in the CEO role. These two industry veterans first knew each other when Delfassy was in charge of the wireless business at Texas Instruments and Rivas held an equivalent role at Motorola. Their friendship was strengthened when Rivas moved to Philips Semiconductor, because the two of them then worked together to provide the baseband and PA for Sony Ericsson handsets.
Rivas took a look at the role, and decided he was interested: “I thought that Anadigics had great technology, great people, and a great possibility. And it’s down my alley as far as expertise was concerned."
When Rivas started at Anadigics on 1 February 2009, morale was rock bottom. There had been lay-offs, quarterly sales had slumped from $80 million to $30 million and the economic outlook was terrible, with the world entering its worst recession since the 1930s.
Rivas cut through the doom and gloom, telling the workforce that they were going to have fun: “We spend too much time at work not to have fun." He also fostered a stronger sense of teamwork, making it clear that every employee has an important contribution to make. The CEO identified three goals for turning the business around: greater intimacy with customers; improved operation; and preservation of cash. “The fact that we were at 30 percent factory utilization allowed us to do a lot of improvement. It’s hard to improve when you are full."
Customers are now starting to come back to Anadigics, and according to Rivas, they never had any doubts about the quality of the company’s technology. Their major concern was trusting Anadigics once more with the role of being a key supplier. Thanks to their faith in the Warren fab, Anadigics quarterly sales are edging towards the $50 million mark, and an associated increase in fab utilization has allowed the company to drop its temporary introduction of one week’s unpaid leave every quarter. What’s more, the company has been recruiting, a move that always boosts morale.
One important change that Rivas has brought to the company is a switch to a hybrid model that combines inhouse production with outsourced manufacture. “When we passed 30 percent fab utilization I announced a strategic agreement with WIN Semiconductors in Taiwan. People were scratching their heads and saying: Are you nuts?" But Rivas knows that it takes a long time for a new factory to optimize its processes for new products that satisfy customer expectations. He was already thinking about the demand for Anadigics products in 2011, and by then he expects that the company will manufacture over 1000 wafers per week, exceeding its in-house capacity.
If the company were to build another fab to meet that demand, it would need to raise $100 million. “In the mean time, you have WIN Semiconductors, which has a very good technology and a very good funding base. You could look at it as free capital for us."
Rivas expects the GaAs IC business to grow over the next few years. He is targeting production of 2500 wafers per week in 2014, a level of production that will allow Anadigics to up its share of the GaAs handset market from today’s single digit figure to 14 percent. If the company executes on this front, it will definitely surpass its current record for annual sales of $250 million.
“I will be pleased if we could break $500 million," says Rivas. “Then we’ll have finally grown up, and no longer be an adolescent."
And let’s hope that is the legacy that Rivas leaves, because when you are approaching 30 you should have thrown off the exuberance of youth, and settled down into a pattern of earning a little more every year.
Anadigics current CEO Mario Rivas (right) invited Ron Rosenzweig (left) onto the stage for ringing the closing bell for the NASDAQ on 22 April 2010