LONDON In a semiconductor industry that is generally reckoned to be maturing, John Daane, chairman and CEO of Altera Corp. (San Jose, Calif.), is one of few semiconductor executives that thinks his company will be able to maintain double digit growth for the next few years. He predicts a compound annual growth rate (CAGR) of 16 percent, roughly the same as the 15 percent that Altera has experienced from 2002 to 2006 and beating the predictions of 6 to 8 percent across the industry.
“Some sectors will grow faster and some slower, but any industry that grows at 6 to 8 percent a year must be considered mature. This era when we have a combination of rising R&D costs and slowing end market is going to force companies to re-evaluate their business and what products they want to be in,” said Daane (left).
“It is also going to force a tremendous consolidation in our industry going forward. And not just in the semiconductor industry. Most of the high tech industry is at a stage where it will continue to grow but it won’t grow at the rapid rate it has in the past. One way to afford the expensive R&D is to merge or purchase other companies to get a higher market share.”
According to Daane some sectors, such as EDA, have already gone through this consolidation but the semiconductor sector still has to go through this.
Daane sees two reasons why a semiconductor consolidation has been delayed. The first is that semiconductor managers have been “in denial” and continue to spend for strong growth rather than to manage a more stable industry. A second reason is that capital has been available, even for companies with broken business models, but that will change Daane said.
Daane said the programmable logic industry has already gone through its consolidation and so the M&A route to growth is blocked, but he believes that Altera will be able to expand its markets by increasingly targeting the ASIC sector.
“When I joined the company six years ago I said that the opportunity was not to steal market share from the other PLD suppliers but to compete for the ASIC market.”
Altera’s approach is based on the Hardcopy structured ASIC which enables designers to prototype with the company’s FPGAs and migrate to a cost-reduced custom device. According to Daane 13 of Altera's top 20 customers are using its Hardcopy route and a number of these have done multiple designs.
At present Hardcopy only provides around 4 percent of Altera’s revenue but Daane expects this to rise over the next few years to between 10 to 15 percent. “This year we expect a very significant increase in our Hardcopy business as we have prototyped a number of designs for the communications and consumer areas.”
“Our aim is to be number one in custom logic,” said Daane. “We have already developed low cost silicon platforms that enable us to move in to other areas with just investment in IP. There are a number of other semiconductor companies, including Infineon and TelASIC, who are becoming meaningful customers of Altera using either programmable devices or Hardcopy to ship their own IP as a standard product solution.”
To become a major player in the ASIC arena, Daane said the company needs to target $7 billion in revenues. “Obviously we are not going to get there in the next three years” and Daane rules out acquisitions to grow market share.
“In the PLD space there is no one to acquire, in the ASIC space we could acquire another vendor but this probably wouldn’t brings us anything we don’t have access to today. We already have the tools and a lot of the IP and the IP we don’t have, we can license. We can grow organically if we focus.”
But if there is room for Altera to grow there is room for competition and for new players to enter the market. Daane reckons that programmable logic is remarkable for its lack of successful startups.
“We estimate that there has been over $1 billion in venture capital money spent in the last ten years on technologies including FPGAs, sea-of-processors, structured ASICs, combinations of structured ASICs and FPGAs, in fact any quasi-programmable vehicle. Over $750 million of this investment was in companies that have gone out of business and a number of big companies, such as STMicroelectrtonics, AMD, Philips and Intel, have also got out of the programmable market. The tools are key and over half of our engineers are software developers,” said Daane.
Things are made more difficult for interlopers by the fact that programmable logic is a moving target that adapts to advances in process technology.
“We have approached each process node with the idea that we cannot just implement what we have done before. We have to re-engineer and re-implement from scratch everything that we do to stay fresh and meet the requirements of the end customers.”
Daane admits that Altera tends not to be the first to ship products based on the latest manufacturing process. “We were not first to ship at 0.13 micron, 90-nm or 65-nm but we have over 50 percent market share in FPGAs at these nodes, compared to our general FPGA market share of 36 percent. We have not suffered from not being first to market at the new nodes. Customers are buying a solution not just a silicon platform.”
Altera started work on adopting 45-nm technology with TSMC early in 2005 and now has 11 joint development teams working on different products. Three test chips have already been produced and five more are in the pipeline. The initial three addressed FPGA-specific design-for-manufacturing processing structures including: optical proximity correction, low-voltage operation of SRAM, a device modeling vehicle, Altera-specific electrostatic discharge protection, power management circuits and the development of a mixed-signal/RF test vehicle.
The company expects to ship 45-nm devices in volume in 2008 but Daane would not reveal if the first would be Stratix or Cylcone class of devices. And the company is already working on 32-nm process development. “Each process technology creates its own unique challenges and it is not just the silicon but the packaging technology and everything you need to create a solution.”
If Altera is not in acquisition mode could it be a target for the bigger semiconductor companies? “It is always possible, if you look for the high profit, high growth areas of the semiconductor industry we are in one of the few,” said Daane. “For a potential purchaser it will all come down to at what price can I buy them, how much can I grow them, how strategic are they to my business and how successful will they be under my ownership. A large semiconductor company might look at us and say strategically I need their platform to continue to deploy my products, I think that might be more of a driver for acquisition than revenue. It is not something that I am focused on personally, my job is to improve the company in the long term.”