It is imperative that North American semiconductor startups quickly abandon a business model that has served the industry well for the last 10 or 15 years but is now “”running out of steam,”” an industry conference heard Tuesday.
The soaring cost of chip development, the increasing number of Asian
foundries and the difficulty of testing new and intricate chips make it next to impossible for North American startups to rely on “”the fabless model,”” said Andrew Chowaniec, chair of the Information Technology Association Canada’s (ITAC) board of directors and chairman of Ottawa’s Tundra Semiconductor Corp.
A decade ago, companies could expect to develop a chip for $5 million, Chowaniec recalled following his address to attendees of ITAC’s ninth annual executive forum on microelectronics. But nowadays it can cost $50 million or more to do the same thing, not to mention the high cost and coordination effort of testing the new chip, which can have hundreds of millions of complex transistors.
“”Back then, the bet was much safer,”” he said. “”Now, you’re going to take a much bigger risk even before you have a customer and (before) you have any traction in the market. If you succeed, investors are going to want a substantial return. Well, if it costs $50 million, you’re going to have to value your company in the hundreds of millions to get that return back. And that’s pretty unlikely.””
Meanwhile, China is currently constructing some 50 new fabrication foundries, creating “”a huge number of players that don’t play by the same rules that we do,”” added Chowaniec. “”So we’re going to have to change the way we compete.””
However, the way in which North American startups should adjust operations in response to outside pressures and cyclical changes in the industry is so far unclear, said Chowaniec, adding “”if I could figure it out, I’d obviously make a huge amount of money very fast.””
The one thing that is for sure: Companies must dissect the high costs inherent in fabrication, design and development, perhaps using older fab technologies to offset the constant siphoning of limited revenues, said Chowaniec. Also setting successful startups apart is how company entrepreneurs can think up new ways to offset the costs of design and testing.
While the design costs are truly challenging for a small company, “”how it maneuvers through China and figures out what its role is, is an even bigger challenge,”” said Jodi Shelton, executive director of the Texas-based Fabless Semiconductor Association. “”It’s no longer about being 100 per cent Canadian with everything you do (located) here except for maybe manufacturing. You have to go global … fast.””
For a small company, it’s extremely challenging to secure foundry partners, but it’s essential to latch on to Chinese foundries, for example, to gain ground, added Shelton. In fact, one thing that must be present in any new-generation business plan is “”a China plan,”” she said. “”For venture capitalists, it’s probably a question on their list: What are you going to do in China?””
That’s why companies such as California-based Silicon Optix Inc. has two employees in Taiwan, and have sales channels in Korea, Japan and China.
Paul Russo, CEO of Silicon Optix, said it’s critical to have customer interaction in the Far East. Yet, “”to have your own people on the ground in China, I’m not sure how critical that is … yet.”” For Russo, “”it’s a case of: Are you better served by a good rep who has more contacts than your own employees?””
Russo disagreed with Chowaniec’s call for a new business model, suggesting instead that the model will stay the same and businesses will adjust accordingly. “”Because it’s so hard to create big chip companies, a large percentage of (startups) will end up being acquired and fewer will go public.””
Comparatively, “”In 2000, most exits were IPOs,”” observed Russo. “”But in the future most exits for clever start-ups will be (mergers and acquisitons). And one in 10 that have really identified technology products and can get to be big enough to create the returns will get the $50- or $60-million and will go public.””
Chowaniec predicted that North American clusters, Ottawa included, will thrive on the ongoing merger of nanotechnology and MEMS (microelectromechanical systems) with standard processing.
“”I see the health-care wave being huge in the next 10 or 15 years,”” he said. “”If we can create nanotechnology sensors that can be combined with IT and microprocessor components, there are whole new generations of applications that can be (developed) rather than building another big complex chip for a communications product where it’s tough even for established companies to do that.””