Right now SiTime is addressing a $6 billion annual market traditionally served by more than 200 manufacturers of quartz crystals. The overall market is growing at CAGR about 5 percent but the MEMS part of the market is growing at a 65 percent CAGR, CEO Rajesh Vashist told eeNews Europe when we caught up with him in London.
Vashist characterized the market as $2 billion for simple resonators, $2 billion for more sophisticated oscillators usually catered for by a crystal and some analog circuitry and about $2 billion for more complex clocks and phase-locked loops (PLLs).
“Clearly as a fabless MEMS company we have an advantage [over crystal manufacturing companies] because we don’t have to own the infrastructure. This means SiTime can use the existing semiconductor and MEMS infrastructure and digital control of its MEMS to configure hundreds of devices from 10 or 12 base components,” Vashist said.
“That $6 billion market will be $8 billion in five years. If we can have 10 percent of that market we will be doing well. The use cases are very wide but the secret for us is to use our flexibility to create timing solutions that don’t exist rather than trying to kick low-cost quartz crystals out of slots.”
Nonetheless it is clearly SiTime’s position that eventually MEMS will displace quartz almost everywhere just as the transistor was able to displace the vacuum tube.
Since its acquisition by MegaChips in 2014 (see Japan’s MegaChips to buy MEMS maker SiTime) SiTime doubled the number of units shipped in 2015 compared with 2014 to 125 million. The company expects to double shipments again in 2016 due to new applications and taking market share from quartz crystals.
The acquisition by has gone well, Vashist said, largely because the parent has left them alone but is there to back them up if they need it. “The cost of capital for us as a private company was huge. Now the cost of borrowing is below 1 percent. I think we have exceeded their expectations.”
Vashist said the executives behind MegaChips were thrusting entrepreneurs and were eager for their company to enter the top ten of fabless chip companies. In 2013 they were ranked 21st (see Fabless chip companies ranked by 2013 sales). Now with SiTime’s help and increasing penetration of MEMS into the timing market there is hope they can continue to climb.
SiTime is in one sense a spin-off from Robert Bosch. Aaron Partridge and Markus Lutz had been working on Bosch R&D in California and when Bosch chose not to pursue the opportunity decided to found SiTime to exploit the technology with Bosch’s cooperation. As a result Robert Bosch Ventures was an investor in the company and Robert Bosch is SiTime’s manufacturing partner making MEMS for SiTime out of its wafer fab in Reutlingen, Germany.
Since then SiTime has added Tower Semiconductor Ltd. as a MEMS foundry and has used TSMC as its supplier of signal conditioning ASICs manufactured on 180nm CMOS.
When asked if with SiTime’s fast pace of growth or Bosch’s growth as the world’s leading MEMS vendor might lead to SiTime being asked to move manufacturing from Reutlingen, Vashist said no. Bosch is not known to be a MEMS foundry for any companies other than SiTime although it did recently indicate that it was open to fabbing for other fabless MEMS companies (see Bosch open to making MEMS for others).
Nonetheless SiTime has the freedom to move its manufacturing if not the need or desire. Vashist said: “The process is 100 percent ours. We own the IP but we also have know-how. We have our own tools and have developed things like simulators for MEMS. In 2008 we helped them out when they had an empty fab. But even if we needed a billion units a year our MEMS are so small that would only be 10,000 wafers. We are not there yet.”
There has been remarkable progress in the performance of MEMS timing circuits over the last ten years producing an 800-fold reduction in jitter and a 250-fold improvement in stability. As a result while there has been a consumer application focus for SiTime the company is now able to address other markets, such as timing for networking and automotive (see SiTime uses dual MEMS for network oscillators).
Given SiTime’s success in timing circuits is there now a temptation to go after other adjacent markets that depend on similar MEMS, such as moving mirrors or actuators? “There is zero temptation. Do one thing well and remain focused,” said Vashist. “For any one else to come in, they would have to replicate $100 million of R&D and 10 years’ work. For the next 5 to 10 years no one can take this market away from us,” Vashist said.
Given SiTime’s self-confidence in the MEMS timing market and its desire to remain focused, Vashist’s response is no surprise. “If we needed to I am sure we could, but who is there to buy? Getting 95 percent yield on 500 million units shipped, that’s the trick.”
Related links and articles: