The announcement that Samsung will license the FDSOI chip manufacturing process as an option at the 28nm node both for foundry customers and for its own chips is, of course, excellent news for FDSOI pioneer STMicroelectronics. It is also good news for fans of technology, and for advocates of choice. It is also excellent news for Soitec, which could use the revenue from volume purchasing of its “smartcut” silicon-on-insulator wafers.
But even better news is the fact that Cadence Design Systems Inc. (San Jose, Calif.) has announced a couple of IP cores suitable for FDSOI production as well as statements to the effect that it has been a long-time supporter of the technology.
Long-time supporter of FDSOI Cadence may have been, but also a darned quiet one until now. Until Samsung stepped forward the reasoning may have been, why upset leading foundry TSMC over something that might not happen.
ST is Samsung’s customer
If and when Synopsys, ARM also come forward then FDSOI will have the start of that much-used idea, an ecosystem. The arrival of the mainstream EDA and IP vendors in the FDSOI camp is in some ways more significant than Samsung's license agreement and words of encouragement.
Bear in mind it costs Samsung relatively little to sign the FDSOI license, transfer in the technology and then wait and see if the fabless and IDM communities want to bite. And ST was already buying a lot of bulk CMOS production from Samsung and could have applied some leverage.
But offering to make is not the same as making. What happened to the great expectations that Globalfoundries would be making FDSOI circuits in Dresden in 2013? What happened was a lack of customers.
Of course, it is the old chicken-and-the-egg problem. The customers are reluctant to come forward without seeing all the needed suppliers in place – and the suppliers are reluctant to commit without a known customer base.
Nonetheless, Jean-Marc Chery, COO at STMicroelectronics, is claiming 17 FDSOI designs are now in progress and we have been told the name of at least one of the customers. That is Mobileye NV (Amstelveen, The Netherlands), which supplies vision-based automated driver assistance systems (ADAS).
Volume manufacturing of FDSOI chips in 2015, courtesy of Samsung, is perhaps a reasonable expectation but there is a risk that any delay will hand the opportunity back to FinFET technology at 16 and 14nm including TSMC.
TSMC's focus on mobile could let FDSOI in
A big opportunity for FDSOI lies in the fact that TSMC has a clear focus on the mobile equipment sector where it is trying to hold on to orders from likes of Apple and Qualcomm and could be sold out for 20nm bulk CMOS and 16nm FinFET production. It's focus is such that it is preparing mobile-optimized processes that give 28nm FDSOI a capacity and technology opportunity.
TSMC may not be too dismayed if FDSOI gains some traction. It an always go back to STMicroelectronics and take a license and start offering the technology with relatively little ground lost. The risk is it will have given competitors in the form of Samsung and perhaps Globalfoundries an entry with some customers and will be asked to pay a higher price for late conversion to the FDSOI religion.
Back in October 2013 I was told by executives from the EDA and IP sector that we would know a tipping point had been reached for FDSOI when they no longer had to be paid to develop FDSOI intellectual property cores and design flows and when they started doing the work off their own initiative.
We don’t know who paid for the development but Cadence has announced Denali DDR4 and USB physical layer IP cores are available on FDSOI. "From early on, Cadence has worked with STMicroelectronics on FDSOI technology and can assure our customers that they can quickly implement these IP solutions and sign off their designs," said Martin Lund, general manager of the Cadence IP Group.
Cadence sounds convinced and if ARM and a few other EDA and IP vendors can put whole-hearted support behind FDSOI, the process could be on its way. That at least gives engineers a choice and then it will be FDSOI customer success, or failure, that will have the last word on the technology.
Related links and articles: