Did Texas Instruments make a smart move when it decided to rely more heavily on its main foundry partners for basic digital CMOS transistor development?
The numbers, such as return on invested capital, certainly support TI’s hybrid approach to manufacturing. Executives argue that at the 90- and 65-nm technology generations, duplication was rampant. TI engineers spent time developing a complex process flow, a foundry partner developed its process, and then the two would get together to figure out how TI’s chips could be made at the foundry, meeting TI’s electrical specifications. Three steps are condensed to one with the new model.
As Bill Krenik, CTO of the wireless business unit, explained at last week’s analysts meeting, the approximately $150 million saved each year by outsourcing transistor development is money that can be used to support small design teams with ideas for innovative new products.
“With that money, we can put more fuel on these small embers,” adding perhaps a dozen engineers to product development efforts, he said.
Krenik is one of a handful of engineers responsible for LoCosto and its follow-ons, Digital Radio Processor (DRP) chips which truly have changed the playing field in cellphone ICs. Back in 2000, Krenik and others began talking about the challenges of integrating the RF function – often done in a BiCMOS process -- with the digitally-intensive baseband IC. The goal was to combine the two in a purely digital CMOS process, with no extra masks to create RF friendly inductors, capacitors, and other devices normally required to create RF circuits.
Dennis Buss, a vice president of silicon technology development, recalled that the LoCosto team came to see him in the early days of their investigations. They asked if TI’s digital CMOS process could support an inductor with 20 nanohenries of inductance. No, he replied, but it could provide 13. Krenik’s team took that value and went back to work.
The question facing TI is whether its new model of relying on foundries would, back in 2000, have resulted in DRP and the LoCosto chips. Could Krenik have gotten his question answered about the value of the on-chip inductor?
Kevin Ritchie, the senior vice president of manufacturing at TI, said such questions will still get answered. TI will have more than 200 people working on digital CMOS, he said, many of them based in Taiwan.
Will the answers come back from Hsinchu as directly and as clearly as if a TI engineer in Dallas had the knowledge directly? Will, for example, the LoCosto design teams of the future find out that the process can support 13 nanohenries? Or will the foundry team guesstimate the answer, reporting an overly optimistic answer to save face? Or an overly pessimistic answer, to provide a guardband?
In essence, TI is saying that its available dollars are better spent on design teams, on innovative engineers like Bill Krenik.
The question that probably will never be answered in any definitive way is whether technology-dependent product innovations like LoCosto will still happen at TI, even as most of the underlying platform is developed over the ocean.
About weQuest: weQuest's are written by G Dan Hutcheson, his career spans more than thirty years, in which he became a well-known as a visionary for helping companies make businesses out of technology. This includes hundreds of successful programs involving product development, positioning, and launch in Semiconductor, Technology, Medicine, Energy, Business, High Tech, Enviorntment, Electronics, healthcare and Business devisions.